Search

The traditional image texture feature recognition method is deeply affected by the complexity of the image background and the amount of noise, resulting in a low recognition rate. To solve the above problems, an image texture feature recognition method based on rotation invariant histogram is proposed. After denoising the image to be recognized, the image texture feature is extracted and the image feature texture set is established. After the rotation invariant region of the image histogram is selected, the classifier is used to recognize the image texture features. Design a comparative experiment with the current recognition method. The experimental results show that the research method can improve the recognition accuracy by about 12.9%, and the recognition time is short and reliable.

In this paper, a scheme of facial texture synthesis and 3D reconstruction based on 2D face image and 3D model under the Unity platform is proposed. First of all, this scheme uses the ASM algorithm to extract the key points of the human face. Next, the facial features in the face image and the standard skin color image are texture blended to get the texture of fused facial image. Finally, the texture of fused facial image is mapped to the 3D face model by texture mapping, and a better personalized 3D face reconstruction effect is obtained by this method. Compared with FaceGen, a software for modeling facial photos, the experimental results show that the proposed algorithm is superior to other existing algorithms in efficiency and has certain application value.

Induction motors (IM) are the most commonly used machines in industry; hence, early diagnosis of IM faults has become a subject of great interest for many researches, since if there is a fault and the IM keep working for long time, they and their associated systems might suffer catastrophic damages. Broken Rotor Bars (BRB) are among the most common and difficult to detect faults. Many methodologies based on Motor Current Signature Analysis (MCSA) for BRB detection have been proposed in recent years, which usually require long execution time and specialized hardware and software for their implementation. Therefore, in this work, a novel and straightforward methodology for BRB detection and classification during the induction motor steady state, based on MCSA, image texture feature (contrast), and a fuzzy logic classifier is proposed. Obtained results demonstrate the suitability of the proposed approach for identifying BRB with high effectiveness.

One of the most important factors for deciding the use of wood product is knowing its strength class. So far, the wood strength has been measured using relatively expensive laboratory tests both for purchase and rent. In addition, this test has also damaged the test wood. An alternative method that can be used to determine the strength class of woods is to observe their characteristics using the naked eye. However, this is also a tough challenge because it must be done by wood experts who truly memorize and understand the characteristics of each wood class. Therefore, an image processing method is needed to identify the strength of wood easily, cheaply and not to damage test wood.The object of this study consisted of seven types of wood: Acacia (Acacia mangium), Teak (Tectona grandis), Mahogany (Swietenia mahagoni), Silk Tree (Paraserianthes falcataria), Sengon Tekik, Rain Tree (Samanea Saman) and Coastal Cottonwood (Hibiscus tiliaceus). Based on the results of specific gravity measurement, the wood strength classes of the test wood were as follows : Acacia, Teak and Mahogany were Class II, Rain Tree and Coastal Cottonwood were Class III, while Silk Tree and Sengon Tekik were Class IV.The parameters used by the author to determine the strength of wood based on image processing are the RGB color and texture of the wood image. We use two feature extraction methods, the average RGB histogram value for each color channel and the static characteristics of the Gray-Level Co-Occurrence Matrix (GLCM) method. The method used for feature selection is Recursive Feature Elimination Cross Validation (RFECV). While for data classification we use five algorithms: K-Nearest Neighbors (KNN), Linear Discriminant Analysis (LDA), Decision Tree (CART), Random Forest (RF), and Support Vector Machine (SVM). The results of the qualifications showed that the highest accuracy was obtained by the KNN of 95%. Therefore, this method can be used as an alternative to determine the strength of wood.

A sign language is a mode of communication in which the intent of the message or the message itself is conveyed through body postures or the movement of the parts of the body like head, eyebrows and cheeks etc. Every expression is distinct and has distinguishable parts of language, its grammatical content fully displayed through gestures. There are more than three hundred sign languages used in the world. The baby sign language is a method of communication between the mothers and their toddlers by means of gestures, clearly expressing their emotions and desires. In the present research work study of the existing literature has been carried out and then prepared a data set of still jpeg images for 60 odd baby signs, performed GLCM(Gray Level Co-Occurrence Matrix) based feature extraction, then performed classification of gestures using KNN and Random Forest based machine learning algorithms. A classification accuracy of 73% has been achieved on the dataset prepared.

