Research On Ghosting Processing Method Of Moving Object In Panoramic Image Mosaic

With the development of digital image technology, image mosaic has become a popular image technology, widely used in many fields, such as:aviation, photogrammetry, remote sensing image processing, medical image processing, machine vision and so on. Image mosaic technique aims to make multiple images together into a high-resolution panorama, it solves the problems of limitations which caused by the view field ofordinary camera when taking pictures, thusthe captured images can not achieve the required precision and large viewing angle. But image mosaic technology makes it possible that it can obtain highly accuratepanoramic images only by ordinary camera.However, not all image sequences captured have stationary background,there alwaysexists moving objects, thus producethe mosaic image ghosting phenomenon, which we need special treatment in the process of the imagemosaic. This paper mainly focuses on the panoramic mosaic algorithm andoptimalize the traditional method, we deal with the problemof panoramic mosaic with dynamic objects. Mainly include the elimination of false matchesand the ghost produced after fusion. The main contentsof this paper are asfollows:Firstly, introduced research status and application technology of image mosaic, summed up the main flow and the key steps ofimage mosaic, and introduced the image pre-processing and camera calibration method, the main method of image matching, as well as the main method of image fusion.Secondly, we proposed an image mosaic method in case of which contain large moving object, It combinedclustering withgeometric characteristics of the image in order to optimize the process, separating the moving object and then fused.Then, for continuous video frame images, the machine learning Markov random motion estimation is applied to extract the binary template of moving target, combined with edge features in image match, using active contour model to generate an optimal seam line in the fusion phase.Finally, we also analyzed the method ofeliminating the accumulated error of image mosaic,and obtain a more precise panoramic image.