A Proposed Methodology for Multi-Temporal Image Change Detection by Using Image Stitching Through Pixel Block Localization

摘要

In alteration discovery analysis, image misalignment still remains a pesky issue, although such discovery would be is entirely determined by the accuracy of image registration. If the procedure of alteration detection is done on images that are not aligned properly, problems may accrue in the form of missed or wrong alterations. This, in turn, would have an adverse effect on the correctness of actual change detection. This paper's aim is to propose a methodology about the recognition of change technique that is expected to curtail the influence exerted by improper alignment of images. And to effect this, the method of detecting alternations or changes based on pixel is being advanced, as this suggested method would deal with errors arising out of mis-registration effectively. The technique proposed here involves image stitching by combining two or more images of the same scene as a single larger image. Block matching technique is a way of locating matching blocks in a sequence of image for the purpose of image estimation. To spot the alteration in image, the use of image differencing is being advocated. It is anticipated that the suggested experimental steps would yield a top rate of accuracy.