There are mainly three methods in Image Retrieval Based on Content, i.e., retrieval technics based on shape, texture and colour features. Different images and uses require selection of retrieval methods through different featuers. Therefore, in this paper, we first analyze the features of the image itself, trying to find its most obvious features, and then, according to the analysis results, apply suitable Retrieval Method Based on Feature to retrieving this image to make the Image Retrieval Technic more intelligent and thus to get better effects. This paper puts forward calculation methods of Colour Histogram, Shape and Dierction Energy Histogram and Cumulative Run-Length Histogram, then briefly describes the Intelligent Selection Method Based on Image Feature, and states that the final result of the test is satisfying.