Study On Image Mosaic Based On SIFT And Contourlet Transformation

Computer vision is through computer simulation of the human eye’s visual function, to extract information from the picture, the objective world of3D scenes and objects form the shape and movement recognition. The contents of this study is an important branch in the field of computer vision which is also based on an important technology in the method of image rendering, image stitching technology. The image stitching technology is a set of images to the same sensor at different times or different sensors at the same time shooting with overlapping regions spliced into a high-definition, wide viewing angle, sharp images. With image stitching technology continues to evolve, in the field of computer vision, virtual reality technology, medicine, meteorology, geological survey, video compression and transmission have been widely used.Image registration and image fusion is image stitching in the most important and foremost the two processes.Firstly, image registration part, the subject image registration method based on image features, the feature extraction method is image based on SIFT features improved operator feature point extraction, SIFT feature descriptor improvement, reducing the characteristic dimension, and then the RANSAC algorithm to purify the feature points to achieve the exact match of the image.Secondly, this topic to this algorithm and SIFT feature-based image registration for a comparative experiment, a series of transformations of the image, obtained through data analysis, the proposed algorithm in the time of registration and registration accuracy are increased.Thirdly, image registration is complete, but in general image registration color patchwork, to make smooth seamless mosaic image, seamless, we need to image registration, image fusion. Used in this paper is based on a combination of Contourlet Transform and Teager Energy First Contourlet transform image decomposition, the low frequency components directly using the weighted average fusion method. Teager Energy for the high frequency components of the row factors and column coefficient calculated according to the calculated results using different fusion rules, the final image fusion.Finally,in this paper, wavelet image fusion comparative experiment, select a certain evaluation criteria for the quantitative analysis of experimental results show that this algorithm is compared with the wavelet decomposition algorithm, not only can save the original image edge and texture information, can also get a good blend of visual effects.The experimental results show that the system is feasible and accurate image stitc hing, and has a good robustness, and image stitching in the complex environment.