Search

Highlights • A novel Multi-Level Semantic Appearance Representation for person re-identification is proposed. • We combine the low-level and mid-level representations for appearance modelling. • A score fusion based on the weighted sum is adopted. • Promissing re-identification performance on challenging dataset. Abstract Person re-identification has become a hot research topic due to its importance in surveillance and forensics applications. The purpose is to find the same person from disjoint camera views at different times and locations. Most of the existing approaches try to re-identify a target by modeling his appearance characteristics based on a single modality, i.e. low-level feature or mid-level feature representation. In this paper, we proposed a novel Multi-Level Semantic Appearance Representation that combines these two complementary characteristics for people appearance modeling. The low-level representation relies on the Multi-Channel Co-occurrence Matrix descriptor extracted from the salient body parts to model the fine characteristics of the appearance in terms of color and texture information. The mid-level representation relies on the Semantic Body Traits descriptor to model the semantic description of the appearance in terms of clothes and accessories patterns. A score fusion method based on the weighted sum was adopted to fuse these characteristics that reflect the impact of each separated modality. The experimental results proved the effectiveness and benefits of the proposed approach in terms of performance and computational complexity through the achieved comparisons with the state-of-the-art approaches using the VIPeR benchmarking dataset. [ABSTRACT FROM AUTHOR]

In this paper, we propose a novel rotation invariant and noise tolerant descriptor for texture classification. This descriptor is based on circular statistics of Fourier phase. This one is recovered from the third order spectrum namely the bispectrum, known to provide invariance properties and to preserve phase information. The computational complexity of two dimensional image is reduced using Radon transform. At first, the input image is decomposed into a set of 1D radon projections. Then, for each projection the bispectrum is computed and the phase vector is recovered. Features vectors contain circular statistics of each phase vector recovered from bispectrum of each 1D projection. The proposed descriptor is evaluated on three test suites from the database “Outex” and compared with three descriptors also based on the phase. According to the classification experiments, our descriptor achieves highest rates under noise, illumination and rotation changes. [ABSTRACT FROM AUTHOR]

In this paper, an efficient multiresolution method for texture classification based on anisotropic diffusion and local directional binary patterns (LDBP) is proposed. The method focuses on recognizing the most dominant LDBP descriptors that characterize the texture in an image by analyzing the effect of neighborhoods and radial distance of LDBP on texture classification. The texture descriptors are evaluated and compared on four texture datasets, namely, Brodatz, Oulu, VisTex and Kylberg. The discriminating power of multiresolution LDBP descriptors is assessed in classification experiments using k-NN classifier. The experimental results show that the proposed approach based on multiresolution LDBP descriptors and anisotropic diffusion yields better classification accuracy with low computational cost. The proposed method is then used in wood identification. The method is tested on fourteen texture images with varying color tones and number of samples per tone. The experimental results demonstrate the effectiveness of the proposed method in achieving the improved classification accuracy. [ABSTRACT FROM AUTHOR]

Deep learning has established many new state of the art solutions in the last decade in areas such as object, scene and speech recognition. In particular Convolutional Neural Network (CNN) is a category of deep learning which obtains excellent results in object detection and recognition tasks. Its architecture is indeed well suited to object analysis by learning and classifying complex (deep) features that represent parts of an object or the object itself. However, some of its features are very similar to texture analysis methods. CNN layers can be thought of as filter banks of complexity increasing with the depth. Filter banks are powerful tools to extract texture features and have been widely used in texture analysis. In this paper we develop a simple network architecture named Texture CNN (T-CNN) which explores this observation. It is built on the idea that the overall shape information extracted by the fully connected layers of a classic CNN is of minor importance in texture analysis. Therefore, we pool an energy measure from the last convolution layer which we connect to a fully connected layer. We show that our approach can improve the performance of a network while greatly reducing the memory usage and computation. [ABSTRACT FROM AUTHOR]

