A method for detecting man made target based on texture perception was proposed, which made use of the texture difference between man made object and natural scene. Firstly, the feature parameters about texture were extracted by grey level co-occurrence matrix, and then the texture feature was estimated by combing the above extracted parameters. Secondly, texture, brightness and direction were used to build spatial salient map by nonlinear feature fusion, and it could enhance the man made object with weak texture roughness and suppress the background with strong texture roughness. Some experimental results were included, and it showed that the method can accurately detect target against complex background.