Matching in high-resolution synthetic aperture radar (SAR) images while being more complicated compared to optical images, is especially important due to its numerous applications. The main aim of current research is to determine improvement of SAR image matching process by deploying texture analysis using gray level's co-occurrence matrix (GLCM). Three parts of the pair of TerraSAR-X images are used to implement the methodology. The results show that for some areas with low texture, the conventional image matching algorithm is not able to detect corresponding points, while using other textural features in image matching process leads to improvement in quantity of acceptable matched points.

Efforts are afoot to design better context-aware human-computer interaction techniques that have knowledge of both their surrounding and the affective state of the user. One of the most important nonverbal behavioural cues for affective human-machine interaction is laughter. Automatic detection of laughter is an interesting, yet challenging problem, which in recent years has gained increased attention from both the academic and industrial communities. The majority of existing laughter detection systems rely on either audio or video modalities. Humans, however, typically rely on audio-visual cues during conversation and/or interaction, thus it is expected that improved results can be achieved if both modalities are used. In this work, we propose a multimodal framework that analyzes audio and video channels separately, then fuses their decisions. Conventional speech spectral and prosodic features are used, whereas new multi -scale multiresolution binarized statistical image features are proposed due to their improved expressive power. Experiments with the publicly available MAHNOB Laughter database show that decision level fusion based on support vector machine classifiers leads to improved performance over single modality approaches, as well as over previously-proposed methods, all whilst requiring just a fraction of the computational power.

In a complicated cloud storage environment in which users upload a large number of files everyday, in order to better solve the challenge of inefficient malicious detection and weak adaptability of multi-platform detection in the traditional way, we propose a malicious file detection method which is based on image texture analysis and BP neural network algorithm. By combining the technology of image analysis and the malicious file detection, the malicious file is converted into grayscale image, the GLCM (Ground Launched Cruise Missile) and the GIST (Generalized Search Trees) algorithms are used to extract the texture features, and the BP neural network algorithm is then used for learning and training. In this paper, we propose and implement a malicious file detection system by means of image texture extraction. Through the experimental analysis on a large number of virus samples from the well-known VirusShare project, the experimental results show that our proposed approach has the characteristics of fast speed, high adaptability and high accuracy.

A statistical classification method using Mel cepstrum and image texture is introduced to passive sonar target recognition. Image texture features are extracted by regarding Mel frequency cepstrum coefficients (MFCC) as a digital image. Bayesian formula and Markov chain are applied to the recognition algorithm. The classification experiments are carried out for three different kinds of targets based on practical data. The experimental results show that the proposed method is useful.

The widespread penetration of counterfeit integrated circuits (ICs) is not only a major threat to the electronic goods supply chain, but also constitute a great threat to national security. Image processing based counterfeit IC design techniques are promising, but currently often suffer from high computational complexity and requirement of expensive image acquisition infrastructure. We describe two techniques based on image texture analysis to automate the process of counterfeit IC detection. The first method employs local textural feature identification to detect counterfeit ICs. The second method includes identification of counterfeit ICs by segmenting the image into regions of different textural features using texture filters. The first method is of lower computational complexity compared to the segmentation method, but the second method is capable of blind identification in the sense that it does not require knowledge of the textural features of a golden IC sample. Our experimental results show that these methods have high detection accuracy, even for images acquired using ordinary digital cameras and low-end digital microscopes.

Specific emitter identification (SEI) aims at discerning radio emitters using external measurement features called the radio frequency (RF) fingerprints of the signal. SEI is widely used in intelligent gathering, network intrusion detection, Cognitive Radio and so on. In this paper, we proposed a novel SEI method based on steady signals, which is more practical. We firstly get the time-frequency-energy distribution (TFED) of the signal using Intrinsic Time-scale Decomposition(ITD), which is more suitable for non-stationary signal than empirical mode decomposition (EMD). Secondly, we transform the TFED of the signal into a gray image and several image texture features are extracted from histogram statistics and Gray-Level Co-Occurrence Matrix (GLCM) of the image. We treat these texture features as the signal's RF fingerprints to identify the specific emitter. Measured ship signals and simulated signals are used to evaluate the performance of the proposed method and another two methods are compared with it. Experimental results demonstrate that the proposed method is more effective.

Public transport vehicle scheduling system plays a very important role in relieving urban traffic pressure, and it's now still a difficult problem. In order to solve the problem effectively, we propose a bus scheduling method based on video processing algorithm in this paper. Firstly, we decode the H264 bare stream which receive from the bus in real-time, and according to the prior knowledge, a certain number of training samples with different density levels were collected, and then extract the improved gray-level co-occurrence matrix and color entropy features of training samples to train the classifier model. Finally, crowd density estimation result can get by processing the real time vehicle video using the trained model. Combined the result of density estimation with vehicle status information, we can realize the vehicle scheduling effectively.

