Abstract: On the basis of the traditional methods for edge detection of potential-field data, various field source edge enhancement techniques have been studied to improve signal-to-noise ratio (SNR) and localization accuracy. But problems still exist, such as noise interference and divergence of source edges. The authors made an improvement based on the traditional methods of edge detection using wavelet combined with traditional methods to enhance the ability of noise suppressing, and put forward power transform to make edges convergent effectively. In this paper, the principle was introduced and the method was applied to edge detection. Three commonly used numerical methods for edge detection were selected and compared with the modified methods in edge detection effect with the model data, gravity anomalies of Sichuan basin, and the magnetic anomalies in Zhurihe area. The results showed that the method of wavelet combined with power transform could effectively detect the edges of geological bodies in the study area. It plays a good role in noise suppressing, and can be used to identify the edges with convergence. The proposed method improves the accuracy of edge detection and obtains satisfactory results.

Key words: edge detection, wavelet transform, power transform, noise suppression, edge convergence