In the recent years,the usage of high resolution images has made a significant contribution to our improved archaeology.Given that the archaeological mark is critical for the analysis and predetermination of ancient sites,its edge feature is required to be detected against the surrounding.In traditional methods,time consuming and costs a lot.The archaeological marks are not always identified,especially when the edge features obscure in low contrast environment.The edge detection technology can not only highlight the edges of ar-chaeological site,but also filter the noises from the surrounding environment.Therefore,it is particularly useful to extract the weak edge information automatically from high resolution image by computer algorithm.The representative methods based on edge gradient are Sobel,Roberts,Prewitt and Laplacian operator.However,they have the same drawback that it is sensitive to noise and the edge is not located accurately.The Canny operator can reduce the noise and keep the edge intact to a great extent and perform well to most of archaeological sites.We choose the classical relics in Xinjiang for comparison of different edge detection methods.It is concluded that the Canny operator achieve the best result in different envi-ronments in high resolution image from experiments.And it is applicable for archaeological prospection and feature extraction.