To determine the real-time traffic state accurately in road network or intersections, traffic state identification method is proposed based on image processing technology. During the image process, by analyzing the image texture features, the multi-scale block local binary patterns are taken as the features. The road traffic state identification model is established based on support vector classification. A road section is taken as an example, the images of three traffic states (free traffic flow, steady traffic flow, and forced traffic flow) are selected to test the established traffic state identification model. This method can be applied to determine the real-time traffic conditions, which can provide information for the traffic management decisions.

A sound signal produces a unique texture which can be visualized using a spectrogram image and analyzed for automatic sound recognition. In this paper, we explore the use of a well-known image texture analysis technique called the gray-level co-occurrence matrix (GLCM) for sound recognition in an audio surveillance application. The GLCM captures the distribution of co-occurring values at a given offset. Unlike most other similar research which derive features from the GLCM, we use the matrix values itself to form the feature vector with analysis carried out in subbands. When compared to a baseline feature from related work, the proposed spectrogram image texture feature (SITF) gives marginally lower results under clean and high signal-to-noise ratio (SNR) conditions but significantly better results are achieved at low SNR, where the baseline feature was seen to be less effective.

This paper contains study and review of various techniques used for feature extraction and texture classification. The objective of study is to find technique or combination of techniques to reduce complexity, speed while increasing the accuracy at the same time. Here we are studying and reviewing the three feature extraction methods: Gray Level Co-occurrence Matrix (GLCM), Local Binary Pattern (LBP) and Gabor filter method. Also classification methods like KNN, SVM, evolving fuzzy neural network (efunn), genetic algorithm and higher-order statistics. These are used on the texture datasets Brodatz, CUReT, VisTex and OuTex for the experimental purpose. In this paper, we present a comparative study of image texture classification techniques which are very much help full for image classifications

In this paper, turning workpiece surface images were captured by the image acquisition system composed of digital microscopy, high-resolution camera and computer. the 14 texture feature parameters based on gray level co-occurrence matrix (GLCM) were extracted by using the method of statistical analysis, The variation between each texture parameter and arithmetic mean deviation (Ra) was investigated, thus the roughness of turning workpiece surface was evaluated qualitatively, the relationship between texture characteristic parameters and roughness evaluation index Ra was analyzed by using multiple regression method, and the linear and nonlinear regression testing model were bulit up. The result shows that the two models have good detection effect, and the nonlinear model performance better than the linear model.

This paper investigates texture feature capabilities from fundus images to differentiate between diabetic retinopathy (DR), age-related macular degeneration (AMD) screening and normal. Our proposed method using improvement of local binary pattern (LBP) with calculation of LBP original value and magnitude value of fundus images. This method is compared with Local Line Binary Pattern (LLBP). In this study, four experiments (DR-Normal, DR-AMD, AMD-Normal, Multiclass) were designed for two databases, DIARETDB0 database and STARE. Kernel PCA is choosed as feature selection method, and three classifiers are tested (Naive Bayes, SVM, and KNN). The experimental results show that our proposed method has higher accuracy than LLBP, with accuracy of binary classification 100% for DR-Normal and AMD-Normal. While, multiclass classification (DR-AMD-Normal) achieves an accuracy 80–84%. These results suggest that our proposed method in this paper can be useful in a diagnosis aid system for diabetic retinopathy.

This paper proposed a texture's feature of gray-level co-occurrence matrix (GLCM) as a calibrator of the sea wave height. Texture in Synthetic Aperture Radar (SAR) images represent the surface roughness of the sea or the wave height. As an archipelago country which has 70% sea of entire region, Indonesia needs and must have early warning system to provide transportation and fisherman safety circumvent the accident. Object of research are the P band full polarization Synthetic Aperture Radar (SAR) image acquired from coastal of East Kalimantan. These polarization are; horizontal-horizontal, horizontal-vertical, vertical-horizontal and vertical-vertical. Extraction of feature is done under observation angle 0° and adjacent pixels is 1. The results showed homogeneity is a consistent and toughest feature in horizontally-vertically polarization and vertical-vertical polarization and this feature can be used as a calibrator.

