Indian Journal of Science and Technology

Abstract

The paper proposes a novel segmentation method for image analysis with classification. Analysis of the scanned compound image and medical image segmentation is greatly interrelated with the subtraction of exact text/graphics, picture and subtraction of the anatomic structures. This research paper proposes an improved fuzzy cluster based segmentation method to repeatedly divide the background and foreground of the images of compound document images and medical images. This improved FCM clustering method is designed by including the spatial neighborhood details such as, a priori probability, spatial weights of the neighboring pixels of the center pixel, fuzzy membership of the current center pixel is calculated for classification. The proposed innovative metrics are used to calculate the exact accuracy of the segmentation scheme. Since the investigation, it is experimental that the proposed metrics are most appropriate for the evaluation of segmentation accuracy. The experimental results achieved from this work, prove that the proposed system performs segmentation effectively and successfully for the different component of compound images and medical images.
Keywords: Classification, Improved Fuzzy C-Means Clustering, Medical Images, Membership Index Creation, Scanned Compound Images