Wavelet Multi-Scale Edge Detection Using Adaptive Threshold

This paper, taking advantage of multi-scale character of wavelet transform, performs multi-scale wavelet transform of the given image. A wavelet transformation of a given image produces an image that is proportional to the gradient of the smoothed image, and then edges can be detected by locating the local maxima of the magnitude of the resultant image. The wavelet function used in the method is derived from a smoothing function. The edges at different scales are extracted and filtered by using an adaptive threshold, and are synthesized to obtain the final edge image. The experiment proves that the method preserves abundant details and good noise resistance simultaneously.