Multilevel Wavelet Decomposition Based Harris Corner Detection Algorithm for Remote-sensing Image

Abstract

Because of the interference of the noise and the subtle texture, the Harris algorithm will generate much more false corners. An improved Harris algorithm based on multilevel wavelet decomposition is proposed. By multilevel wavelet decomposition, the high frequency components of image are extracted. These high frequency components with different scales are parsed, and the noises and subtle textures are suppressed while the original image edge features are kept. The final experiment shows that the improved corner detection algorithm can effectively reduce the number of false corners.