Search

Tuna is superior among the marine fishes that are exported in the forms of raw fish and processed food. Separation of Tuna into their species is done in industries manually, and the process is tiresome. This work proposes an automated system for classifying Tuna species based on their images. An ensemble of region-based deep neural networks is used. A sub region contrast stretching operation is applied to enhance the images. Each fish image is then divided into three regions and is augmented before giving as input to pre-trained convolutional neural networks (CNN). After fine-tuning the models, the output from the last convolutional layer is given to a grouped 2D-local binary pattern descriptor (G2DLBP). Statistical features from the descriptor are applied to different classifiers, and the best classifier for each image region model is identified. Different ensemble methods are subsequently used to combine the three CNN-G2DLBP models. Among the ensemble techniques, super learner ensemble method with random forest (RF) classifier using 5-fold cross-validation shows the highest classification accuracy of 97.32%. The performance of different ensemble methods is analyzed in terms of accuracy, precision, recall, and f-score. The proposed system shows an accuracy of 93.91% when evaluated with an independent test dataset. An ensemble of region-based CNN with textural features from G2DLBP is applied for the first time for fish classification.

This paper presents a new facial feature descriptor called Fused Cross Lattice Pattern of Phase Congruency (FCLPPC) for high accuracy homogeneous and heterogeneous illumination invariant intra/inter-modality face recognition. Using the dimensionless phase congruency features, an effective homogeneous and heterogeneous illumination invariant local feature extractor has been devised. Finally, a novel multi-directional binary pattern named Cross Lattice Pattern (CLP) has been proposed. CLP is applied to the previously extracted invariant phase congruency feature map to generate the CLPPC images. Weighted alpha-blending has been performed on the CLPPC maps to generate Fused CLPPC (FCLPPC) map. Recognition results on Extended Yale-B, TUFTS, CMU-PIE, and CASIA NIR-VIS dataset has been presented to depict the superiority of the proposed scheme over other state-of-the-art methods. Additionally, the proposed FCLPPC has been combined with a lightweight Convolutional Neural Network to further augment the recognition accuracy.

The main subject of this paper are binary pattern sequences, that is, sequences of the form ( − 1 ) # ( n , A ) where A is a set of strings of 0 s and 1 s, and # ( n , A ) denotes the total number of times patterns from A appear in the binary expansion of n. A sequence is said to be noncorrelated if the corresponding spectral measure is equal to the Lebesgue measure. We show that it is possible to algorithmically verify if a given binary pattern sequence is noncorrelated. As an application, we compute that there are exactly 2272 noncorrelated binary pattern sequences of length ≤4. If we restrict our attention to patterns that do not end with 0 , we put forward a sufficient condition for a pattern sequence to be noncorrelated. We conjecture that this condition is also necessary, and verify this conjecture for lengths ≤5.

This paper presents a novel face image descriptor called Gammadion Binary Pattern of Shearlet Coefficients (GBPSC) for illumination and noise invariant, homogeneous and heterogeneous face recognition. Exploiting the energy concentration property of the Digital Shearlet Transform, an efficient illumination and noise invariant feature extractor has been devised. Finally, inspired by the Gammadion structure, a robust multi-directional local binary pattern named Gammadion Binary Pattern (GBP) has been proposed. GBP is applied on the previously extracted illumination and noise invariant feature map to generate the GBPSC images. Recognition results on Extended Yale B and TUFTS dataset indicate the primacy of the proposed scheme in terms of common feature representation under varying illumination, and modality. Furthermore, the merger of the proposed GBPSC and Convolutional Neural Network (CNN) consistently outperforms other state-of-the art methods.

Local binary patterns and their extensions are heavily used in image texture classification. However traditional LBP codes are sensitive to noise and have limited capability of discriminating various texture patterns. This study presents a novel Completed multiple adaptive threshold patterns (CMATP) descriptor for texture classification to ensure higher classification rates. For each patch of pixels statistical parameters like mean, standard deviation and value of centre pixel are computed to determine structural properties. Based on these observations pixel patches are categorized in to uniform, moderate and non-uniform structures. A suitable adaptive threshold is then obtained for each class of pixel groups, which can generate binary pattern that are more uniform and can produce distinct binary code for different texture patterns. Further to extract centre pixel and contrast information from each patch, multiple threshold based complementary sign and magnitude patterns are generated in a similar way. The sign and magnitude pattern are then combined form CMATP descriptor. We have compared the proposed CMATP pattern with some state of the art LBP based descriptors on some benchmark datasets like Outex, Brodatz and UMD. The experimental results shows superior performance of the CMATP as compared to others.

In this work, the evaluation of segmentation techniques towards the detection tumours or gliomas on BraTs 2018 data sets on T2 weighted 3D MRI images is carried out. Techniques experimented include pixel wise Linear Binary Pattern (LBP), region growing and Otsu thresholding to compare the efficiency of the same with proposed approach. In this work, 3D MRI images are initially subject to slicing from which about 10% from the top and bottom are normalized as those slices does not possess any significant details (empty slices). Further, the enhancement of tumour/gliomas regions is performed on normalized slices using high boost ‘convolution kernel processing in various trials with varying kernel sizes of high boost convolution kernel. The glioma regions enhanced using kernel are subject to local adaptive thresholding technique for extraction of gliomas from normalized slices. The detected gliomas are compared with the results of ground truth segmentation to determine the effectiveness of the proposed technique. It is observed in 80% of slices are successful in detection of gliomas both in case of Low Grade Glioma (LGG) and High Grade Glioma (HGG). By analyzing the results, it is observed that proposed method has outperformed with average Jaccard and Dice similarity recorded as 87.20% and 89.18% for HGG and where as in case of LGG it is found to be 83.77% and 87.74%.

