论文部分内容阅读
What image feature extract is the basis of image recognition,image data mining,image retrieval,is the most crucial step of pattern recognition and classification.In this paper,we is performed on wavelet-based and gray level co-occurrence matrix(GLCM)to extract texture feature of liver's region of interest(ROI).ROI corresponding to normal liver,mono-hydatid cyst and multiple daughter hydatid cyst.For each ROI,first a wavelet-based texture feature set is derived from two level discrete wavelet transformed approximation(low frequency part of the image)sub image;Then by using GLCM extracted feature of the sub image;Meanwhile constructing a decision tree C4.5 algorithm characteristics of classifier to verify the ability of classification.The result of decision tree C4.5 for extract feature is processed classify.normal liver classification accuracy rate reached to 89%;mono-hydatid cyst classification accuracy rate reached to 93%;multiple daughter hydatid cyst classification accuracy rate reached to 97%.The experimental results show that wavelet co-occurrence matrix and decision tree C4.5 could process classification for the different types of liver images,to some certain extent for reference on the radiologists to provide diagnosis and decide,for the subsequent retrieval efficiency of content-based image retrieval system laid a foundation.