This paper examines 2D multifractal model based on operations taken from mathematical morphology. Firstly, there is given an extension of the existing multifractal model based on Legendre transform and dilation and complementary model using erosion is investigated. Two additional methods are proposed by introducing a new definition of local multifractal measures and two realizations of coverings of local dimensions images used for calculation of global dimensions. One of local measures is implemented using dilation and the other using erosion. Derived multifractal spectra are used as descriptors of textural images. Experiments conducted on referent textural datasets and in comparison to state of the art descriptors provide classification results of proposed descriptors that exceed performance of both multifractal and non-fractal descriptors from literature. [ABSTRACT FROM AUTHOR]

In traditional iris recognition systems, majority of researches improves the recognition performance for frontal images by ignoring the challenging issues including corneal refraction, 3D iris texture, limbus occlusion, and blur. When comparing images from the same angle such as frontal, all the challenging effects have similar distortions (i.e., corneal refraction and 3D texture) or minimal effects (i.e., limbus effect and blur) on all iris images. However, in off-angle iris recognition they have a significant negative impact on the accuracy and performance and they require additional treatments. In this paper, we first investigate how eye structures related iris recognition affects the performance of iris biometrics for different gaze angles and then quantify the effect of gaze angle on the inter-class and intra-class Hamming distance distributions. Based on our results from real images, as gaze angle of the probe image increases, the Hamming distance scores increases in intra-class distribution. We further found that average Hamming distance of the inter-class decreases as image captured from stepper angles and the distribution moves towards the intra-class distribution that causes an increment in the false-match-rate due to the existence of less area in the segmented off-angle images compared with the frontal images. [ABSTRACT FROM AUTHOR]

The iris segmentation module plays a crucial role in iris recognition system as it allows to define the exact iris texture region in the image of the eye. Usual iris segmentation methods tend to fail on challenging eye images captured in less constrained environment or at-a-distance. In this paper, we propose a new robust model to segment degraded iris images. Its main characteristics are as follows: (1) we explore the use of advanced statistical model for unsupervised iris segmentation and more particularly, we focused on Hidden Markov Chain. (2) Novel adequate image scanning procedure and initialization step for implementing this model are developed. (3) The implementation of the proposed model can be performed on reduced image resolutions allowing limiting the processing time without degradation of the performance. A novel recognition system can therefore be obtained by adding this unsupervised iris segmentation module as a preprocessing in the open-source recognition model OSIRIS-V4. Extensive experiments on two large near infra-red databases ICE2005 and CASIA-IrisV4-distance demonstrate a significant improvement of the recognition performance with this novel system compared to OSIRIS-V4 and recent region-based iris verification systems, showing this way the potential of such statistical models for iris recognition. [ABSTRACT FROM AUTHOR]

Local Parameter Histograms (LPH) based on Gaussian–Markov random fields (GMRFs) have been successfully used in effective texture discrimination. LPH features represent the normalized histograms of locally estimated GMRF parameters via local linear regression. However, these features are not rotation invariant. In this paper two techniques to design rotation invariant LPH texture descriptors are discussed namely, Rotation Invariant LPH (RI-LPH) and the Isotropic LPH (I-LPH) descriptors. Extensive texture classification experiments using traditional GMRF features, LPH features, RI-LPH and I-LPH features are performed. Furthermore comparisons to the current state-of-the-art texture features are made. Classification results demonstrate that LPH, RI-LPH and I-LPH features achieve significantly better accuracies compared to the traditional GMRF features. RI-LPH descriptors give the highest classification rates and offer the best texture discriminative competency. RI-LPH and I-LPH features maintain higher accuracies in rotation invariant texture classification providing successful rotational invariance. [ABSTRACT FROM AUTHOR]

Projections being based on global image representative are considered as simplified and efficient stabilization solutions under poor textural or imaging conditions. The existing projection algorithms are investigated only for translation or rotation stabilization, but the image captured with unsteady camera also suffers from undesired zoom-in/out effect. This paper presents a new zoom estimation algorithm for extended projection applicability. Camera focusing generally being effective around its optical axis, produces zooming around the image center. This image-centered zoom results in an opposite relative motion (stretching/compression) in two halves of the image projections. A novel concept of utilizing this relative longitudinal projection motion is proposed for camera zoom estimation. Derivative dynamic time warping is used for image projection alignment and the effect of mismatched-warping singularities leading to motion estimation error is removed by selecting some key warping vectors labeled as Interest Warping Vectors . Slope values of these interest warping vectors in the two projection-halves are utilized for zoom factor estimation and the approach is also generalized for arbitrary zoom-center under forward translation. Robustness of the proposed algorithm tested with artificial textural artifacts is verified with motion accuracy analysis on various sequentially zoomed test and real-world images within 0.8 to 1.2 scale factor range. [ABSTRACT FROM AUTHOR]