The signal processing is one the very important research area in the computer sciences and artificial intelligence. Because, audio recognition, voice activity detection, disease diagnosis, brain activity detection and predictions methods are evaluated using signal processing methods. Nowadays, deep methods have been become popular in the signal processing applications. In this article, a novel hybrid feature extraction network by using novel approximations and multiple pooling method. The proposed method uses both binary pattern (BP) and ternary pattern (TP) as feature extractor. In order to extract variable and distinctive features, dynamic center based feature extraction strategy is used. Hence, the proposed feature extraction network is called as dynamic center based binary and ternary pattern network (DC-BTPNet). The proposed DC-BTPNet is consists of 9 layer. Also, a novel multiple pooling method is used in DC-BTPNet. In order to select features, neighborhood component analysis (NCA) is utilized. Finally, the extracted features are forwarded to polynomial kernel support vector machine (SVM). In order to evaluate performance of the proposed method, a novel dataset is created. The proposed DC-BTPNet based multiple learning method achieved 89.0% accuracy rates and it was compared to other state-of-art convolutional networks. Other well-known conventional classifiers are also used for instance linear discriminant analysis (LDA), k nearest neighbor (KNN) and bagged tree (BT) classifiers are used to compare performance of the classifiers. The comparisons and results clearly proved success of the DC-BTPNet. These results demonstrated that the proposed methods can be achieved successful results in larger datasets.

Phase measuring deflectometry (PMD) is a high precision measurement method for optical surface shape, which has been used in the fabrication and application of high quality optical element. For the on-line surface measurement of transparent optical elements in the laser inertial confinement fusion facility, PMD is a very promising method. However, the parasitic reflection from the rear surface is an existing problem for PMD to measure transparent element. In this paper, a parasitic reflection eliminating method with the use of binary pattern is proposed, the principle of which is described in detail. The ability of the proposed method to extract the front surface information is demonstrated by the numerical simulation. And the proposed method is used to measure a transparent window glass with a thickness of about 10 mm. The surface shape result shows a good agreement with the interferometer data with a sub-wavelength level accuracy.

A model is presented of how humans learn repeating patterns made of only two qualitatively distinct elements, for example: aabbaabab. The model proposes hierarchical coding principles or axioms that code the pattern into larger groups of elements at higher levels until perfect prediction is possible. Prior to solution the learner uses the transition probabilities associated with the coded elements present at the lower levels to make predictions. A simple rationale for weighting the contribution of the lower levels allows one to predict the proportion of errors that should occur at each pattern position, that is, to predict the error profile of a pattern, without using estimated parameters. Apparently no other such model exists to describe this basic type of human learning. Results published by Restle (1967) showing the error profile of three different binary patterns were used to test the model. Each error profile was successfully predicted, with r ranging from .91 to .94. Modifications, implications, limitations and qualifications of the model are discussed.

The number of individuals who have lost their fingers in our world is quite high and these individuals experience great difficulties in performing their daily work. Finger movements classification and prediction are one of the hot-topic research areas for biomedical engineering, machine learning and computer sciences. This study purposes finger movements classification and prediction. For this purpose, a novel finger movements classification method is presented by using surface electromyogram (sEMG) signals. To accurately classify these movements, a novel binary pattern like textural feature extractor is presented and this textural micro pattern is called as multi-centered binary pattern (MCBP). In the MCBP, five odd-indexed values of a block are utilized as center. The proposed MCBP based multileveled finger movements classification method evaluate by three cases. In the first case, the raw sEMG signals are utilized as input. In the second and third case, sEMG signals are divided into frames and these frames are utilized as input. A two-layered feature selector is used to choose the most valuable features. The purpose of using these two feature selectors together is to choose the optimum number of features. In the classification phase, two fine-tuned classifiers have been used and they are k-nearest neighbor (k-NN) and support vector machine (SVM). The proposed MCBP based method achieved 99.17%, 99.70% and 99.62% classification rates using SVM classifier according to Case 1 , Case 2 and Case3 respectively. The results show that the study is a highly accurate method.

Aimed at the interference of scene and the local overly split, we fully excavate the information of spatial and spectral domain to construct the feature descriptor called local and neighbor multiscale, and construct a hierarchical structure model for multispectral target detection. Based on multidimensional differential distance measure, we put forward Space and Spectrum Differential Structure (SSDS) operator to extract small scale fine structure and robust feature of multispectral targets. Based on the idea of binary patterns, we propose Local and Neighbor Binary Pattern (LNBP) operator for large scale neighborhood structure extraction of multispectral targets. Finally, we construct pyramid hierarchical structure model (MH-LS, Multi-scale Hierarchical-LS), design a multiscale and multilevel computing architecture, and fully unite the characterization and separability of two operators on the multidimensional multiscale structure. Also, MH-LS can make robust matching based on small sample coming true and improve the detection accuracy and efficiency. Experiments show that MH-LS does not need a large number of sample training, and can effectively detect multispectral objects in different scenes, postures, views and scales based on a small number of template sets.

Background and Objective In the last decade, Ultrasound-Guided Regional Anesthesia (UGRA) gained importance in surgical procedures and pain management, due to its ability to perform target delivery of local anesthetics under direct sonographic visualization. However, practicing UGRA can be challenging, since it requires high skilled and experienced operator. Among the difficult task that the operator can face, is the tracking of the nerve structure in ultrasound images. Tracking task in US images is very challenging due to the noise and other artifacts. Methods In this paper, we introduce a new and robust tracking technique by using Adaptive Median Binary Pattern(AMBP) as texture feature for tracking algorithms (particle filter, mean-shift and Kanade-Lucas-Tomasi(KLT)). Moreover, we propose to incorporate Kalman filter as prediction and correction steps for the tracking algorithms, in order to enhance the accuracy, computational cost and handle target disappearance. Results The proposed method have been applied on real data and evaluated in different situations. The obtained results show that tracking with AMBP features outperforms other descriptors and achieved best performance with 95% accuracy. Conclusions This paper presents the first fully automatic nerve tracking method in Ultrasound images. AMBP features outperforms other descriptors in all situations such as noisy and filtered images.

