Research On The Seamless Mosaic Method For Extended Depth Of Field Images

Image mosaicing is a technology that finds out the same characters on the part-overlapped region of two adjacent images to stitch them into a large view image or panorama seamlessly. It's used to show the super angle vision as well as complete 360 degree angle vision of the scene. It has been widely used in space exploration, remote sensing satellite, geological survey, medical diagnosis and so on. Therefore, it has great significance to research on image mosaicing.Image mosaicing methods for extended depth of field images are discussed in this article. Several classical methods of image mosaicing and image fusion are analyzed in the paper, the image mosaicing on feature points is described in detail. Firstly, an improved way of Harris operation that the numbers of corners can be controlled is introduced. Secondly, a matching algorithm based on calculating the cross correlation coefficients between corners and a checking method of correct corner couples by calculating the distance between corner couples on images are discussed. Thirdly, the method of RANSAC (Random Sampling Consensus) to select the correct corner couples, to eliminate the wrong corners and to make sure the image translation matrix correctly is discussed. Finally, the seamlessly image fusion is discussed. Then a fast and precise task of the seamlessly image mosaicing with a big depth of field is accomplished.