Image registration method based on feature points

Abstract

The invention discloses an image registration method based on feature points, and relates to the technical field of image registration. The method comprises the following steps: a, preprocessing a reference image and an image to be registered by adopting a wavelet transform threshold denoising method; b, extracting feature points by an SIFT algorithm; c, describing the feature points by adopting adeformation dimension reduction method; d, performing rough matching on the feature points according to cosine similarity; and e, eliminating part of mismatching by adopting an improved RANSAC algorithm to obtain a matching point pair with relatively high matching precision. Images are preprocessed by using a wavelet transform threshold denoising method before feature points are extracted, part of noise in the image is eliminated, so that the purity of the feature points is improved, a deformation dimension reduction method is provided for describing the extracted feature points when the feature points are described, SIFT descriptor dimensions are reduced, the algorithm operation time is shortened. The matched feature point pairs are optimized by adopting a method of combining rough matching and fine matching, so that the matching accuracy of the algorithm is improved.

Description

technical field [0001] The invention relates to the technical field of image registration, in particular to an image registration method based on feature points. Background technique [0002] Image registration is the process of matching and superimposing two or more images acquired at different times, different sensors (imaging devices) or under different conditions, and is widely used in the fields of computer vision and image processing. Image registration mainly includes three parts: (1) feature extraction (2) feature matching (3) parameter estimation. At present, image registration methods mainly include: 1) region-based registration method; 2) feature-based registration method. Among them, the region-based registration method mainly uses the gray information of the image to establish the similarity between two images. Similarity measure, and then count the maximum or minimum similarity measure to transform the parameter values ​​of the model to achieve the purpose of ...

AI Technical Summary

Problems solved by technology

Zhang Wenyu et al. proposed a scene matching algorithm for UAVs based on CenSurE-star for the situation that the traditional scene matching algorithm with local invariant features has many redundant points, poor real-time performance, and not outstanding resistance to geometric transformation. Although the matching accuracy is improved, the real-time performance Better, but at the expense of time

Images