Optimal threshold selection algorithm in edge detection based on wavelet transform

This paper presents an optimal threshold selection algorithm, which selects the de-noising threshold according to the turbulent degree of detected edge points, in edge detection based on wavelet transform. First of all, adjacent domain division algorithm (ADDA) and parabola fitting algorithm (PFA) are used to separate edge curves from each other after wavelet transform. Then, the entropies, corresponding to different possible thresholds are computed according to the number and length of all the edge curves detected above. The threshold, which giving the minimum entropy, is selected as the optimal one to filter the noises. The experimental results show that our method can get better threshold than other ones, in a subjective view.