Abstract: Aiming at the problem that the traditional discontinuous edge detection algorithm was used to detect the edge contrast of the enhanced image, but it was only suitable for detecting the edge of the image with no strong change of gray value and ordinary noise, and the detection performance had the limitation, we proposed an improved algorithm for intelligent detection of discontinuous edges of blurred images. Firstly, the estimation of the noise variance of the image was obtained by the gene
ralized cross validation criterion to discriminate the Gauss noise in the image, and the adaptive fuzzy filter was used to make fuzzy filtering of the noisy image. Secondly, the improved image edge detection algorithm of blurred image was used to acquire the edge of the blurred image by formulating the image edge detection strategy according to the situation of image noise. Finally, the discontinuous edge detection algorithm of blurred image of gray morphology was used to detect the expansion, corrosion and morphologic gradient discontinuous edges of the blurred image edge caused by the uneven change of gray value. The experimental results show that the proposed algorithm has high noise immunity, and the results of discontinuous edge detection of the blurred image are clearer and more complete.

Key words: blurred image, discontinuous edge, edge detection, generalized cross validation criterion, adaptive fuzzy filter, gray morphology