Towards better segmentation of object of interest using histogram equalisation and morphological reconstruction

Abstract

This study investigates a segmentation algorithm with histogram equalisation and morphological reconstruction filters to extract meaningful Object of Interest (OOI) from the low depth of field images. Low Depth of Field is a popular photographic technique to focus the foreground and defocus the background. The proposed method is designed to enhance the contrast using histogram equalisation and extract the edge using edge detection. The method removes the holes and bright patches using the morphological reconstruction filters. After the filter process, the method merges the focused foreground objects from defocused background. The image analysis was conducted on 50 natural images using the proposed method. The results show that the accuracy is close to the manual segmentation. The accuracy of segmented image using the proposed method and manual segmentation was also compared with other state–of–art methods.