In order to further improve the performance of noise suppression and edge detection in image processing, a new morphology image edge detection algorithm based on multi-scale and multi-direction structure elements is proposed. The method fully utilizes the erosion, dilation, opening, closing and its transformation and combination operation based on the directional differences of structural elements in mathematical morphology. In order to improve the signal-to-noise ratio and edge details of image, the image denoising, edge extraction and other preprocessing operations are carried out. The image initial contour is obtained by using multi-scale and multi-direction structure elements recursive morphological filtering. Image edge detection is carried out by using multi-scale morphology and multi-direction structure elements. Experimental results show that the proposed algorithm has more powerful anti-noise performance and can extract the edge information effectively and accurately.