In order to solve the problem of incomplete edge and unclear outline in remote sensing image edge detection using classical edge detection operator

an edge detection method based on multi-scale adaptive equalization is proposed using the multi-scale feature of remote sensing image ground objects. Firstly

morphology reconstruction and gradient operation of gray-scale remote sensing image is carried out by adopting multi-scale and different directional structural elements. Multi-images owning edge features of different scales and directions is obtained. Secondly

compute the weight ratio of grey value variance of each image

adaptive equalization summing is carried out using weight ratio on multi-images

acquire edge profile. The experimental results show that

comparing with classical edge detection operator

the proposed method can detect image edge more clearly and completely

and can effectively suppress the salt and pepper noise caused by the collection and transmission of the remote sensing image

and has certain anti-noise performance.