An evaluation method based on the normal gradient feature of image texture is proposed to describe the imaging sharpness of stereo objects. The method can be summarized by four steps. Firstly, median filter is utilized to remove noise and threshold segmentation is applied to binary the image. The second step is edge detection which is realized by Canny algorithm. In the following procedure, the gray value in a fixed range along the normal direction of the edge is extracted for least square method curve fitting. Lastly, the average index coefficient D value is taken as the sharpness evaluation index for the image. In this paper, the sharpness evaluation method based on the normal gradient feature of image texture is dedicated to the local region of interest evaluation in different planes, so as to achieve the accurate evaluation of the sharpness requirements for specific needs, such as high-definition required object recognition. Experiments show that the proposed method can accurately evaluate the image definition. The method also has the advantages of single peak, high sensitivity and strong stability, which may shade lights in local definition evaluation and high-accuracy object recognition.

In the interior of image texture analysis methods, many types, behavior, strategies, modeling, and characteristics are employed. Several surveys and reviews are illustrated in the literature regarding image texture and feature extraction using an agent and multi-agent-based systems. Nevertheless, few determinations are made in the context of the image texture feature extraction. Hence, the overall aim of this paper is to provide an overview of the image texture and feature extraction concepts depicted in the literature for characterizing regions in images by exploring the role and strategies of the agent and multi-agent systems (MAS). This paper tackles two behavior types, reactive agent (RA) and goal-directed agent (GDA), in terms of the natural surroundings of agent and multi-agent communications. In terms of MAS modeling, this paper tackles three types, including facilitator agent architecture (FAA), mediator agent architecture (MAA), an autonomous agent architecture (AAA). While for the MAS characteristics, five types are presented: interaction, coordination, cooperation, collaboration, and collectiveness. Finally, a discussion for agent and multi-agent-based feature extraction in image analysis is presented by comparing the strategies and the frequent use of the strategies in the literature. Based on this review, most of the feature extraction multi-agent-based systems rely on either coordinating or emergent-based strategies, while feature selection agent-based systems rely on collaborative strategies. However, there are several aspects that we can consider to be classified as agent-based strategies. This review develops a classification scheme for systems used in specific tasks of feature extraction.

Camouflage plays an indispensable role in modern military. Generally, a camouflage pattern includes a plurality of shapes units, which are split by elemental colors. The fixed camouflage pattern is low adaptability and concealability for changeable battlefields. Moreover, the conventionally camouflage design is time and resource consuming for manual drawing. In this paper, we propose a dynamic camouflage synthesizing method. Firstly, the texture patterns of one class of battlefield images is extracted using convolutional transfer network. We use 3*3 convolution kernels to extract texture features. And the covariance matrix is used as loss function to calculate the loss of different image texture features. Colors construction and their distribution of the specified battlefield images is extracted according to a clustering-based algorithm. Finally, the statistical color units are embedded into the texture patterns. We assess the adaptability and the concealability of synthesized camouflage using Eye Tracking based criteria. Comparing to the People’s Liberation Army Type 87 woodland pattern and the digital camouflage pattern synthesized using our previous method, testing results of our synthesized camouflage patterns were better in different saccade indicators. This demonstrates camouflage pattern synthesize proposed got better concealment and the validity of proposed method. Besides, we also compared our method with local binary pattern for extracting texture features. The experiments results indicated the method proposed had better consistency to the scene images.

Current steganalyzers based on deep learning mostly adopt wider or deeper designs to improve detection performance. However, an overly complex network increases the training cost and is not conducive to its expansion and optimization. Moreover, steganalysis pays more attention to high-frequency information corresponding to the image texture. However, the deeper the network, the more likely it is to learn low-frequency information corresponding to the image content, which is inconsistent with the goal of steganalysis. In response to these problems, a multi-frequency residual deep convolutional neural network for steganalysis of color images called MFRNet is proposed in this paper. We apply the idea of multi-frequency analysis to steganography detection for the first time, effectively controlling the network scale. By designing columns of different depths, it can learn different frequency components of steganographic noise at the same time. The detection performance is better than the existing networks that only learn a single frequency component of steganographic noise at the same depth. Therefore, it can achieve a good detection performance with a lighter architecture. In addition, by designing residual basic blocks with different residual shortcuts, different scales of steganographic noise residuals can be calculated at the same time, which can effectively suppress the interference of image content, so as to better reduce the impact of steganography algorithm mismatch and payload mismatch than the existing methods. The experimental results on PPG-LIRMM-COLOR showed that the proposed MFRNet outperformed the state-of-the-art model WISERNet.

