Performance analysis of texture classification techniques using shearlet transform

Orientation, scale, sharp image transitions or singularities such as edges, and the other visual appearance are the major problem in texture classification. Texture classification is one of the most importantissue in image processing and computer vision. The applications of texture classification depends on classifying patterns or identifying objects in the field of image editing and completion, industrial and biomedical inspection, segmentation of aerial imagery, remote sensing, document analysis, identifying ground objects and content based access to image database. Shearlets are the extension of wavelets that are widely use for image characterization. In this project, the texture classification and retrieval methods that are used to obtain higher classification rate are modeled using the adjacent shearlet subband dependences.For classification of image textures the energy features such as mean and standard eviation are used, that represent each shearlet subband. The textures are classified using minimum distance classifier and linear regression moeling. Shearlet transform has been widely recognized as an advanced tool in texture analysis, due to its efficient multiscale directional representation. The validation of experiments performed using different classifiers proves that the proposed method outperforms the current methods and it is proved that shearlets are highly sensitive to directions.