Edge Detection Based On Wavelet Transform And Mathematical Morphology

In transient state signal and in image analysis, the point of discontinuity often is one of important characteristics, they are located the edge of the important structure frequently. The edge precision, restraining noise, timeliness and the stability are the main performance index of edge detection. Traditional derivative operator, Like Roberts, Sobel, Laplacian and so on, timeliness is good, but the anti-jamming is bad. Afterwards appeared Canny, the operator although surpasses other derivative operators, but localization aspect unsatisfactory.Like this appeared two kind method of edges detection, one is obtains from the wavelet theory, detecting its edge through the mold maximum value method. This method may obtain the good effect, but the choice of wavelet has a stricter request. Another method is from the mathematical morphology, detecting the edge by the operations of dilation and erosion. The merit of the method is the high edge precision, but the extraction of detail edge has inseparable relation to the structure element.In this paper, we first propose a kind of simplification algorithm by wavelet mold maximum value; Then by the improvement of the morphology edge examination operator, we propose a kind of revision edge examination operator; Finally we unify the two methods, applying the wavelet decomposition technique, a grey image is decomposed into a set of wavelet approximation coefficients and wavelet detail coefficients. According to the feature of the image, we can strengthen these coefficients moderately. By reconstructing these coefficients, we propose a revisedalgorithm of edge detection based on wavelet theory and mathematical morphology.Furthermore, we analyze the result of the experiment, which prove the algorithm proposed in this paper is efficient on edge precision, detail edge detection and restraining noise.