Medical Image Retrieval based on LBP Histogram Fourier features and KNN classifier

Abstract

In the recent past, Content Based Image Retrieval Systems has become much important in various applications such as medical images. Such systems must effectively retrieve medical images of the same kind. In this paper, we propose to extract local binary pattern (LBP) Histogram Fourier features from each image and the query image, perform the histogram intersection and then apply KNN classifier over the output images of the histogram intersection such that only nearby distance images with respect to the query image is retrieved, thus improving the performance of the system in terms of precision and recall. The proposed technique is compared with that of the histogram intersection based classification and KNN classification separately. Experimental results demonstrate better accuracy of the proposed technique when compared to the existing techniques.