In view of the existing texture feature extraction method of computational complexity and accuracy problems, this paper proposes a calculation method fused with Complete Local Binary Patterns (CLBP) and Gray-level Co-occurrence Matrix (GLCM). This method uses the rotation invariant CLBP operator to process the texture image and get the CLBP image, then calculate the GLCM of the CLBP image, use the contrast, correlation, energy and inverse difference moment to describe the image texture feature. The experimental results show that the method can reduce the feature parameters at the same time, also improved the texture description ability. Introduction Texture is an important visual cue. It is feature which is widespread in the image and difficult to describe. Texture feature extraction is one of the hot topics in computer vision, image processing, image analysis and image retrieval. T. Ojala in 1996 proposed the Local Binary Patterns algorithm (Local Binary Patterns, LBP) [1], for the description of the texture feature. LBP algorithm is simple and easy to understand, the computational complexity is small, and it can well describe the local texture features of the image. So it attracts the attention of many research scholars. In the past many years, the researchers studied the LBP algorithm, and proposed several improved algorithms such as FLBP, LBP, FPLBP, MS-LBP, CLBP [2] and so on, and it is widely applied in image segmentation, face recognition, image retrieval and other fields. Compared with other improved LBP algorithms, CLBP algorithm is more comprehensive and precise in local texture description and texture feature extraction, and has achieved good results. But the shortcoming of CLBP algorithm is that the feature dimension is large in the process of texture feature description, which brings great difficulty to the calculation, and reducing the feature dimension will inevitably lead to the loss of texture features. Haralick feature is proposed by Haralick for analyzing the 14 features of Gray-level Co-occurrence Matrix (GLCM). It is found that there are only 4 features (contrast, correlation, energy and inverse difference moment) in the 14 texture features, which are not only convenient for calculation, but also can provide a high classification accuracy. This paper proposes a calculation method fused with CLBP and GLCM, using the rotation invariant CLBP operator to process the texture image and get the CLBP image. At last calculate the GLCM of the CLBP image, and use the contrast, correlation, energy and inverse difference moment to describe the image texture feature. Local Binary Patterns Traditional Local Binary Patterns (LBP). The basic idea of LBP is to calculate the LBP operator by comparing the image pixels and the pixels around it. Take this pixel as the center and compare the adjacent pixels. If the gray value of the central pixel is greater than or equal to its adjacent pixels, it marked as 1, otherwise marked as 0. As shown in Fig. 1: This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). Copyright © 2017, the Authors. Published by Atlantis Press. 167 Advances in Computer Science Research (ACSR), volume 76 7th International Conference on Education, Management, Information and Mechanical Engineering (EMIM 2017)
Paper
Full text
Texture Feature Extraction Research Based on GLCM-CLBP Algorithm
Semantic Scholar · Computer Science · 2017
Abstract
In view of the existing texture feature extraction method of computational complexity and accuracy problems, this paper proposes a calculation method fused with Complete Local Binary Patterns (CLBP) and Gray-level Co-occurrence Matrix (GLCM). This method uses the rotation invariant CLBP operator to process the texture image and get the CLBP image, then calculate the GLCM of the CLBP image, use the contrast, correlation, energy and inverse difference moment to describe the image texture feature. The experimental results show that the method can reduce the feature parameters at the same time, also improved the texture description ability. Introduction Texture is an important visual cue. It is feature which is widespread in the image and difficult to describe. Texture feature extraction is one of the hot topics in computer vision, image processing, image analysis and image retrieval. T. Ojala in 1996 proposed the Local Binary Patterns algorithm (Local Binary Patterns, LBP) [1], for the description of the texture feature. LBP algorithm is simple and easy to understand, the computational complexity is small, and it can well describe the local texture features of the image. So it attracts the attention of many research scholars. In the past many years, the researchers studied the LBP algorithm, and proposed several improved algorithms such as FLBP, LBP, FPLBP, MS-LBP, CLBP [2] and so on, and it is widely applied in image segmentation, face recognition, image retrieval and other fields. Compared with other improved LBP algorithms, CLBP algorithm is more comprehensive and precise in local texture description and texture feature extraction, and has achieved good results. But the shortcoming of CLBP algorithm is that the feature dimension is large in the process of texture feature description, which brings great difficulty to the calculation, and reducing the feature dimension will inevitably lead to the loss of texture features. Haralick feature is proposed by Haralick for analyzing the 14 features of Gray-level Co-occurrence Matrix (GLCM). It is found that there are only 4 features (contrast, correlation, energy and inverse difference moment) in the 14 texture features, which are not only convenient for calculation, but also can provide a high classification accuracy. This paper proposes a calculation method fused with CLBP and GLCM, using the rotation invariant CLBP operator to process the texture image and get the CLBP image. At last calculate the GLCM of the CLBP image, and use the contrast, correlation, energy and inverse difference moment to describe the image texture feature. Local Binary Patterns Traditional Local Binary Patterns (LBP). The basic idea of LBP is to calculate the LBP operator by comparing the image pixels and the pixels around it. Take this pixel as the center and compare the adjacent pixels. If the gray value of the central pixel is greater than or equal to its adjacent pixels, it marked as 1, otherwise marked as 0. As shown in Fig. 1: This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). Copyright © 2017, the Authors. Published by Atlantis Press. 167 Advances in Computer Science Research (ACSR), volume 76 7th International Conference on Education, Management, Information and Mechanical Engineering (EMIM 2017)