Moving Image Feature Extraction Method Based on Wavelet Transform

The accuracy of traditional image feature extraction algorithm for accurate recognition of moving images is not high. To solve such problem, a target recognition algorithm based on edge contour feature segmentation using wavelet transform technology is proposed in this paper. Firstly, through multi-scale moving target edge detection, the process of moving target image acquisition and preprocessing is standardized. Then, using the information redundancy of stationary wavelet transform to remove noise, an image denoising method based on adaptive soft threshold is put forward. Finally, the improved method is tested in the simulation environment. The results show that the root mean square error of pixel feature matching is low, the corner extracted under different thresholds is reliable, and the motion estimation accuracy is better than that of similar algorithms.