An improved ant based algorithm and its implementation for the Broken Edge Linking Problem

Abstract

As a robust image processing technique, the broken edge linking technique is considered to be the complementary one to the edge detection technique. Here in case of the edge linking problem, we form closed contours by linking the broken edges. This action indeed needed to split the areas in the image into different parts known as segment. However the traditional edge linking technique is always succeeded by the traditional edge detection technique. The traditional edge linking technique, for all time, neglects some significant part of the boundaries to consider, as a result, appropriate and perfect solutions to the edge related linking problem cannot be reached always. With this article, we presented an improved method (technique) for the edge related linking difficulty (problem), which is actually the expansion of the original old Ant System (AS) algorithm. Here in the experiment, we mainly consider the two decisive and significant factors: first one is the length of the linking edge as well as second one is the grayscale visibility of the pixels, apart from an sobel edge binary image , so that the effective solution evaluation and enhancement of the overall performance can be achieved. The experiment showed the expected result, are presented herewith this to ensure the successfulness of the projected improved technique /algorithm.