This paper reports a face identification system for visible, look–alike and post–surgery face images of individuals using some novel variants which are exploited from local graph structure (LGS). The proposed LGS variants attempt to improve the performance of the face identification system under the influence of pose changes, facial expression changes, illumination variation, makeup, accessories (glasses) and facial complexity (look alike and plastic surgery). The idea is to represent each pixel along with its neighborhood pixels of a face image based on the regenerated directed local graph structure. From the newly defined local graph structure a binary pattern is generated for each pixel and this binary string is then converted into a decimal value and generates a transformed pattern. Finally, this transformed pattern is used to generate a concatenated histogram which is then used for matching and identification by using three well-known classifiers, namely, locally scaled sum of squared differences (LSSD), locally scaled sum of absolute differences (LSAD), and histogram intersection (HI). Unlike prior works, face images do not have to undergo the preprocessing stages as each novel variant deals with local structure of a face image by disregarding other effects. The UMIST, the JAFFE, the Extended Yale Face B, the Look-alike and the Plastic Surgery face databases are used for the evaluation. Extensive experiments on face databases exhibit promising and convincing results. Further, the experimental results have led to a robust identification system which is found to be invariant to different challenges made of due to capturing environment and face modality changes.

Dielectric properties of red banana are studied at different ripening temperatures.Different image processing techniques are used for ripening grade determination.Tsallis entropy technique is proposed for red banana ripening grade determination. In this research work, the dielectric properties of red banana fruit are studied at different ripening temperatures for developing a rapid and non-destructive assessment method to measure the ripening stages of red banana. A 5 volt sine wave AC power supply and a rectangular parallel plate capacitor circuit are used to measure the difference in dielectric properties caused by the introduction of a red banana in between the plates. The values of properties like capacitance and relative permittivity are increased continuously whereas impedance and admittance are decreased gradually with increase in ripening stages of red banana. In image processing approach, Noise Reductant Local Binary Pattern (NRLBP), Local Binary Pattern (LBP), Completed Local Binary Pattern (CLBP) based techniques are used for red bananas ripening grade determination. The processing stages involved are enhancement, Binary Pattern generation and classification. The variant Binary patterns are tested on noisy as well as noiseless condition and the results are compared. A novel enhancement technique for banana ripening grade determination is proposed based on segmentation using Tsallis entropy. Also novel idea on the automation of q parameter involved in Tsallis Entropy is implemented. The threshold parameter of the Noise Reductant Local Binary Pattern (NRLBP) varied and its effect on classification rate is studied. A new modification is proposed and implemented on NRLBP to accommodate uniform background and areas with the image. Classification is done using Chi-Square distance/nearest neighbor and Fuzzy C means (FCM) clustering. The results are compared and superiority of FCM method for banana ripening grade determination is noted.

This paper proposes the definition of the architecture of an evolving fuzzy neural network based on self-organizing direction aware data partitioning through stochastic processes based on the dataset used in the model. The choice between four different types of logical neurons in the second layer and the six different types of neuron activation function that compose the artificial neural network of aggregation are defined in a procedure of generating random pairs of combinations between the two factors. The combination that obtains the maximization of training accuracy is chosen to compose the network and perform the composition of the model structure, whose components can be transferred to readable IF-THEN rules for interpretability purposes. Furthermore, the model is able to adapt its parameters and evolve its structure autonomously with new data samples through self-organized direction-aware data partitioning (SODA). In this context, we also propose a technique to measure the degree of changes of neurons (rules), which could be used for structural active learning purposes (e.g, to request user feedback in the case of significant changes). To compare the proposed approach, binary pattern classification tests were performed, and the results were compared with other models of fuzzy neural networks and neural networks, obtaining satisfactory results with the stochastic definition of elements that compose their architecture comparing the final accuracy of the model when classifying real datasets. The proposed model obtained the best result in 3 of the four synthetic bases evaluated, in addition to the best accuracy results in the classification of patterns in five of the nine evaluated real datasets. It highlights the accuracy in problems in patients who underwent breast cancer surgery (72.44%), diabetes evaluation (67.87%), Australian (72.27%) and German (80.51%) credit ratings evaluation, and finally in the classification of radar signals in the ionosphere (90.46%). It is also noteworthy to obtain fuzzy rules in evolution extracted from a real problem of the identification of respiratory diseases through the collection of saliva with 76.67% of accuracy. The results presented by the model were superior to traditional artificial intelligence models, while it was possible to extract knowledge in form of interpretable rules and to realize how these changed over time.

This work presents novel LBP variant so-called Graph Based Structure Binary Pattern (GBSBP) in different challenges. By considering 3 × 5 & 5 × 3 pixel windows, 4 graphs are created & each graph structure is injected with 3 different edges (→, ↔ and ―) which corresponds to single edge (any one) from one pixel position to another pixel position. Each injected edge specifies different measures & produces different values by utilizing 2 pixels. From both the windows 8 & 6 values are produced after applying different edge measures for each pixel position. By using these values 4 transformed images are produced. Then extracted histograms (from images transformed) are integrated to make the GBSBP feature size. Further enhancement in accuracy is achieved when GBSBP features are incorporated with LPQ features.

As one of the most critical components in rotating machinery, it is essential to determine the health of rolling bearings on time. The effective feature extraction method is considered significant for fault diagnosis. In order to extract sufficient and effective bearing information, a novel extraction method called multiscale one-dimensional hybrid pattern (1D-HBP) is proposed for fault diagnosis of rolling bearings. The proposed method extracts the local and global texture statistical information of signals to reflect different bearing conditions. Considering the inherent multi-scale characteristics of the vibration signals, multiscale analysis is employed to obtain discriminative features of different scales. Two rolling bearings sets with changeable loads and rotating speeds verified the effectiveness and practicability of the proposed diagnostic model. Compared to other models examined for the same dataset our proposed model achieves a remarkably high classification accuracy.

Background Heart valve diseases are commonly seen ailments, and many people suffer from these diseases. Therefore, early diagnosis and accurate treatment are crucial for these disorders. This research aims to diagnose heart valve diseases automatically by employing a new stable feature generation method. Materials and method This research presents a stable feature generator-based automated heart diseases diagnosis model. This model uses three primary sections. They are stable feature generation using the improved one-dimensional binary pattern (IBP), selecting the most discriminative feature with neighborhood component analysis (NCA), and classification employing the conventional classifiers. IBP uses three kernels, and they are named signum, left signed, and right signed kernels. By applying these kernels, 768 features are generated. NCA aims to choose the most discriminative ones, and 64 features are chosen to employ NCA. The k nearest neighbor (kNN) and support vector machine (SVM) classifier are employed in the classification phase. Open access (public published) Phonocardiogram signal (PCG) sound dataset is used to calculate this model's measurements. This dataset contains 1000 PCGs with five categories. Results The presented IBP and NCA-based heart valve disorders classification model tested using kNN and SVM classifier and attained 99.5% and 98.30% accuracies, respectively. Conclusions Per the results, the presented IBP and NCA-based PCG sound classification is a successful method. Moreover, this model is basic and high accurate. Therefore, it is ready for the development of real-time implementations.