The presence of reflection on a surface has been a long-standing problem for object recognition since it brings negative effects on object’s color, texture and structural information. Because of that, it is not a trivial task to recognize the surface structure affected by the reflection, especially when the object is entirely reflective. Most of the cases, reflection is considered as noise. In this paper, we propose a novel method for entire reflective object sub-segmentation by transforming the reflection motion into object surface label. To the best of our knowledge, the segmentation of entirely reflective surfaces has not been studied. The experimental results on specular and transparent objects show that the surface structures of the reflective objects can be revealed and the segmentation based on the surface structure outperforms the approaches in literature. [ABSTRACT FROM AUTHOR]

This paper proposes a novel methodology of characterizing and recognizing plant leaves using a combination of texture and shape features. Texture of the leaf is modeled using Gabor filter and gray level co-occurrence matrix (GLCM) while shape of the leaf is captured using a set of curvelet transform coefficients together with invariant moments. Since these features are in general sensitive to the orientation and scaling of the leaf image, a pre-processing stage prior to feature extraction is applied to make corrections for varying translation, rotation and scaling factors. Efficacy of the proposed methods is studied by using two neural classifiers: a neuro-fuzzy controller (NFC) and a feed-forward back-propagation multi-layered perceptron (MLP) to discriminate between 31 classes of leaves. The features have been applied individually as well as in combination to investigate how recognition accuracies can be improved. Experimental results demonstrate that the proposed approach is effective in recognizing leaves with varying texture, shape, size and orientations to an acceptable degree. [ABSTRACT FROM AUTHOR]

Predicting the quality of user-generated answers is definitely of great importance for community-based question answering (CQA) due to the frequent occurrence of low-quality answers. Most existing answer quality prediction works combine non-textual features of user-generated answers directly without considering the diversity of non-textual features. In this paper, we propose two co-training approaches: random subspace split-based co-training (RSS-CoT) and content and social split-based co-training (CS-CoT) to predict the quality of answers by mining the relationships of non-textual features and unlabeled data in CQA. Our results demonstrate that both appropriate combination of non-textual features and unlabeled data can promote the prediction performance of answer quality. [ABSTRACT FROM AUTHOR]

This paper presents a fast precise unsupervised iris defects detection method based on the underlying multispectral spatial probabilistic iris textural model and adaptive thresholding applied to demanding high resolution mobile device measurements. The accurate detection of iris eyelids and reflections is the prerequisite for the accurate iris recognition, both in near-infrared or visible spectrum measurements. The model adaptively learns its parameters on the iris texture part and subsequently checks for iris reflections using the recursive prediction analysis. The method is developed for color eye images from unconstrained mobile devices but it was also successfully tested on the UBIRIS v2 eye database. Our method ranked first from the 97+1 recent Noisy Iris Challenge Evaluation contest alternative methods on this large color iris database using the exact contest data and methodology. [ABSTRACT FROM AUTHOR]

This paper proposes an effective scale invariant texture representation based on frequency decomposition and gradient orientation. First, the image intensities are decomposed into different orientations by using wedge filters in the frequency domain, and the N-nary coding method is adopted for the vector quantization. Second, the scale invariant gradient orientation is generated by selecting the most stable value of the gradient orientation with different Gaussian scales. Finally, the 2D joint distribution of the two types of local descriptors is used as the representation. The performance was evaluated on texture classification using a nearest neighbor classifier. Simple but not ordinary, our method achieves state of the art classification performance on the KTH-TIPS dataset under the traditional experimental design. Moreover, the main experiments were conducted on the KTH-TIPS and KTH-TIPS2-b datasets with the experimental designs of scale invariance validation. Compared with the methods of basic image features (BIFs) and local energy pattern (LEP), the proposed representation achieves superior performance with much lower dimension of representation. [ABSTRACT FROM AUTHOR]

