Edge detection belongs to the category of image segmentation techniques which extracts edges/lines within the image. Generally speaking, an edge consists of the boundary pixels which connect two separate regions with different attributes such as gray levels, tristimulus values, or texture. Despite the large number of available researches, edge detection still remains a challenging endeavor in image processing for noise corruption and boundary uncertainty. In this paper, a new gradient-based framework is proposed for edge detection. In this paper, we proposed a new edge detection algorithm. First, the image is smoothed by a Gaussian filter for noise suppression. After computation of the image gradient, a non-maximum suppression algorithm is performed on the gradient amplitude to extract some edge seeds. Once the smoothed image is filtered by using the LoG filter, every seed point with no zero-crossing match in the resultant image is discarded. Finally, the edge map of the image is obtained by using an edge following algorithm. This method has considerable potential dissociation between retail removed edges because it follows the edge of the algorithm. Experimental results show that the proposed method performs better detection and removal dissociation between edges than all these operators under almost all scenarios.

Keywords