Edge detection is an important part of the image processing, pattern recognition and computer visions. The edge detection results of the traditional image edge detection methods in general. In order to better detect the image edge, presented a method of image edge detection based on edge image fusion on the basis of the analysis of traditional edge detection algorithm. The original image is decomposed by dyadic wavelet get low-frequency sub-image at first. Then the original image and the low-frequency sub-image are enhanced by histogram equalization method, the enhanced original image and the enhanced low-frequency sub-image are detected using Canny operator get edge images. Finally, the two edge images are fused into one complying with relevant rules get an intact edge image. Results show that this method is much better than directly using alone the Canny operator or wavelet transform edge detection method to the original image.