Describing textures is a challenging problem in computer vision and pattern recognition. The classification problem involves assigning a category label to the texture class it belongs to. Several factors such as variations in scale, illumination and viewpoint make the problem of texture description extremely challenging. A variety of histogram based texture representations exists in literature. However, combining multiple texture descriptors and assessing their complementarity is still an open research problem. In this paper, we first show that combining multiple local texture descriptors significantly improves the recognition performance compared to using a single best method alone. This gain in performance is achieved at the cost of high-dimensional final image representation. To counter this problem, we propose to use an information-theoretic compression technique to obtain a compact texture description without any significant loss in accuracy. In addition, we perform a comprehensive evaluation of pure color descriptors, popular in object recognition, for the problem of texture classification. Experiments are performed on four challenging texture datasets namely, KTH-TIPS-2a, KTH-TIPS-2b, FMD and Texture-10. The experiments clearly demonstrate that our proposed compact multi-texture approach outperforms the single best texture method alone. In all cases, discriminative color names outperforms other color features for texture classification. Finally, we show that combining discriminative color names with compact texture representation outperforms state-of-the-art methods by 7.8%, 4.3% and 5.0% on KTH-TIPS-2a, KTH-TIPS-2b and Texture-10 datasets respectively. [ABSTRACT FROM AUTHOR]

This paper proposes a DEMD-based textured video coding scheme for efficient representation of textured data compatible with an H.264/MPEG framework. We encode the DEMD residue image of all non- I frames ( P and B frames in H.264/MPEG) instead of the original frame. All non- I frames are synthesized using correlation search and patch-based multi-level IMF synthesis from the DEMD representation of key frames ( I frames, in the H.264/MPEG framework) and the DEMD residual image of the frame being synthesized. Experimental results show that the proposed scheme provides better compression results as compared to MPEG/H.264 with acceptable reconstruction quality. [ABSTRACT FROM AUTHOR]

Deriving an effective image representation is a critical step for a successful automatic image recognition application. In this paper, we propose a new feature descriptor named Local Directional Texture Pattern (LDTP) that is versatile, as it allows us to distinguish person’s expressions, and different landscapes scenes. In detail, we compute the LDTP feature, at each pixel, by extracting the principal directions of the local neighborhood, and coding the intensity differences on these directions. Consequently, we represent each image as a distribution of LDTP codes. The mixture of structural and contrast information makes our descriptor robust against illumination changes and noise. We also use Principal Component Analysis to reduce the dimension of the multilevel feature set, and test the results on this new descriptor as well. [ABSTRACT FROM AUTHOR]

Abstract: Texture analysis is a very important area in the field of computer vision and related fields. There are a good number of databases developed by different research groups for various texture analysis, in the field of medical analysis, robotics, recognition, analysis, image processing, etc. However, till-to-date, there is no comprehensive works covering the important databases and analyze these in various perspectives. In this paper, we consider this important task so that it becomes helpful for a researcher to choose and evaluate having crucial evaluating aspects in mind. We categorize and critically survey based on many references of the state-of-the-art related to the databases and other texture works. We strongly believe that this elegant survey will be a great contribution for the vision community, especially in the arena of texture analysis. [Copyright &y& Elsevier]

Abstract: There is no single best representation of images that can separate different classes with well defined boundaries in the feature space. Therefore, content-based image retrieval (CBIR) using conventional distance metric is not efficient in the low level image feature space viz. texture. Classifier based retrieval approaches (classification followed by retrieval) classify the query image and retrieve images only from the identified class. The performance of such approaches greatly relies on the performance of classifier. For each correct classification of query image, these systems yield high retrieval accuracy and for each misclassification the systems result in complete failure. It results huge variance in performance. This paper proposes a novel approach to content-based image retrieval called “Class Membership-based Retrieval” that addresses the limitations of both conventional distance based and conventional classifier based retrieval approaches. The proposed method consists of two steps. First, the class label and fuzzy class membership of query image is computed using neural network. In the second step, the retrieval is performed using a combination of simple and weighted (class membership based) distance metric in complete search space unlike the conventional classifier based retrieval techniques. The proposed technique also provides flexibility in reducing the search space in steps increasing the speed of retrieval at the cost of gradual reduction in accuracy. The performance of the method is evaluated using three texture data sets varying in orientations, complexity and number of classes. Experimental results support the proposed technique favorably when compared with other promising texture retrieval schemes. [Copyright &y& Elsevier]

