Texture Study

Motivation

Texture almost presents everywhere in natural and real world images. Texture, therefore, has long been an important research topic in image processing. Successful applications of texture analysis methods have been widely found in industrial, medical and remote sensing areas. For example, texture analysis techniques had been used to evaluate roentgenograms in order to classify normal and abnormal interstitial pulmonary patterns . In addition, texture analysis techniques had also been applied in the remote sensing area for the identification of crop types by using radar imagery . Furthermore, the recent emerging of multimedia and the availability of large image and video archives has made content-based information retrieval become a very popular research topic. Texture is also deemed as one of the most important features when performing content-based information retrieval. As texture is an essential feature when performing image retrieval, the study of texture analysis becomes critical and important. In the past, lots of researches have been devoted to solve the texture analysis problems, such as texture segmentation, texture classification and texture primitive detection etc.

State of the Problems

Generally speaking, texture analysis methods can be classified into three major categories, namely, statistical, structural and spectral. In statistical approaches , texture statistics such as the moments of the gray-level histogram, or statistics based on gray-level co-occurrence matrix are computed to discriminate different textures. For structural approaches , “texture primitive”, the basic element of texture, is used to form more complex texture pattern by grammar rules which specify the generation of texture pattern. Finally, in spectral approaches , the textured image is transformed into frequency domain by a Fourier transform. Then, the extraction of texture features can be done by analyzing the power spectrum. In addition to the above-mentioned methods, there are also other texture analysis approaches proposed .

Recently, multichannel and multiresloution-based approaches have drawn lots of attention in the field of texture analysis . Wavelet transform is one of the multiresolution-based approaches. Most of the wavelet-based methods use a pyramidal type of decomposition to transform the input image into an image of wavelet coefficient at different resolutions. The wavelet coefficients are then transformed into texture-specific features.

In our study, we havel proposed methods to deal with the problems of texture segmentation, coarse classification of textures and texture primitive extraction. In addition, we will provide a method to compute the texture browsing descriptor of MPEG-7. These four problems are defined as follows:

(1) Texture segmentation: given a texture image consisting of several regions of different textures, to develop a method to automatically segment these texture regions out.

(2) Coarse classification of textures: given a texture image, to develop a method to classify it into one of the three classes: directional, periodic and random.

(3) Texture primitive extraction: given a regular texture, to develop a method to extract its primitives used to reproduce the texture image by a certain displacement rule.

(4) Texture browsing descriptor computation: based on the syntax and semantics of MPEG-7 standard, compute the texture browsing descriptor.

As mentioned above, lots of research efforts have been spent on these problems. However, some points still remain to be solved:

(1) Texture segmentation: traditional approaches for texture segmentation via wavelet transform usually adopt textural features to achieve segmentation purposes. However, for a natural image, the characteristics of the pixels in a texture region are not similar everywhere from a global viewpoint, over-segmentation often occurs.

(2) Coarse classification of textures: it will be very useful to provide a preliminary texture classification method based on the three most important dimensions of human texture perception, i.e., periodicity, directionality and randomness. In addition, coarse classification of textures is also one of the applications of the texture browsing descriptor of MPEG-7.

(3) Texture primitive extraction: t o lessen the computational load, most of the methods proposed usually adopt coarse quantization when extracting texture primitives, therefore the accuracy of texture primitive extraction is sacrificed.

 

Department of Computer and Information Science, National Chiao Tung University, 1001 Ta Hsueh Rd., Hsinchu 30050, Taiwan, R.O.C.
國立交通大學工程三館EC120 TEL: 03-5712121 轉 54744
Contact Us | ©2006AIP Lab@nctu