Enhanced Edge Detection Technique for Satellite Images

Abstract

Traditional Canny edge detection algorithm is sensitive to noise, therefore when filtering out this noise weak edge information gets lose easily. In response of these problems an improved canny edge detection algorithm was proposed by Weibin Rong, Zhanjing Li, Wei Zhang and Lining Sun. The improved canny algorithm introduces the concept of gravitational field intensity to obtain the gravitational field intensity operator while replacing image gradients. Based on standard deviation and the mean of image gradient magnitude were put forward for two kinds of typical image among which one has the rich edge information and another has relatively poor edge information. The experimental results says that algorithm preserve more edge information but it’s computing speed was relatively slow. In response of these problem this paper proposes an Enhanced edge detection algorithm which uses the concept of double derivative Gaussian filter and is much faster than the improved canny algorithm. The Experimental Analysis has been done based on time, peak-signal to noise-ratio (PSNR) and entropy which states that algorithm preserves more edge information and is more robust to noise.