Brain activities can be evaluated by using Electroencephalogram (EEG) signals. One of the primary reasons for traffic accidents is driver fatigue, which can be identified by using EEG signals. This work aims to achieve a highly accurate and straightforward process to detect driving fatigue by using EEG signals. Two main problems, which are feature generation and feature selection, are defined to achieve this aim. This work solves these problems by using two different approaches. Deep networks are efficient feature generators and extract features in low, medium, and high levels. These features can be generated by using multileveled or multilayered feature extraction. Therefore, we proposed a multileveled feature generator that uses a one-dimensional binary pattern (BP) and statistical features together, and levels are created using a one-dimensional discrete wavelet transform (1D-DWT). A five-level fused feature extractor is presented by using BP, statistical features of 1D-DWT together. Moreover, a 2-layered feature selection method is proposed using ReliefF and iterative neighborhood component analysis (RFINCA) to solve the feature selection problem. The goals of the RFINCA are to choose the optimal number of features automatically and use the effectiveness of ReliefF and neighborhood component analysis (NCA) together. A driving fatigue EEG dataset was used as a testbed to denote the effectiveness of eighteen conventional classifiers. According to the experimental results, a highly accurate EEG classification approach is presented. The proposed method also reached 100.0% classification accuracy by using a k-nearest neighborhood classifier.

Binary pattern family is considered as a powerful tool for visual texture classification. Most popular methods improve the classification performance by multi-feature fusion. However, many sub-features are redundant and low-discriminative and the classification system has high computational complexity and unsatisfactory results. To handle above problems, this paper proposes a compact multi-pattern encoding descriptor for visual texture classification. First, we develop local extremum patterns and local center pattern to represent the neighborhood intensity changes. Then, we design a compact encoding scheme to encode local maximum, minimum and center patterns into a three-bit binary code, named MMC pattern. Finally, a compact multi-pattern encoding descriptor is proposed by combining the traditional local sign pattern and MMC pattern. Experimental results on five representative texture databases demonstrate that our method achieves the state-of-the-art texture classification performance.

Effective feature representations and similarity measurements are crucial for large-scale image retrieval, and conventional methods often learn hash functions from a predefined hand-crafted feature space. Meanwhile, the spatial structure in raw images always lost in most previous methods. Encouraged by the recent advances in convolutional neural networks (CNNs), a novel Spatial Pyramid Deep Hashing (SPDH) algorithm is developed for the task of fast image retrieval. In our SPDH algorithm, the CNN with a spatial pyramid pooling and a locally-connected layer with binary activation functions is utilized to build the end-to-end relation between the raw image data and the binary hashing codes for fast indexing. Different from the fully-connected layer, the locally-connected layer can consider each local spatial bin as an independent unit and only connect the local bin to preserve the spatial pyramid structure for hash codes. The learning of both the hash function and the feature representations are jointly optimized via backward propagation with classification or similarity loss function on the large-scale labeled dataset such as ImageNet. Moreover, a spatial pyramid binary pattern matching algorithm is developed to achieve partial local similar matching among the images. Our experimental results have shown that our SPDH method can outperform several state-of-the-art hashing algorithms on the CIFAR-10, SIVAL and the Oxford buildings datasets.

In this paper, existing local structure patterns (LSPs) are reviewed and categorized into two types: intensity-based LSPs (I-LSPs) and gradient-based LSPs (G-LSPs). I-LSPs include local binary pattern (LBP), modified census transform (MCT) and generalized binary pattern (GBP) methods that compare the intensities of differently selected neighboring pixels within a 33 neighborhood with differently-formed reference intensities, encoding 256, 511, and 19,162 binary patterns respectively. G-LSPs include local gradient pattern (LGP), modified gradient pattern (MGP), and generalized gradient pattern (GGP) methods that compare the gradient magnitudes of differently selected neighboring pixels within a 33 neighborhood with differently-formed reference gradients, encoding 256, 511, and 19,162 binary patterns respectively. We extend all these LSPs to multi-scale block LSPs (MB-LSPs) that concatenate multiple block-based LSPs. Finally, we propose several hybrid LSPs that combine I-LSPs and G-LSPs by means of the AdaBoost feature selection. In experiments using AR564, BioID, ColorFERET and LFW databases to evaluate the eye detection accuracy of the proposed LSPs, the I-LSPs were good for detecting the eyes in low quality images, the G-LSPs were good for detecting the eyes in high quality images, and the hybrid LSPs achieve state-of-the-art eye detection accuracy across image qualities. Existing local structure patterns are reviewed and categorized into two types.We propose several hybrid LSPs by the AdaBoost feature selection.I-LSPs were good for detecting the eyes in low quality images.G-LSPs were good for detecting the eyes in high quality images.Hybrid LSPs provided state-of-the-art eye detection accuracy across image qualities.

This paper proposes a new local texture descriptor, called quad binary pattern (QBP). Compared with local binary pattern (LBP), the QBP is with stronger robustness for feature extraction under complex scene (e.g., luminance change, similar target and background color) and with lower computational complexity. To demonstrate its effectiveness, the proposed QBP is further applied on the mean-shift tracking, in which a joint color-QBP model is developed to effectively represent the color and texture characteristics of the target region. Extensive simulation results have demonstrated that the proposed algorithm is able to improve the tracking speed and accuracy, compared with the standard mean-shift tracking and joint color-LBP model based mean-shift tracking.

In this paper a novel method based on facial skin aging features and Artificial Neural Network (ANN) is proposed to classify the human face images into four age groups. The facial skin aging features are extracted by using Local Gabor Binary Pattern Histogram (LGBPH) and wrinkle analysis. The ANN classifier is designed by using two layer feedforward backpropagation neural networks. The proposed age classification framework is trained and tested with face images from PAL face database and shown considerable improvement in the age classification accuracy up to 94.17% and 93.75% for male and female respectively.