The color and contrast of objects in the image will be affected by meteorological factors, especially rain and snow will block part of the image, which will change the information contained in the image. Image restoration under bad weather conditions has practical application value. At present, most of the research focuses on the removal of fog, and the research on complex rain and snow is relatively less. Rain has more prominent features in gradient domain, and it is more distinct from non-rain image texture. In this paper, Generative Adversarial Networks is used to combine the information of image in gradient domain and spatial domain to get better performance of rain removal. Gradient aided coding is used in the generator to generate depth features that are more conducive to rain removal. In the discriminator, the gradient is used as an additional input to provide more recognizable rain and non-rain information, which enhances the discriminator’s ability to distinguish the image generated by the generator and the ground truth. By modifying the network structure of the expanded spatial pyramid pooling (ASPP), the abnormal rain removal results produced by the generator are reduced. Experimental results show that the proposed method improves the performance of rain removal and the visual quality of the generated image.

There are lots of image data in the field of remote sensing, most of which have low-resolution due to the limited image sensor. The super-resolution method can effectively restore the low-resolution image to the high-resolution image. However, the existing super-resolution method has both heavy computing burden and number of parameters, which greatly limits the super resolution method in the mobile terminal. For saving costs, we propose the information multiple distillation network based on feedback mechanism (Feedback-IMDN), which considers the feedback mechanism as the framework to attain lower features through high-level refining. Further, for high-level feature extraction, we use the information multiple distillation blocks (IMDBs) to carry out hierarchical feature extraction with the method of course learning in the case of a small number of parameters. Compared to other state-of-the-art lightweight algorithms, our proposed algorithm can reach convergences more rapidly with fewer parameters, and the performance of the network can be markedly enhanced on the image texture and object contour reconstruction with better peak signal-to-noise ratio (PSNR) and structural similarity (SSIM).

The traditional binocular vision stereo matching replaces and responds to the maximum point as the feature point, which is prone to uneven distribution of feature points, especially when the distribution of image texture strength is large. The Euclidean distance needs to be calculated when the feature points are matched, and the constraints considered are basically position relationships, ignoring the color block constraints. Based on the image texture and color distribution characteristics, the texture strength area and RGB color block are respectively calculated, and the image is segmented. The multi threshold feature points detection is used, and the matching points are calculated according to the color segmentation template. The experimental results show that compared with the traditional ORB algorithm, the proposed algorithm improves the uniformity of feature points and the matching accuracy.

Aiming at the problem that the image texture is complex, the grayscale feature is very close to the wharf, and the accurate segmentation is very difficult in the process of port ship detection in high-resolution optical remote sensing image, a port ship detection method combining visual significance model and center dark channel priori is proposed. The watershed segmentation algorithm based on spectral intensity and image texture information achieves land and sea segmentation. The land area is masked to suppress the interference of complex land background. Extracting fitting line and smoothing port contour line. According to the characteristics of the straight shape of the port and the straight shape of both sides with the ship, the center line point of the line is defined to locate the port, and the set of candidate areas of interest to the port is obtained. Combined with the visual model and the dark channel prior feature, the ship initial significance map and the central dark channel map were fused to complete the ship target detection. Finally, through the bow characteristics of the ship, the adjacent ship is separated to eliminate false alarm and improve the detection accuracy. Experiments show that the algorithm is robust, has a good detection effect, and its accuracy and recall rate are greatly improved.

A serious issue in developing automatic palmprint verification systems is the accurate and robust palm image cropping and feature extraction in order to produce high recognition accuracy. Feature extraction such as local binary patterns (LBP) can be used to describe palm image texture characteristics since the palm print image has a rich number of texture features. However, using LBP only to represent a local feature is not enough information to categorize the palmprint image. Our paper proposes a combination of LBP with KAZE feature detection to extract the features exist in palm image after applying the LBP. KAZE feature detection is able to select the most discriminant features which produce better class separation in the feature space. KAZE is used to select the most dominant features exist in LBP image descriptor. The performance evaluation is performed using benchmark PolyU dataset. The best result of the experiment work using PolyU palmprint database is 98% recognition rates.

Local binary pattern (LBP) is an important image texture feature extraction method. This paper proposes a combination of adaptive threshold and directional local image feature representation, and to realize the image retrieval. This local combination feature is composed of two parts: 1)The standard deviation of the gray value of all pixels in the local neighborhood of the image and the gray value of the center pixel are used as the comparison threshold for encoding. This threshold is adaptively changed following the grayscale changes in the neighborhood, and can indicate the severity of the gray scale change; 2)directional local pattern: by comparing the grayscale changes of local neighborhood pixels in different directions, discrimination and binary encoding are performed to indicate the local texture features directionality. Finally, the two features are combined to form the local features of the image, and the features are used for image retrieval. Experiments on the Corel-1k database show that this method has a better image retrieval effect than existing local feature representation methods.

