Edge Detection Algorithm for Uneven Lighting Image Based on Vision Theory

Abstract

Many commonly used differential operators for edge detection which are effective and efficient on simple images can’t deal with relatively complex images. This paper put forward an edge detection algorithm for uneven lighting image based on vision theory. Firstly this paper introduces this vision theory, and established an effective computation model, through the preprocessing and threshold correction, the probability of losing edge at dark region and edge redundancy at bright region is reduced. At last, the algorithm based on modified Canny operator is given. Experiments indicated that this method, which is adaptive to complex image, has excellent brightness compatibility and has the ability to outstand the target. The division result is more consistent with person's visual feeling and has built a foundation for the following recognition.