Abs-Laplacian and Robert’s cross operator offers high speed edge detection capabilities with comparable speed-quality tradeoffs

Abs-Laplacian is a newer technique for detecting edges. While the Sobel and Prewitt seems to be used predominantly in image processing due to better edge detection. However they need higher amount of complexity twice than that of abs-Laplacin. Moreover, qualities of images are no better than the newer kernel. While the new technique seems to perform better in speed and quality, there seems a bit of competition with the 1 order kernel i.e. Robert’s cross operator. In terms of speed, Robert needs only 3 computational units against 7 for abs-Laplacian, but in terms of image quality analyzed by determining the intensity deviation of Robert’s edge or abs-Laplacian edges with respect to the Sobel’s edges shows that Robert’s kernel show major deviation indicating a low quality image while the new technique showed negligible deviation. Intensity profiles of edges detected for various kinds of images shows similar behavior for each kernel which also suggests that the inherent nature of kernels tend to remain the same for whatever complex images we deal with. In conclusion, the results show that there exists a speedquality tradeoff between the abs-Laplacian and the Roberts operators.