The following functions are used for detecting edges in an image by convolving eight kernels in sequence over the image matrix and then comparing the results.
Comparison edge detection can be used for enhancing image features that depend on boundaries and directions of boundary gradients.
• compgrad(M)—Performs edge detection by comparing the gradients of the eight neighbors on matrix M.
• kirsch(M)—Performs edge detection by kirsch convolution and comparison on matrix M.
• robinson3(M)—Performs edge detection by 3 x 3 Robinson convolution and comparison on matrix M.
• robinson5(M)—Performs edge detection by 5 x 5 Robinson convolution and comparison on matrix M.
Arguments
• M is an image matrix.
Additional Information
• The functions return a matrix containing the absolute maximum among the eight convolutions at each pixel. This makes these edge finders sensitive both to large differentials and to the direction of the differential. The perimeter of the image is untreated, since the kernels do not fully overlap there.
• The calculation time of these functions can be long because eight convolutions are performed per pixel.