Edge detection: Sobel, Prewitt, and Roberts; a comparison
- Three different approaches to gradient as an edge detector were implemented: Roberts, Prewitt and Sobel operators (in four
different biasing).
Furthermore, normalization was applied to each operator result to better show the edge on 110.tif. This
normalization procedure consists of calculating the max value of the result matrix and dividing all onto the value, and then
the absolute value of such matrix will be the normalized result. The same names were considered for each code. Since these
operators, each consists of 4 biased operators, their results were combined using
the as a final result. Results are shown for 110.tif:
- The combined results of above-mentioned operators were fed to the program for snake calculation and all failed, since
edge map is not quite suitable to be applied for GVF calculation (on 110.tif). Results:
- A Matlab code to calculate gradient of an image using directly the vertical and horizontal axes first derivative.
This program is called "gradientm." Again the result of applying this code on 110.tif was used for GVF
snake algorithm and again it failed since it does not clearly detect the edges.
