Texture segmentation of ultrasound B-scan image by sum and difference histograms

摘要

A technique is described for segmenting B-scan images to extract a tissue signature by texture analysis. Gray-level sum and difference histograms are used to extract texture features. A K-mean clustering algorithm is used to partition the B-scan image into homogeneous zones of texture. The performance of the method is tested by the analysis of natural texture and abdominal images.