Study On Point Feature-based Image Registration And Application Of Land Use

Remote sensing image registration is the process of aligning operations between two images, which is a variety of remote sensing image processing techniques(such as change detection, image stitching, image updates, etc.) basis and prerequisite. Since the 1990 s, land use and land cover change has been a global change research frontier and hot. Land use change information automatically detect, based on remote sensing image data with high accuracy for the premise of registration. Thesis by analyzing and summarizing remote sensing image registration techniques, focusing on the characteristics of point-based image registration techniques.The key question paper through a comprehensive description of the feature point selection, description and matching feature-based image registration technique, devised SURF feature points based image registration technique improvement. Paper describes the detection of land-use changes, the use of a comprehensive experimental study of the contents verified. Firstly, the use of the improved image registration algorithm for remote sensing image with high accuracy automatic registration, then the type of land use change detection using change vector analysis.Achieved results and contents of the paper include the following:①In the feature point extraction process, designed the feature point extraction block-based strategy approach. To ensure that the extracted feature points distributed uniformly in the entire image, thereby improving image registration accuracy.②In the feature point matching process, devised a method to improve the use of the relative distance of the image registration techniques. This improvement by removing the mismatching points, effectively improve the accuracy of image registration.③In the land use change detection process, a method using a paper change vector analysis. Change vector analysis of the most critical is to determine the optimal segmentation threshold issue, the paper has a certain choice of straightforward universal cycle segmentation method to automatically determine the best segmentation threshold.Papers by land use change detection experiments designed to verify the paper image registration techniques in getting a high-precision registration of remote sensing image after the image change detection of land use types using change vector analysis, the final results obtained meet Land use change detection requirements.