The positioning accuracy of the gravity-aided navigation is closely related to the selection of local gravity maps. In this paper, the local gravity maps are converted into 8-bit gray images. The navigability features of the local gravity maps are extracted by using the image texture feature analysis methods, such as the gray histogram complexity, the Sobel operator, and the gray level co-occurrence matrix, and the navigability comprehensive evaluation of each local gravity map is obtained by using the projection pursuit model. According to the nebula model, a gravitation field algorithm is proposed. The gravitation field algorithm, genetic algorithm, and firefly algorithm are used to obtain the optimal projection direction of the projection pursuit model, respectively. We compared the optimizing results and gave the navigability evaluation of local gravity maps that provide the basis for the selection of local gravity map. The comparison results show that the gravitation field algorithm has the best performance in obtaining the optimal projection direction, and the contributions of the navigability features in the navigability evaluation are most equal. Under the same experimental conditions, the local gravity map selected by the method proposed in this paper has the highest positioning accuracy and the best matching track.

Facade structural features can represent the overall framework of buildings. However, structural features extracted by the current methods contain a quantity of trivial unstructured information. In this study, we proposed an accurate extraction method for structural features of building facades through texture fusion. By performing texture fusion on building facade images, the interference of textural elements on structural feature extraction could be eliminated. After texture fusion, the line segment detector (LSD) algorithm is used to extract the structural features from the building facade images, and random sample consensus (RANSAC) is used to improve the continuity of structural features. The accuracy and effectiveness of the proposed method was demonstrated by comparing results with the state-of-the-art methods, such as LSD, MLSD, CannyLines, and MCMLSD. Value setting of three important parameters is discussed in detail. The imagery facade features extracted through the proposed method provide valuable support for many fields, such as image feature registration and 3D reconstruction of building surfaces.

This paper proposes an innovative method in the investigation and analysis of real time transformer oil images at different temperatures along with different ages using haralick image texture features. Haralick texture feature method based on Gray-Level Co-occurrence Matrix (GLCM) used in this paper to enumerate the spatial relation between the neighborhood pixels in an image. A theoretical examination performed on oil test images to characterize its textural properties. The statistical features extracted for original as well as filtered transformer oil image at different temperatures, and features of one year to twenty five year aged oils determined. The results through this analysis indicate the identification of significant textures of the test images. The experimental results demonstrated that texture feature extraction derived from the haralick features realize a new technique in the analysis of transformer oil images under different ages as well as operating conditions.

As one of the Minimally Invasive Surgery (MIS), laparoscopy is widely applied in gastrointestinal surgery. Benefits from its minimally invasive procedure and fast postoperative rehabilitation for patients. However, it is a huge challenge for surgeons to operate slender surgical instruments under a limited field of view taken by laparoscopy which requires the surgeon to have rich surgical experience. In laparoscopic surgeries, an accurate 3D reconstruction mode with internal anatomy structure and laparoscopic positions in the abdominal cavity can effectively help the surgeon to reduce the dependence on surgical experience. In addition, reconstructing the structure of the surgical scene is also a key step for data registration in surgical navigation and augmented reality surgery. In this paper, a novel Simultaneous Localization and Mapping (SLAM) method based on feature patch tracking by Kernel Correlation Filter (KCF) is proposed for 3D dense point clouds reconstruction, which is called KCF-SLAM. It reconstructs the laparoscopic surgery scene accurately and densely under stereo laparoscopic conditions. The proposed method is validated on a public in-vivo data set captured by a stereo laparoscope. Compared with the feature-based SLAM, the proposed method can work efficiently when the image texture is missing or insignificant. The results suggest that the proposed method can reconstruct the dense point cloud of laparoscopic scene stably and accurately.

Biometrics identification technology has gradually become a research hotspot in the field of information processing. As a step of biometrics identification technology, feature extraction processing plays a vital role. Aiming at the shortcoming of existing feature extraction algorithms are vulnerable to noise interference, large amount of calculation, high dimension and incomplete features, this paper proposes an improved MB-LBP feature extraction algorithm based on reduced-dimensional HOG. The algorithm uses MB-LBP to extract texture features of the image, and uses reduced-dimensional HOG to extract edge features. Through serial fusion, complete image features are formed. The proposed algorithm is verified by experimental simulation comparison with HOG feature extraction, dimensionality reduction HOG feature extraction and MB-LBP feature extraction. The algorithm in this paper has the characteristics of strong anti-interference ability, low dimension and complete features in feature extraction.

