Hybrid Image Segmentation Method Based on Global Thresholding Method and Edge Detection using Canny Operator

Abstract

This paper presents a hybrid image segmentation technique based on edge detection and global thresholding methods. The segmentation process is simplified and becomes very fast due to applying the proposed method to discontinuous relationship between image pixels called edge pixels instead of whole pixels. Also, Our method is robust for selecting the initial estimate of threshold value depends on the similar relationship between pixels. The proposed method begins by applying edge-based segmentation by splitting the object using canny operator and merging the lines or edge linking using Hough transform. The resulted image will be binary image and represent the object that was separated from its background. Then intensity-based segmentation method is applied by partitioning the image histogram using global thresholding technique. The performance of the proposed method is illustrated using acquired images from a range of different shapes and compared with proposed methods.