Most of the local descriptors exists today are based on the relationship between the neighborhood pixels and the center pixel by considering the 3 × 3 spatial window. Moving one step forward from the previous methodologies used, this research paper introduces the novel descriptor for Face analysis called as orthogonal difference-Local binary pattern (OD-LBP). In OD-LBP, initially the 3 gray level differences are computed for each orthogonal position (of 2 groups, in 3 × 3 pixel window) by subtracting the 2 closest neighborhoods and the center pixel from the respective orthogonal value. Then these respective gray level differences are compared with the value produced after employing the novel comparison method. Finally the 1 × 24 size row vector is produced by concatenating the binary patterns produced from both the orthogonal groups. Eventually the 24 bit row vector is split into three 8 bit binary patterns from which the 3 OD-LBP codes are produced. By employing this concept for each pixel position eventually results in the 3 OD-LBP transformed images. All the 3 transformed images are then divided into 3 × 3 sub-regions for histogram extraction. The combined sub-regional histograms from all the transformed images are the entire feature size of the OD-LBP descriptor. To achieve the compact feature representation PCA is applied further and then classification is performed by the SVMs. The entire concept is tested on 5 challenging databases which include ORL, GT, JAFFE, MIT-CBCL and Yale. The proposed descriptor (i.e. OD-LBP) achieves very promising results on all 5 databases as it comprehensively outperforms the other 10 state of art descriptors LBP, CS-LBP, 6 × 6 MB-LBP, OC-LBP, MBP, HOG, ELBP, OS-LTP, WCLBP and RLBP on majority of sub-sets. Numerous states of approaches from the literature are also outclassed by proposed descriptor.

A comprehensive feature representation for electroencephalogram (EEG) signal to achieve effective epileptic focus localization using a one-dimensional quad binary pattern (QBP) is proposed in this work. The wavelet packet decomposition (WPD), entropies, and QBP methods are applied to EEG signals for identifying the non-focal and focal classes. The proposed approach consists of three strategies, based on the local pattern transformation of EEG signals using QBP. The time-domain EEG signals are transformed into the QBP domain, and histogram features are extracted in the first strategy. Secondly, the nonlinear features like sample entropy, log energy entropy, fuzzy entropy, permutation entropy, and approximate entropy are computed from the EEG signals and concatenated with the QBP features. In the third strategy, the EEG signals are decomposed into sub-bands by employing WPD, and histogram features based on QBP are extracted from each sub-band. The computed feature sets are used to classify the non-focal and focal EEG signals using an artificial neural network (ANN). For the selection of best settings in WPD, the EEG signals are decomposed in seven commonly used wavelet families with different decomposition levels, and the wavelet family providing the highest classification accuracy with the minimal computational cost is found out. The proposed approach is validated using the widely recognized Bern-Barcelona EEG dataset. The proposed method achieved a classification accuracy of 95.74 % on this dataset using the proposed WPD based QBP approach. In conclusion, the proposed approach is efficient in identifying the focal EEG signals and can be useful for accurate detection of focal regions in the epilepsy diagnosis.

This research paper introduces two novel color descriptors for Face Recognition (FR) called as Color ZigZag Binary Pattern (CZZBP) and Color Median Block ZigZag Binary Pattern (CMBZZBP). The pixels aligned in ZigZag manner are utilized for producing the feature size of both the proposed descriptors. For first proposed descriptor, the ZigZag features are extracted from 3 color components of the RGB color space format, which are concatenated to produce the entire feature size of the CZZBP descriptor. Specifically, on 3 color components i.e. R, G and B, the features are extracted by utilizing three different type of ZigZag binary patterns called as ZZBP1, ZZBP2 and ZZBP3. The ZZBP1 is applied on the R component, ZZBP2 is applied on the G component and ZZBP3 is applied on the B component. In the second proposed descriptor, the same concept is employed on medians computed from 9 different blocks of 9 × 9 pixel window, where each block size is 3 × 3. To produce compact feature size, PCA is employed and classification is performed by SVMs and NN. The 2 databases used for performance evaluation are GT and Faces94. Experiments reveals effectiveness of the proposed descriptors than compared descriptors. Among all descriptors, CMBZZBP is most impressive.

Induction motors, which are widely used in industrial applications, are indispensable tools of the industry. Induction motors work in almost every part of the industry, such as production, packaging, and service. In this study, an acoustic-based method is proposed for the detection of the rotor and bearing faults of three-phase induction motors. In the first stage, two fault sound datasets were collected and these datasets are called near and far. For extracting features from these sounds, a multilevel feature generation method is presented and this method uses Discrete Wavelet Transform (DWT) and Local Binary Pattern (LBP) methods together. Neighborhood Component Analysis (NCA) method was used to select the most informative features. Selected features are utilized as the input of SVM (Support Vector Machine) and KNN (K Nearest Neighborhood) classification algorithms. 99.8% classification success was achieved as a result of the SVM algorithm and the KNN algorithm reached 99.9% classification accuracy.

The recently proposed optimized dithering techniques can improve measurement quality. However, the objective function in these optimization methods just qualifies the global similarity of the pattern while the global optimization methods ignore the influences of local structure. To complete the phase-based optimization for high-quality three-dimension (3D) shape measurement, this paper presents a novel objective function which consists of global intensity term and local texture term, as well as a novel framework. Then, both simulation and experimental results show that the proposed method is able to improve quality and speed of 3D reconstruction for optimized dithering techniques.

First-order Riesz transform based monogenic signal representation has been widely used in image processing and computer vision, however it only characterizes image intrinsic one-dimensional structure, and is incapable of describing intrinsic two-dimensional structure. To this end, a novel feature extraction approach, named Riesz Binary Pattern (RBP), is proposed for face recognition based on image multi-scale analysis and multi-order Riesz transform. RBP consists of two complementary components, i.e., local Riesz binary pattern (LRBP) and global Riesz binary pattern (GRBP). LRBP is obtained by performing local binary coding operator on each Riesz transform response to extract image intrinsic two-dimensional structure features. While GRBP is the global binary coding of joint information of image pixel multi-scale analysis and multi-order Riesz transform. Histogram of LRBP and GRBP are concatenated to form face image RBP description. Experimental results on three databases demonstrate that our proposed RBP descriptor is more discriminant in extracting image information and can provide a higher classification rate compared to some state-of-the-art image representation methods. A new image descriptor RBP is presented for face recognition in this paper.RBP is based on image multi-scale analysis and multi-order Riesz transform.RBP consists of two complementary components, i.e., local Riesz binary pattern (LRBP) and global Riesz binary pattern (GRBP).Experimental results on four databases demonstrate the superiority of our RBP compared with other image representation methods.

