Sonar Image Edge Detection Method Based On Wavelet Analysis

The edge,which is used to describe the main outline of the image is the key to subsequent image segmentation and pattern recognition.For sonar images,extracting edge is equivalent to extracting bright and dark areas of the target from complex submarine reverberation zones.Speckle noise has the greatest influence on the sonar image.Effectively removing speckle noise is essential for improving the edge detection of sonar images.Since the wavelet transform has good localization characteristics in both the time domain and the frequency domain,it is easy to detect the detailed information of the image.Therefore,the wavelet transform is used to filter and detect the edges of sonar image.The main research contents of this thesis are as follows:In the aspect of sonar image denoising: Based on the semi-soft threshold function,a new threshold function is proposed to obtain one optimal method.Combining the one optimal method,the improved Frost filter is applied for another method.Experiment results show that the edge preservation ability and the amount of information of the two improved methods are superior to the classical algorithm and the comparison algorithm.In the aspect of sonar image edge detection:The first derivative of Gaussian function is used as the wavelet function and the standard deviation σ is the scale parameter.Combined with the gray-scale characteristics of the image,an algorithm for adaptive scale parameter σ has been proposed.This algorithm optimizes the edge of the adaptive scale parameter,and the edge detection of the second method's filtering result was done.The initial edge is obtained.The initial edge are segmented and the final edge is obtained using the block threshold method.Experiments show that the edge extracted by the improved method are superior to the classical algorithms mentioned in this thesis in terms of positioning and signal-to-noise ratio.