Nonlinear gradient-based edge detection algorithms in the telesign system

Abstract

Edges characterize object boundaries, and an edge point can be thought of as the pixel location of abrupt gray-level changes. Most edge detection algorithms are based on generation of gradient images and thresholding of gradient images. There are two problems in detecting edges of a video sequence in a telesign system. One is that the linear gradient does not match Weber's law, and the other is difficulty in finding an appropriate threshold efficiently. In this paper, we introduce a concept of nonlinear gradient to match Weber's law, i.e., to detect perceptual edges. It is easy to determine a proper threshold for the nonlinear gradient images. Based on the advantages of the nonlinear gradient, the author found an important application in the telesign system.