Multi View Optical Image Registration And Object Change Detection

With the development of computers and communication technologies,the fields of application of images and video have become more widespread.However,humans need more and more information from images,and single-view images cannot meet these needs.In order to obtain more information on a target,people usually shoot the same scene from different angles and get more information on the same target.The long-term development of image registration algorithm has also formed a stable framework,which has been widely used in image fusion,image correction and other occasions.However,the accuracy of traditional registration algorithm for multi-view optical image will be affected,So registration based on this multi-view optical image has become a hot topic of research.At the same time,the object change detection method has been widely used as the basis for dynamic image analysis and target tracking.Therefore,it is of great theoretical significance and practical value to study a registration algorithm which can be more accurate and stable for multi-view images.In this thesis,based on scale-invariant feature registration algorithm,a multi-view image registration algorithm suitable for object change detection is studied.The main research work completed is as follows:(1)The basic theory of image registration algorithm is studied,the technological process of image registration algorithm is analyzed and studied,and the evaluation criteria of image registration algorithm are studied.(2)The feature-based image registration algorithm and related algorithms are studied.The image registration algorithm based on SIFT feature points,the image registration algorithm based on SURF feature points,the feature point matching algorithm and the solution of transformation matrix are analyzed and studied more deeply.Some simulation experiments were performed,and the intuitive conclusion was obtained based on the principle of the algorithm.(3)Based on scale-invariant feature extraction algorithm,combined with nearest neighbor matching algorithm and perspective transformation,the original scale-invariant feature-based registration algorithm is improved to be more suitable for multi-view image registration.Combined with the actual image data and compared with the original algorithm to carry out a test.The comparison includes the accuracy test of the algorithm,the time complexity analysis of the algorithm and the robustness of the algorithm.(4)The basic principle and main steps of object change detection are studied,and applied multi-view image registration algorithm to the object change detection method.(5)Complete the basic model and algorithm library of image processing and computer vision,and complete the design,development,testing and integration of image feature extraction module,image registration and image fusion module.