Research On Image Fusion Algorithm Based On SIFT And Application In Ethnic Characteristics

Image mosaic technology has a wide range of application in many areas, such as satellite remote sensing image, medical image processing and so on. In order to realize image mosaic, we need to recognize images to search the overlapped contents of images and analyse the complex the transformation relationship between images from the overlapped contents. SIFT is an image recognition method based on multi-scales proposed by Low who summarized the existing image recognition technology. This algorithm can effectively remove the influence of light, rotation and scale, and then realizes image matching.In this paper, we view SIFT as the local features of images to construct an affine calculation method based on it. We propose the consistency filtering algorithm based on K-Means according to SIFT feature matching. Using the method, we can search the stable matching and filter the invariant SIFT feature points in images. In allusion to the instability of the edge of multiple image mosaic, we use the grid to overlay the matching feature points, and select affine points in matched SIFT feature points by the method, then, build the corresponding affine transformation. In the test experiment, this paper gives the method of quantitative measurement the stability of the filter results and fitting degree of edges. Experimental results show that the proposed method can effectively deal with the problem of image mosaic that is based on invariant SIFT feature points.Finally, the paper proposes a new method of ethnic recognition of human faces based on SIFT local features. And gives the system of local invariant feature extraction and analysis to extract the SIFT local invariant feature points of faces, and then completes position matching of local features in faces.