In this paper, a novel method for classification of Ultrasound Kidney images using texture properties based on Logical operators is presented. Different regions of an image are identified based on texture properties. This algorithm mainly deals with operators and that are constructed from logical building blocks. The different classes of images are convolved by these operators and the resulting responses are converted to standard deviation matrices computed over a sliding window. Features for classification of images are extracted from these standard deviation matrices using zonal masks. Feature selection process is applied to these zonal sampling features and new set of features are used for classification. This work proposes an algorithm for classification of textures of three different categories namely Fine, Coarse and Rough. Based on this texture classification, this algorithm applied to medical images. Three kinds of Ultrasound kidney images namely Normal (NR), Medical Renal Diseases (MRD) and Cortical Cysts (CC) images are classified based on texture properties. This algorithm involves only convolution and simple arithmetic in various stages which leads faster implementation. The efficient feature space is created for textures as well as US kidney image classification. For classification, zonal mask sum features gives efficient classification for texture images. For kidney images difference features gives better results. This algorithm has higher classification accuracy and computational superiority.
展开▼