Rotation invariant texture descriptors based on Gaussian Markov random fields for classification

Local multiple patterns based multiresolution gray-scale and rotation invariant texture classification

2010 17th IEEE International Conference on Image Processing, ICIP 2010, Hong Kong, 26-29 September 2010

In this paper, we propose a novel feature extraction scheme for texture classification, in which the texture features are extracted by a two-level hybrid scheme, by integrating two statistical techniques of texture analysis. In the first step, the low level features are extracted by the Gabor filters, and they are encoded with the feature map indices, using Kohonen's SOFM algorithm. In the next step, the encoded feature images are processed by the Gabor filters, Gaussian Markov random fields (GMRF), and Grey level co- occurrence matrix (GLCM) methods to extract the high level features. By integrating two methods of texture analysis in a cascaded manner, we obtained the texture features which achieved a high accuracy for the classification of texture patterns. The proposed schemes were tested on the real microtextures, and the Gabor-GMRF scheme achieved 10 percent increase of the recognition rate, compared to the result obtained by the simple Gabor filtering.

Extracting efiective features to describe texture is always a key problem in writer identiflcation. This paper proposes a novel method for texture feature extraction by integrating Gabor fllter and Gauss Markov Random Field (GMRF). That is to say, the handwriting images are flrstly flltered by a bank of Gabor fllters in which global features such as directional information can be detected; then GMRF models are developed for every flltered image that is ∞exible enough to capture the local spatial structure and the model parameters of all GMRFs are concatenated as texture features for writer identiflcation, and flnally, Support Vector Machine (SVM) is applied to evaluate the performance of calligraphic artist identiflcation. The experimental results show that this method can achieve a classiflcation rate of 98%, which has outperformed the traditional Gabor fllter method.

Texture feature of the froth image is widely used in the working condition recognition of froth flotation. However, due to the complexity of the froth image, the current texture features vary greatly and are difficult to identify the work condition accurately. Therefore, we propose a dynamic texture feature named LBP on the TOP and GLCM Histograms (LTGH) which integrates the local binary patterns (LBPs) and gray-level co-occurrence matrix (GLCM) histograms on the three orthogonal planes (TOP). First, we use the rotation invariant LBPs to enhance rotation invariance and illumination robustness. Then, we implement the TOP on the enhanced texture feature map to generate the multiple dimensional enhanced feature maps. After that, we calculate the GLCM and supplementary features (SFs) on the multiple dimensional enhanced feature map. Finally, we integrate the histogram of the GLCM and SFs to discriminate the texture feature. The LTGH feature considers the froth structures both in the macrolevel and microlevel and captures the temporal information between the froth images. Experiments have demonstrated the effectiveness and stability of the proposed texture feature for work condition recognition in froth flotation. Compared with other traditional texture features, the accuracy of the LTGH feature has been increased by at least 7.76%.

The analysis of image feature robustness using cometcloud

Is local dominant orientation necessary for the classification of rotation invariant texture?

The purpose of this study is to develop a dissimilarity measure for the classification of trabecular bone (TB) texture in knee radiographs. Problems associated with the traditional extraction and selection of texture features and with the invariance to imaging conditions such as image size, anisotropy, noise, blur, exposure, magnification, and projection angle were addressed. In the method developed, called a signature dissimilarity measure (SDM), a sum of earth mover's distances calculated for roughness and orientation signatures is used to quantify dissimilarities between textures. Scale-space theory was used to ensure scale and rotation invariance. The effects of image size, anisotropy, noise, and blur on the SDM developed were studied using computer generated fractal texture images. The invariance of the measure to image exposure, magnification, and projection angle was studied using x-ray images of human tibia head. For the studies, Mann-Whitney tests with significance level of 0.01 were used. A comparison study between the performances of a SDM based classification system and other two systems in the classification of Brodatz textures and the detection of knee osteoarthritis (OA) were conducted. The other systems are based on weighted neighbor distance using compound hierarchy of algorithms representing morphology (WND-CHARM) and local binary patterns (LBP). Results obtained indicate that the SDM developed is invariant to image exposure (2.5-30 mA s), magnification (x1.00 - x1.35), noise associated with film graininess and quantum mottle (< 25%), blur generated by a sharp film screen, and image size (> 64 x 64 pixels). However, the measure is sensitive to changes in projection angle (> 5 degrees), image anisotropy (> 30 degrees), and blur generated by a regular film screen. For the classification of Brodatz textures, the SDM based system produced comparable results to the LBP system. For the detection of knee OA, the SDM based system achieved 78.8% classification accuracy and outperformed the WND-CHARM system (64.2%). The SDM is well suited for the classification of TB texture images in knee OA detection and may be useful for the texture classification of medical images in general.

