Texture Analysis of Non-Uniform Images using GLCM G.Arockia Selva Saroja 1 Dr.C.Helen Sulochana 2 1 Associate Professor / ECE./ Noorul Islam Centre for Higher Education, Kanyakumari District,India 2 .Professor / ECE./St.Xavier!s Catholic College of Engineering,Kanyakumari District,India Abstract:Texture analysis finds central role in automatic inspection, medical image analysis, document processing and remote sensing. The result of deviation in illuminant direction affects the texture appearance. The images present in the universe are not uniform because of changes in scale, orientation and lighting conditions. The feature extraction of the non uniform images was done using gray level co-occurrence matrix(GLCM) for the different datasets for the non ideal images.The results showed that performance result varied with the texture datasets used. Keywords: Texture analysis, feature extraction, gray level cooccurrence matrix . I INTRODUCTION Textures are characteristic intensity variations that typically originate from roughness of object surfaces[1].Generally it can be defined as a regular repetition of elements or pattern on a surface. These texture images vary in brightness, color, shape, size, etc[2].Texture analysis plays an important role in texture classification. Texture analysis can be defined as a set of mathematical procedures used to extract feature information from the input texture image. They provide information about the local spatial organization of spatially varying spectral values. Recently, several important advancements are made in the field of texture analysis [3]- [7].Analyzing a texture is a complicated process because of the variations in the periodicity, directionality and randomness of the image[8]. Small changes in the surrounding light conditions also affects the spectral quality of the texture image. So, the texture analysis methods should be illumination invariant,. Most of the methods fail when the images are under real conditions. Number of solutions have been put forward for those problems. But except few techniques, most of the methods failed to perform well under non-uniform conditions. In our work, statistical approach is considered for texture analysis[9].They provide information about the spatial distribution of pixels in an image[10].Statistical approach yield details about the characteristics of texture, whether it is smooth, coarse, fine, etc. Previously, spatial gray level dependence method(SGLDM)[11], gray level run length method(GLRLM)[11],local pattern spectrum[12],wavelet decomposition with energy signatures[13][14] were used under uniform conditions. When these methods are tested under non-uniform conditions, only few techniques performed well. In this paper, the performance of statistical feature extraction method is analyzed using K-NN classifier. II GRAY LEVEL CO-OCCURRENCE MATRIX Gray level co-occurrence matrix(GLCM) was suggested by Haralick[15].It is one of the widely used texture analysis algorithm because it can be implemented easily. GLCM contains information about the positions of pixels having similar gray level values[16]. GLCM extracts the structural information about the texture pattern to be analyzed at different scale and orientation . This made the GLCM more effective,but at the cost of significantly increased computationsThis method extract the information by considering only a pixel pairs. GLCM provides a tabulation about how often different combinations of pixel intensity value occur in an image. A co- occurrence matrix is given as Proceedings of 2013 IEEE Conference on Information and Communication Technologies (ICT 2013) 978-1-4673-5758-6/13/$31.00 ' 2013 IEEE 1319