Performance of Ant System over other Convolution Masks in Extracting Edge

front end of most vision systems consists of edge detection as preprocessing. The vision of objects is easy for the human because of the natural intelligence of segmenting, pattern matching and recognizing very complex objects. But for the machine, everything needs to be artificially induced and it is not so easy to recognize and identify objects. Towards Computer vision, the Machine needs pattern recognition; extracting the important features so as to recognize the objects, where the boundary detection or the edge detection is very crucial. Edge detection is finding the points where there are sudden changes in the intensity values and linking them suitably. This paper aims, at presenting a comparative study on the Gradient based edge detectors with a swarm intelligence. Though, these detectors are applied on to the same image, they may not see the same edge pixels. Some detectors seems to be good only for simple transparent images which are less noise prone, and marks pseudo and congested edges in case of denser images. Hence it would be appreciated, to have an edge detector, which is sensitive in detecting edges in majority of the common types of edges. With this in mind, the authors propose a new edge detector based on swarm intelligence, which fairly detects the edges of all types of images with improved quality, and with a low failing probability in detecting edges.