A self-adaptable improved arithmetic based on wavelet transform in the image edge detection

Compared with traditional edge detection methods, the method based on wavelet transform is a high precision way. The scale, however, impacts the effect of output imagine. The high frequency could be controlled when the scale is large, and the SNR of the image is relatively great. But the precision of location is lower. On the other hand, the location could be realized precisely when the scale is small, and the detail of the image could be extracted. But the ability of denoising is weak. A self-adaptable scale selection method is given in this paper. The edge detection of an image can be realized well by different properties of wavelet transform on different scales. Accordingly, the gauss filter is adopted to pre-treat the figure, and the self-adaptable multi-scale wavelet transform is introduced to process the after-pretreat figure. Result of experiment shows that a highly quality edge figure by smooth disposal of different scales is received.