VLSI Implementation of Improved Sobel Edge Detection Algorithm

Image processing techniques are widely used to extract relevant information from digital images and also to enhance the image. Edge detection algorithms are used to detect the edges of images and its applications include image segmentation and data extraction. Sobel is one of the most widely used edge detection algorithms. But some edge information will be lost in the traditional Sobel because only the edges in X and Y directions are detected. To overcome this, improved Sobel algorithm having 8-directional template is used. In this work, we analyze the performance of improved Sobel with other traditional edge detection algorithms. By comparing the results in Matlab 2019b, it is seen that the Sobel in eight directions gives the best results in terms of MSE, RMSE, entropy, SNR and PSNR. The improved Sobel has a high entropy of 0.6-0.8 implying more information in output. Then we implement this improved Sobel technique using ModelSim 10.4a and synthesis is done using Xilinx Vivado 2020.2.