A weak texture image registration method based on a spatial time sequence model

Abstract

The invention provides a weak texture image registration method based on a spatial time sequence model. The weak texture image registration method comprises the following steps: obtaining a referenceimage and an image to be registered; Detecting feature points of the reference image and the to-be-registered image by using a spatial time sequence model method; Determining the directions of the feature points, and constructing a composite feature point descriptor containing gradient direction information; Carrying out feature point matching on the reference image and the to-be-registered imagethrough feature searching; Removing wrong matching pairs to obtain an optimal matching point pair and a transformation model; And transforming the to-be-registered image according to the transformation model to obtain a registration result. Compared with a region, the method has the advantages that image features are used for replacing direct gray scale operation, and the feature method is high inefficiency and accurate; Compared with a feature method, the dependency degree of the algorithm on the significant features is reduced, manual adding of auxiliary features is not needed, part scratching is avoided, the robustness of the algorithm is improved, and automatic registration can be achieved.

Description

technical field [0001] The invention relates to the field of image registration, in particular to a weak texture image registration method based on a space time series model. Background technique [0002] Image registration is the process of superimposing two or more images of the same scene at different times, different viewing angles, and different sensors. It realizes the geometric alignment of the reference image and the image to be registered. Image registration has a wide range of applications, including machine vision, 3D reconstruction, remote sensing image processing, object classification and retrieval, image understanding and fusion, etc. Image registration methods include registration methods based on regions, based on line features or surface features, and based on point features. [0003] The region registration method directly uses the gray information of the image to calculate the similarity of the image. The processing process is as follows: firstly select ...

AI Technical Summary

Problems solved by technology

However, the processing "traces" contained in the part image are usually relatively weak, and the gray level of the image is not much different. Even if preprocessing methods such as image enhancement are used, the edge is still difficult to detect

In addition, the processing "trace" has obvious periodicity, so only using the information of the edge itself to construct the feature point descriptor may lead to the failure of registration

Images