Abstract: Two novel methods for extracting distinctive invariant features from interest regions are presented in this paper. The idea of these methods are associated with that measuring similarity between visual entities from images can be based on matching the internal layout of Local Self-Similarities. The main contributions are two-folds: firstly, two new texture features called Local Self-Similarities (LSS,C) and Fast Local Self-Similarities (FLSS,C) based on Cartesian location grid, are extracted, which are the modified versions of the well-known Local Self-Similarities (LSS,LP) feature based on Log-Polar location grid. To combine the powers of the SIFT and LSS (LP), LSS and FLSS are used as the local features in the SIFT algorithm. Secondly, different from the natural LSS (LP) descriptor that chooses the maximal correlation value in each bucket to get photometric translations invariance, the proposed LSS (C) and FLSS (C) adopt distribution-based representation to achieve more robust geometric translations invariance. In the contexts of image matching and object category classification experiments, the LSS (C) and FLSS (C) both outperform the original LSS (LP), and achieve favorably comparable performance to the SIFT. Furthermore, these descriptors are low computational complexity and simpler than the SIFT. [Copyright &y& Elsevier]

Abstract: The local binary pattern (LBP) operator is a very effective multi-resolution texture descriptor that can be applied in many image processing applications. However, existing LBP operators can not use the information of non-uniform patterns efficiently and they are also sensitive to noise. This paper proposes a noise tolerant extension of LBP operators to extract statistical and structural image features for efficient texture analysis. The proposed LBP operator uses a circular majority voting filter and suitable rotation-invariant labeling scheme to obtain more regular uniform and non-uniform patterns that have better discrimination ability and more robustness against noise. Experimental results on the Brodatz, CUReT and MeasTex databases show the improvement of the proposed LBP operator performance, especially when a large number of neighbors are used for extracting texture patterns. [Copyright &y& Elsevier]

Abstract: In this paper, we introduce a novel approach to grade prostate malignancy using digitized histopathological specimens of the prostate tissue. Most of the approaches proposed in the literature to address this problem utilize various textural features computed from the prostate tissue image. Our approach differs in that we only focus on the tissue structure and the well-known Gleason grading system specification. The color space representing the tissue image is investigated and basic components of the prostate tissue are detected. The components and their structural relationship constitute a complete gland region. Tissue structural features extracted from gland morphology are used to classify a tissue pattern into three major categories: benign, grade 3 carcinoma and grade 4 carcinoma. Our experiments show that the proposed method outperforms a texture-based method in the three-class classification problem and most of the two-class classification problems except for the grade 3 vs grade 4 classification. Based on these results, we propose a hierarchical (binary) classification scheme which utilizes the two methods and obtains 85.6% accuracy in classifying an input tissue pattern into one of the three classes. [Copyright &y& Elsevier]

Abstract: In this paper, we propose a method of automatic detection of texture-periodicity using superposition of distance matching functions (DMFs) followed by computation of their forward differences. The method has been specifically devised for automatically identifying row and column periodicities and thereby the size of periodic units from textile fabrics belonging to any of the 17 wallpaper groups and is a part of automatic fabric defect detection scheme being developed by us that needs periodicities along row and column directions. Overall row-DMF (or overall column-DMF) is obtained based on superposition of DMF of all rows (or columns) from the input image and its second forward difference is computed to get the overall maximum which is a direct measure of periodicity along row (or column) direction. Results from experiments on various near-regular textures demonstrate the capability of the proposed method for automatic periodicity extraction without the need of human intervention. [Copyright &y& Elsevier]