Texture classification algorithms using local binary pattern (LBP) and its variants usually can achieve attractive results. However, the selected rotation invariant structural patterns in numerous LBP variants are not absolutely continuous invariant to any rotation angle. To improve the classification effectiveness on this occasion, in this paper, we introduce a robust descriptor based on the principal curvatures (PCs) and rotation invariant version of CLBP_Sign operator in completed LBP (CLBP), namely PC-LBP. Different from the original LBP and many LBP variants, PCs are employed in this paper to represent each local structure information due to their continuous rotation invariance. Simultaneously, both micro-and macro-structure texture information can also be captured through PCs, which comprise maximum and minimum curvatures. Inspired by the similar coding strategy of the CLBP_Sign operator, a new operator CLBP_PC is developed. By exploiting complementary information resulting from the two operators combination, the final PC-LBP descriptor has the properties of conspicuous rotation invariance, strong discriminativeness, gray scale invariance, needless of pretraining, and high computational efficiency. In addition, to improve the robustness of texture classification with multiresolution, a multiscale sampling approach is designed by adjusting three parameters accordingly. Experimental results demonstrate that the proposed multiresolution PC-LBP approach achieves comparable performance or outperforms a large number of state-of-the-art methods. Impressively, the classification accuracy of the proposed method performed on Outex_TC_00010 test suite is 100%.

Double-smoothing for bias reduction in local linear regression

Local Binary Pattern (LBP) has achieved great success in texture classification due to its accuracy and efficiency. Traditional LBP method encodes local features by binarying the difference in local neighborhood and then represents a given image using the histogram of the binary patterns. However, it ignores the directional statistical information. In this paper, some directional statistical features—including the mean and standard deviation of the local absolute difference—are integrated into the feature extraction to improve the classification ability of the extracted features. In order to reduce estimation errors of the local absolute difference, we further utilize the least square estimate technique to optimize the weight and minimize the local absolute difference, which leads to more stable directional features. In addition, a novel rotation invariant texture classification approach is presented. Experimental results on several texture and face datasets show that the proposed approach significantly improves the classification accuracy of the traditional LBP.

This paper is concerned with a systematic exposition of the usefulness of two-dimensional (2-D) discrete Gaussian Markov random field (GMRF) models for image processing applications. Specifically, we discuss the following topics; notion of Markovianity on a plane, statistical inference in GMRF models, and their applications in several image related problems such as, image synthesis, texture classification, segmentation and image restoration.

Texture is a natural characteristic of all surfaces, which describes the visual patterns, and each has homogenization properties. In this paper, our concern is with the Regional Agricultural Land texture classification using grey level co-occurrence matrices (GLCMs). Firstly, the Gabor filter is applied to the test image in order to allow a certain band of frequencies and reject the others. Then texture discrimination is performed to partition a textured image into areas, each related to a homogeneous texture where samples of four different textures are extracted from the image. For each patch, the four features of the GLCM matrices namely, dissimilarity, correlation, angular second moment, and homogeneity are computed. Finally, for texture classification we have used two well-known methods: Support Vector Machine and decision tree induction. The results show that these texture features have high discrimination accuracy and classification using support vector machines gives better results as compared to the decision tree induction classifier.

Bare hand detection system stages many challenging factors such as illumination, position and rotation variation, complex background etc. Present study explores 4 gray-level texture features and 5 color-texture features for robust detection of hand. In this paper, hand detection primarily aims to develop a 2D based hand gesture recognition system. Existing system uses red-markers, data-glove, and 3D cameras etc., which are not cost effective and user friendly for a realtime application. Present system uses a classification based approach using Naive Bayes, Real AdaBoost, Gentle AdaBoost and Modest AdaBoost for analyzing the notable information present in the features. AdaBoost is an efficient boosting algorithm primarily developed for binary classification. A bare hand detection system is effected by even a slight variation in the environmental conditions. Present study highlights the performance of the features under varying non-ideal conditions. An effective and flexible pair-wise comparative analysis method called Analytic Hierarchy Process (AHP) is used for pair-wise comparisons among classifiers and features. Results suggest that Modest AdaBoost is the most stable classifier among others, while Laplacian of Gaussian (LoG) based Gabor feature is the most efficient feature with the maximum accuracy of 97%. Features and classifiers are also evaluated based on time-complexity where Naive Bayes system has better training time-efficiency as compared to AdaBoost systems and color-texture features has less time complexity as compared to texture features due to lower dimensions.

Principal curvatures based rotation invariant algorithms for efficient texture classification