Deformable image registration remains a challenging research area due to difficulties associated with local intensity variation and large motion. In this paper, an Accurate Inverse-consistent Symmetric Optical Flow (AISOF) method is proposed to overcome these difficulties. The two main contributions of AISOF include the following: (1) a coarse-to-fine strategy for an inverse-consistent symmetric method and (2) a novel Hybrid Local Binary Pattern (HLBP) to the classical Lucas–Kanade optical flow method. The HLBP consists of a median binary pattern and a generalised centre-symmetric local binary pattern. The generalised centre-symmetric local binary pattern has two thresholds, and this pattern can capture more information than the classical centre-symmetric local binary pattern, which has one threshold. The proposed HLBP can cope well with high contrast intensity and local intensity variation. Because the inverse-consistent symmetric method can reduce inverse consistency errors in Markov random fields based registration methods, we adopted this method to improve the accuracy of registration. In addition, a coarse-to-fine strategy was adopted to handle large motion. The proposed AISOF method was evaluated for 10 publicly available 4D CT lung datasets from the DIR-Lab. The mean target registration error of the AISOF method is 1.16 mm, which is significantly superior to the error of the classical Lucas–Kanade optical flow method, 2.83 mm. Moreover, this error is also the smallest of all unmasked registration methods using these datasets.

Display Omitted A novel Video Copy Detection (VCD) is proposed.The VCD is based on the Weber Binarized Statistical Image Features descriptor (WBSIF).The matching process is performed by using of the 2 test.Results outline the robustness and effectiveness with high accuracy.Proposed method shows a noteworthy performance against currently existing techniques. This paper presents a new video copy detection system based on testing similarities between textural feature vectors which are extracted from videos. Herein, the proposed method is based on Weber Binarized Statistical Image Features (WBSIF) which is an interior improvement of Weber Local Descriptor (WLD). Actually, the orientation gradient in WLD is substituted by a recent Binarized Statistical Image Features (BSIF) as a local textural descriptor. The WBSIF approach is tested on three databases and evaluated through several attacks. Moreover, the proposed method is compared to the recent existing approaches, especially those mostly used in the literature, which are based on the binary pattern descriptors. The obtained results outline the robustness and the effectiveness of the proposed video copy detection system in terms of precision, recall, Fscore, accuracy and collision test. This study shows clearly a noteworthy performance of the proposed scheme against currently existing techniques.

In this paper, we propose a novel feature type, namely Motion Binary Pattern (MBP) and different computation strategies for the well-known Volume Local Binary Pattern (VLBP). MBPs are a combination of VLBPs and Optical Flow. By combining the benefit of both methods, a simple and efficient descriptor is constructed. Motion Binary Patterns are computed in the spatio-temporal domain while the motion in consecutive frames is described. Finally, a feature descriptor is constructed by a histogram computation. Volume Local Binary Patterns are a feature type to describe object characteristics in the spatio-temporal domain. But apart from the computation of such a pattern further steps are required to create a discriminative feature. These steps are evaluated in detail and the best strategy is shown. For MBPs and VLBPs, a Random Forest classifier is learned and applied to the task of human action recognition. The proposed novel feature type and VLBPs are evaluated on the well-known, publicly available KTH dataset, Weizman dataset and on the IXMAS dataset for multi-view action recognition. The results demonstrate challenging accuracies in comparison to state-of-the-art methods.

In this paper, we propose a novel descriptor called spatially weighted order binary pattern (SWOBP) for color texture classification. The SWOBP descriptor not only encodes color order information in different channels but also encodes color order relationships in the spatial domain. To achieve these goals, we introduce a color gradient channel to complement the traditional color channels and explore a multi-channel color order pattern to jointly encode inter-channel features. Furthermore, we decompose local color differences into spatially weighted binary templates and use them to encode color order information in a local neighborhood. Finally, we aggregate all the encoded features into image histograms as texture descriptor. Experiments on five benchmark databases demonstrate that the proposed SWOBP descriptor achieves the state-of-the-art performance for color texture classification.

In this paper, we propose a texture descriptor, called Wavelet Domain Directional Binary Pattern using Majority Principle (WDDBP θ ) which integrates the discrete wavelet transform and Local Binary Pattern (LBP). The discrete wavelet transform provides the directional information, while LBP describes the texture features predominately. The proposed approach consists of four phases. In the first phase, the image is subjected to single level decomposition; resulting in discrete wavelet coefficients. The wavelet coefficients present at various distances; one, two and three are compared among themselves which results in sign pattern. In the second phase, the difference of wavelet coefficients located at different distances are compared with global difference of wavelet coefficient and results in magnitude pattern. In the third phase, the majority principle is applied for both sign and magnitude pattern resulting in WDDBP_S and WDDBP_M pattern. Finally, the histogram of both WDDBP_S and WDDBP_M is concatenated. The proposed method is also evaluated by involving the wavelet coefficients present at twelve different directions. The K-NN algorithm is applied to classify the input and database images. The experimental results demonstrate that the proposed method outperforms the other existing methods on Brodatz and Outex database with improved classification accuracy.

