Improved extension of local binary pattern in wavelet domain for texture classification

Abstract

In recent past the volume of storing digital images in database has been galloping leaps and bounds. Content based image retrieval is a wellknown method to retrieve digital images using color, texture, shape, etc. Texture classification is widely focused by many researchers and adopted in many applications such as face recognition, tea leaves identification, palm newlineprint identification, ear recognition, etc. The extraction of texture feature is a prominent phase in texture classification. The texture classification is an extensive field of study in the present scenario. The texture features are extracted in spatial or frequency domain. The methods existing to extract the texture features in spatial domain are Gray Level Co occurrence Matrix, local binary pattern, etc. Among the available spatial domain methods; local binary pattern is a robust method in extracting the texture image features. The texture features in frequency domain are extracted through various methods such as Gabor filters, discrete wavelet transform, etc. In frequency domain discrete wavelet transform is widely used. The Local Binary Patterns (LBP) based techniques offer the required texture information in the spatial domain. In the transform domain, the wavelet based techniques describe the texture of an image with enhanced features. Researchers have applied LBP and discrete wavelet transform as it derives specific benefits. newline newline newline newline