Extraction of Shape Components for Classification of Textures Based on Texture Elements

摘要

Shape representation or description is a central and challenging problem in Image Processing and Computer Vision which arises in many applications such as image classification, analysis, interpretation etc. Texture is an important spatial feature useful for identifying objects or regions of interest in an image. The present paper derives a new set of texture features, which are shape components derived from the texture elements of a 3x3 mask. The proposed texture elements extract textural information of an image with a more complete respect of texture characteristics in all the eight directions instead of only one displacement vector. The present paper evaluated five simple shape components on each of the derived patterns. The experimental results on the five groups of texture images clearly show the efficacy and simplicity of the present method.