In this paper, a new coding scheme, local Gabor maximum edge position octal patterns (LGMEPOP) is proposed for content based image retrieval. The standard local binary pattern (LBP) collects the sign edge (binary code) information between the center pixel and its surrounding neighbors in an image. Further, the concept of LBP is extended to local maximum edge binary pattern (LMEBP) which collects the sign code (binary code) using the magnitude edges. These magnitude edges are collected based on the maximum edges between the center pixel and its surrounding neighbors in an image. In this paper, we propose a new feature descriptor, octal code which is coded based on the maximum edge positions (MEP) on Gabor responses. Specially, each pixel of every Gabor response gains eight edges based on the relationship between the referenced pixel and its neighbors. LGMEPOP utilizes the first three dominant (maximum) edge positions in an octal code generation. Then, these three maximum edge positions are encoded into three-eight octal numbers to produce the LGMEPOP. Further, the LGMEPOP is classified into two categories which are named as sign maximum edge position octal pattern (SMEPOP) and magnitude maximum edge position octal pattern (MMEPOP). The SMEPOP and MMEPOP are coded based on the sign and magnitudes of dominant edges respectively. The performance of the proposed method is tested on two benchmark databases. The results after being investigated show a significant improvement as compared to other existing methods (LBP and LBP variants) in terms of average retrieval precision (ARP) and average retrieval rate (ARR) on Corel-5000 and Corel-10000 databases.

A layered cascade optimization model is developed for facial emotion recognition.Two layered cascade-based evolutionary algorithms are proposed for feature selection.They focus on within-class and between-class variations for feature optimization.Both a neural network and an adaptive ensemble classifier are employed for expression recognition.Superior performance is shown in both frontal and 90? side-view expression recognition. In this research, we propose a facial expression recognition system with a layered encoding cascade optimization model. Since generating an effective facial representation is a vital step to the success of facial emotion recognition, a modified Local Gabor Binary Pattern operator is first employed to derive a refined initial face representation and we then propose two evolutionary algorithms for feature optimization including (i) direct similarity and (ii) Pareto-based feature selection, under the layered cascade model. The direct similarity feature selection considers characteristics within the same emotion category that give the minimum within-class variation while the Pareto-based feature optimization focuses on features that best represent each expression category and at the same time provide the most distinctions to other expressions. Both a neural network and an ensemble classifier with weighted majority vote are implemented for the recognition of seven expressions based on the selected optimized features. The ensemble model also automatically updates itself with the most recent concepts in the data. Evaluated with the Cohn-Kanade database, our system achieves the best accuracies when the ensemble classifier is applied, and outperforms other research reported in the literature with 96.8% for direct similarity based optimization and 97.4% for the Pareto-based feature selection. Cross-database evaluation with frontal images from the MMI database has also been conducted to further prove system efficiency where it achieves 97.5% for Pareto-based approach and 90.7% for direct similarity-based feature selection and outperforms related research for MMI. When evaluated with 90? side-view images extracted from the videos of the MMI database, the system achieves superior performances with >80% accuracies for both optimization algorithms. Experiments with other weighting and meta-learning combination methods for the construction of ensembles are also explored with our proposed ensemble showing great adpativity to new test data stream for cross-database evaluation. In future work, we aim to incorporate other filtering techniques and evolutionary algorithms into the optimization models to further enhance the recognition performance.

This paper addresses the challenging problem of the recognition and classification of textured surfaces given a single instance acquired under unknown pose, scale and illumination conditions. We propose a novel texture descriptor, the Adaptive Median Binary Pattern (AMBP) based on an adaptive analysis window of local patterns. The principal idea of the AMBP is to convert a small local image patch to a binary pattern using adaptive threshold selection that switches between the central pixel value as used in the Local Binary Pattern (LBP) and the median as in Median Binary Pattern (MBP), but within a variable sized analysis window depending on the local microstructure of the texture. The variability of the local adaptive window is included as joint information to increase the discriminative properties. A new multiscale scheme is also proposed in this paper to handle the texture resolution problem. AMBP is evaluated in relation to other recent binary pattern techniques and many other texture analysis methods on three large texture corpora with and without noise added, CUReT, Outex_TC00012 and KTH_TIPS2. Generally, the proposed method performs better than the best state-of-the-art techniques in the noiseless case and significantly outperforms all of them in the presence of impulse noise. HighlightsNew method for texture analysis and classification.Adaptive approach to extract the local patterns is proposed.Combining different information with the joint histogram.New multiscale scheme of the texture cue is proposed.The classification accuracy reach high rates and robustness for impulse noise.

A center sliding Bayesian design adopting orthogonal polynomials for binary pattern classification is studied in this paper. Essentially, a Bayesian weight solution is coupled with a center sliding scheme in feature space which provides an easy tuning capability for binary classification. The proposed method is compared with several state-of-the-art binary classifiers in terms of their solution forms, decision thresholds and decision boundaries. Based on the center sliding Bayesian framework, a novel orthogonal polynomial classifier is subsequently developed. The orthogonal polynomial classifier is evaluated using two representative orthogonal polynomials for feature mapping. Our experimental results show promising potential of the orthogonal polynomial classifier since it achieves both desired accuracy and computational efficiency. HighlightsProposed a center sliding Bayesian classifier with deterministic solution.Proposed to adopt orthogonal polynomials for efficient feature expansion.Developed an easy tuning mechanism for the classification framework.Provided an extensive evaluation of the proposed classifier.

Recently, finger vein recognition has received considerable attention in the biometric recognition field. Originating from fingerprint recognition, minutiae-based methods are recognized as an important branch, which attempts to discover minutia patterns from finger vein images for matching and recognition. However, the accuracy of these methods is generally unsatisfactory. One of the most challenging problems is that, the correspondences of two minutia sets are difficult to obtain resulting from the rotation, translation and deformation of the finger vein images. Another critical problem is that, the current available feature descriptors for minutia representation are weak and insufficient. In this paper, we propose SVDMM, a singular value decomposition (SVD)-based minutiae matching method for finger vein recognition, which involves three stages: (I) minutia pairing, (II) false removing and (III) score calculating. In particular, stage I discovers minutia pairs via SVD-based decomposition of the correlation-weighted proximity matrix. Stage II removes false pairs based on the local extensive binary pattern (LEBP) for increasing the reliability of the correspondences. Stage III determines the matching score of the input and template images by the ‘average’ matching degree of all their precise minutia pairs. Extensive experiments demonstrate that our work not only performs better than the similar works in the literature, but also has great potential to achieve comparable performance to other categories of state-of-the-art methods.

This paper proposes an efficient indexing technique for fingerprint database using minutiae based geometric hashing. This technique consists of two stages, known as indexing and searching. For an accurate match at the time of searching, it has proposed a fixed length feature vector built from each minutia, known as Minutia Binary Pattern. Unlike any existing geometric based indexing technique, the proposed technique inserts each minutia along with the feature vector exactly once into a hash table. As a result, it reduces both computational and memory costs. Since minutiae of all fingerprint images in the database are found to be well distributed into the hash table, no rehashing is required. Experiments over FVC 2004 datasets prove the superiority of the proposed indexing technique against well known geometric based indexing techniques.