Adaptive PRNU Denoising (APD) is one of the strongest forms of state–of–the–art counter–forensic attacks, that has largely succeeded to prevent Photo Response Non–Uniformity (PRNU) based source identification of digital images. A major way to thwart such attack would be finding a successful mapping of a counter-forensically modified image, back to its source. In other words, such attack will be rendered obsolete if accurate source identification can be made possible with the counter–forensically modified images. For the first time, we perform a feature based identification of image sources, working directly with counter–forensically modified images, without trying to recover their authentic versions. We quantize the effects of APD on forensic source identification techniques, and explore how image texture helps in source identification of APD counter–forensic images. We propose Local Binary Pattern (LBP) extraction on APD image texture layer, the key component of our work, enabling successful counter-forensic image source attribution. Our experimental results prove that the proposed methodology is highly effective to mitigate the threats posed by APD, one of the strongest counter-forensic attacks on camera model identification today.

The textured regions embedded a watermark have better visual quality than the smooth regions in an image. To take advantage of the image texture being easy to hide the watermark, accurately locating the regions in an image with rich texture is significant. This paper proposes an optimized local image watermarking algorithm combining feature point and texture. The SURF feature points extracted from an image with moderate scales are selected to obtain initial watermark embedding regions. A scoring scheme by comprehensively analyzing texture, scale, and position of a region is proposed to evaluate the regions around each initial embedding region, and select the regions with the highest score from them to constitute the candidate embedding regions. Finally, the same watermarks are embedded in multiple non-overlapping embedding regions to guarantee the imperceptibility and improve the robustness. The simulation experiments on 100 images show the superiority of our proposed method compared with the state-of-the-art method in terms of imperceptibility and robustness.

Power transformers are important assets in power grid. Winding deformation is one of main fault types of transformer internal failures. The accurate diagnosis of transformer winding deformation is significant and meaningful. In this study, an improved method of classifying winding deformation types is proposed. The polar plot is first plotted by using the amplitude and phase information of the measured frequency response analysis (FRA) traces, then the digital image processing technology is used to extract three image texture features from polar plot, and three support vector machines (SVMs) are independently trained by using the extracted texture features. As a result, a strong classifier is eventually obtained by combining the three trained SVMs, for fault type classification and recognition. The proposed method is verified on the experimental FRA data obtained from an actual model transformer, which demonstrates that the proposed method has more excellent performance compared with the traditional method based on the FRA trace and single SVM.

Image style conversion refers to a technique that uses an algorithm to learn the style of a famous painting and then applies it to another picture. Current generative adversarial networks have been widely used for image style conversion. However, Cycle Generative Adversarial Networks (CycleGAN) is not very clear in processing image textures. This paper proposes a method of adding a Local Binary Pattern (LBP) algorithm and adds a cyclic LBP algorithm to the generator that generates the adversarial network to improve the effectiveness of the cyclically created adversarial network to extract image texture feature content. Experimental results show that adding the LBP algorithm can improve the image quality and make the trueness and falseness between the input picture and the picture generated by cyclegan closer.

Filtering of spam image-based email remains a major challenge for researchers. This paper presents a proposed work which is based on several facts such that spam images containing a large percentage of text which has characteristics or features different from other types of images. In addition to that, there is much similarity between the features of these images. These facts can be used to distinguish text regions spam images from others. A hybrid method based on combined features vector from text regions and features of the image is proposed. Two types of features are extracted. The first features extraction method is the local binary pattern (LBP) with extricating the image texture features directly, while the second is utilised to extricate features of image text regions only. The extracted features are used in individual and combination style in order to learn classifiers at the training stage. A one-class KNN classifier and two-class KNN classifier are applied separately. Each classifier was used in three fashion, with the text-regions features, with texture features in the image, and with merging both those features respectively. Experimental results showed that the appropriation of both image and text features together improves the effectiveness of the classification concerning the case in which only image or text features are used.

