The Study Of The Edge Detection Based On Mathematical Morphology And Morphological Wavelets

Content:In this paper, we propose two kinds of edge detection algorithm. One is the edge detection based on the edge focusing, its basic thought is as follows: it focuses edge results detected by the LoG operator and the Roberts operator, instead of the iterative process of the multi-scale algorithm, and introduces mathematics morphology operator to enhance its anti-noise ability . The classical edge detection operators, such as the LoG operator, the Canny operator and so on, are sensitive to noises. Multi-scale edge detection algorithm can decrease the affection of noises, but it has high timing complexity because it has to iterate many times. Our algorithm overcomes above problems. The other one is morphological wavelets based on the nonseparable sampling. Its basic thought is as follows: first choosing the sampling matrix in the interesting edge direction; then analyzing the original image by nonseparable morphological wavelets to get the high frequency data with different directions; after that taking the low frequency part to zero; finally synthesizing , so we can get the edge of the original image. The advantage of the algorithm is to avoid the edge excursion to a certain extent since using small convolution kernel. This is difficult to do by space filters.