Biometrics and Bioinformatics

Abstract

Edge detection is one among the all fundamental issues of digital image. This paper provides an algorithm that improved the classical Sobel operator whose defect is rough effect for edge. In this paper a palm image is used and the sobel operator is improved with the combination of Otsu method and means filter this improved algorithm makes the edge detection for salt and pepper noise image and it effectively overcome the problem that the sobel operator is only sensitive to vertical and horizontal direction and it combines the advantage of mean (Average filter) to remove the salt and pepper noise and thus through Otsu a segmented image is formed through histogram thresholding and then the effect of the whole process is better than the other edge detection methods. And at last the improved sobel operator is also compared with the other edge detection methods (Roberts, LOG, prewit) and also with canny but the results with these operators are not satisfactory and thus the improved sobel operator detects the edged properly.

Keywords

Full Text