For single-features based on finger vein image texture features (LBP), image gradient features (HOG), and other features with poor image contrast, the feature extraction effect is not ideal, and the sparse representation classification method based on residual size classification is used in low-quality images. This paper proposes a method for finger vein recognition based on improved weighted sparse representation. In order to reduce the heterogeneity of the samples participating in sparse reconstruction, this paper weights each training sample based on the sum of existing sparse coefficients as a classification basis to reduce the heterogeneity of samples participating in sparse reconstruction. On this basis, this paper uses the sum of intra-class coefficients instead of reconstruction errors as the basis for classification judgment. It has a high recognition rate for finger vein images with poor contrast.

In this paper, a new feature descriptor, named correlated primary visual texton histogram features (CPV-THF), for image retrieval is proposed. CPV-THF integrates the visual content and semantic information of the image by finding correlations among the colour, texture orientation, intensity, and local spatial structure information of an image. Based on texton theory, box-shaped structural elements are designed for image texture analysis. The colour, texture orientation, and intensity feature histograms that are proposed in CPV-THF are represented by correlated attributes of the co-occurrence matrix. The performance of the proposed descriptor is compared with those of state-of-the-art texture, colour, shape, and local-pattern-based CBIR descriptors. Experiments are conducted on multiple standard natural image datasets, namely, corel1k, corel5k, and corel10k. Experimental results indicate that the proposed descriptor outperforms various state-of-the-art descriptors, such as GLCM, MTH, MSD, MS-LSP, SED, STH, and MTSD.

A biomedical image retrieval technique using novel multi-scale pattern based feature is proposed. The introduced technique, in each scale, employs arbitrary shaped sampling structures in addition to a classical circular sampling structure in local bit-planes for effective texture description, and named as the multi-scale local bit-plane arbitrary-shaped pattern (MS-LBASP). The proposed feature descriptor first downsamples the input image into three different scales. Then the bit planes of each downsampled image are extracted and the corresponding bit-planes are locally encoded, characterizing the local spatial arbitrary and circular shaped structures of texture. The quantization and mean based fusion is utilized to reduce the features. Finally, the relationship between the center-pixel and the fused local bit-plane transformed values are encoded using both sign and magnitude information for better feature description. The experiments were conducted to test the performance of MS-LBASP. Two benchmark computer tomography (CT) image datasets and one magnetic resonance imaging (MRI) image dataset were used in the experiments. Results demonstrate that the MS-LBASP outperforms the existing relevant state of the art image descriptors.

Finding the edges of the images is an important process, especially medical images, to access the image data accurately to find the objects in the image, especially in the analysis and diagnosis processes Detecting edges in noise images is still a very difficult process. Several methods have been introduced to solve the problem, such as Sobel, Prewitt, and canny, but it is still difficult to process noisy images. The current research presents a proposed system designed to detect edges in a medical image and has been applied to medical MRI images. The proposed system consists of two phases, the first phase is the initial processing phase and this process aims to prepare the image for the processors, the second stage of the proposed system, it is the edge detection stage. The edges of the images entered into the system are detected by finding an edge map, and depending on the texture characteristics of the image, as the method used for edge detection is the Laws method and according to the following steps: 1. Create two-dimensional filters.2. The process of circumventing images and laws filters. The efficiency and effectiveness of the proposed system were analyzed through the results of experiments that showed that the performance of the system is very good and gives excellent and accurate results, by comparing the results of the proposed system with traditional methods of edge detection. Results like sobel, prewitt and canny showed that the proposed system gives a better and faster performance.

In order to improve the recognition rate of surface texture defects of machined parts, a machine vision-based method for identifying surface texture defects of machined parts is proposed. By analyzing the characteristics of the detection content of small parts, the image is filtered, the improved edge detection algorithm is used to accurately extract the edge position of the surface, the image is divided into different regions by the edge, the candidate defects are screened, and the BP network is finally used Algorithms classify and mark defects. The experimental results show that the method can stably recognize and detect the surface defects of small parts, the defect recognition rate is up to 97.05%, and the detection speed is also fast, reaching the expected detection target.

When traditional information extraction methods are used to extract image feature points, image denoising degree is not sufficient, and there are still small scale noise points in detail images, which leads to the spatial dimension error of information extraction. To solve this problem, a new method based on global optimization texture feature is proposed to extract the details of double lift image. Using laser scanning measurement technology, the data of double lift image is collected, the point cloud data of image is preprocessed, and the detail feature points are extracted. The texture sequence of feature points is obtained by using the global optimization texture feature technology, and the detail color and texture are mapped into the grid to obtain the detail information of the lifting image. The experimental results show that the spatial dimension error of this method is smaller than that of the two traditional methods, which makes the extracted information meet the requirements of high precision and increase the reliability of information extraction.