Nowadays, the high speed (e.g., kilo-Hertz) refreshing rate of the digital micro-mirror device (DMD) has enabled superfast 3D shape measurement using the binary defocusing technique. This research finds that when the system reaches its extreme binary pattern refreshing rate, the transient response of the DMD induces a coupling effect (i.e., two neighboring patterns blend together) that may cause substantial measurement error. Since this transient response repeats itself, this systematic measurement error is substantially reduced to a negligible level when the timing between the projector and the camera is properly adjusted. Experimental results are presented to demonstrate the observed phenomena, and the success of utilizing the proposed method to overcome the problems associated with the transient response of the DMD.

The digital binary phase-shifting technique has been demonstrated for its merits over the conventional sinusoidal phase-shifting method in terms of measurement speed and simplicity. Yet, the measurement depth range is small when a squared binary method is used. Our recent research focuses on improving its measurement accuracy without sacrificing measurement speed, and increasing its depth range without losing measurement quality. This paper will summarize our recent work on the following three major areas: (a) realization of kHz 3D shape measurement with binary phase-shifting methods; (b) binary pattern improvement with pulse width modulation and binary dithering/halftoning techniques; and (c) applications of superfast 3D shape measurement techniques. Principle of each technique will be presented, and experimental results will be shown to verify its performance.

One of the most important authentication approaches is the iris recognition system (IRS), which is based on the iris of aperson for the authentication.In this paper, we propose a new iris recognition system using a novel feature extraction method. The proposed method, Neighborhood-based Binary Pattern, compares each neighbor of the central pixel withthe next neighbor to encode it by 1 if it is greater or 0 if it is lower than the central pixel. The obtained binary code is converted into a decimal number to construct the NBP image. In order to deal with the rotation problem, we propose an encoding process to obtain a rotation-invariant image. This image is subdivided into several blocks and the mean of each block is calculated. After, the variations of the means are encoded by a binary code. The obtained binary matrix is considered as feature descriptor of the iris. In the evaluation part, the CASIA iris database [10] has been used to evaluate the performance of the proposed IRS. The experiments demonstrate that the proposed method gives better recognition rate compared to the LBP method. Experimental results show also that the proposed system especially the feature extraction method gives promising results.

This paper presents a novel method to construct binary patterns for high-quality 3D shape measurement. The algorithm generates small patches using symmetry and periodicity, randomly initializes each pixels, optimizes the small patches through mutations, and finally tiles the optimized patches into full size patterns using again symmetry and periodicity. We will demonstrate that the proposed method can achieve substantial phase quality improvements over the dithering techniques for different amounts of defocusing.

We complete the classification of binary patterns in partial words that was started in earlier publications by proving that the partial word avoidability index of the binary pattern ABABA is two and the one of the binary pattern ABBA is three.

This paper presents an efficient face recognition method where enhanced local Gabor binary pattern histogram sequence has been used for efficient face feature extraction and generalized neural network with wavelet as activation function is being used for classification. In this method the face is first decomposed into multiresolution Gabor wavelets the magnitude responses of which are applied to enhanced local binary patterns. The efficiency has been significantly improved by combining two efficient local appearance descriptors named Gabor wavelet and enhanced local binary pattern with generalized neural network. Generalized neural network is a proven technique for pattern recognition and is insensitive to small changes in input data. The proposed method is robust-to-slight variation of imaging conditions and pose variations. Performance comparison with other existing techniques shows that the proposed technique provides better results in terms of false acceptance rate, false rejection rate, equal error rate and time complexity.

Grayscale digital halftoning produces bi-level representation of original continuous tone images. This process plays pivotal role for devices like printers, plasma panels, LCD displays, etc. The bi-level images can be considered as binary images where '0' and '1' correspond to black and white, respectively. This paper investigates potential of binary particle swarm optimization (BPSO) to generate faithful binary halftone patterns. The cost function addresses important characteristics of original images and pleasant visual appearance of halftone images. The paper also shows the application of pattern look-up-table (p-LUT) approach to address the high processing time of BPSO optimization and simple gradient-based edge enhancement for improved edge retention. Results are evaluated subjectively by statistical measures and psychovisual test. Results are evaluated objectively using image quality evaluation metrics as well. The comparisons with state-of-the-art techniques are also drawn. The evaluation results along with the comparisons show the competitive potential of the presented technique.

This paper proposes a high-order Texture Pattern Flow (TPF) for complex background modeling and motion detection. The pattern flow is proposed to encode the binary pattern changes among the neighborhoods in the space–time domain. To model the distribution of the TPF pattern flow, the TPF integral histograms are used to extract the discriminative features to represent the input video. The Gaussian Mixture Model (GMM) is exploited to calculate an adaptive threshold in propagation way for the histogram similarity measure to decide which part/pixel is background or moving object. Experimental results on the public databases testify the effectiveness of the proposed method in comparison to LBP and GMM based background modeling methods.

The ability to simulate and control complex physical situations in real time is an important element of many engineering and robotics applications, including pattern recognition and image classification. One of the ways to meet specific requirements of a process is a reduction of computational complexity of algorithms. In this work we propose a new coding of binary pattern units (BPU) that reduces the time and spatial complexity of algorithms of image classification significantly. We apply this coding to a particular but important case of the coordinated clusters representation (CCR) of images. This algorithm reduces the dimension of the CCR feature space and, as a consequence, the time and space complexity of the CCR based methods of image classification, exponentially. In addition, the new coding preserves all the fundamental properties of the CCR that are successfully used in the recognition, classification and segmentation of texture images. The same approach to the coding of BPUs can be used in the Local Binary Pattern (LBP) method. In order to evaluate the reduction of time and space complexity, we did an experiment on multiclass classification of images using the “traditional” and the new coding of the CCR. This test showed very effective reduction of the computing time and required computer memory with the use of the new coding of BPUs of the CCR, retaining 100% or a little less efficiency of classification at the time.