Affine parameter estimation-based image registration method

Abstract

The invention belongs to the field of computer image processing, and relates to an affine parameter estimation-based image registration method, which comprises the following steps of: extracting characteristic points of images by adopting an SIFT algorithm, and matching the characteristic points of different images; establishing a mathematic model according to the concrete condition of affine transformation, and estimating affine transformation parameters therein; and performing motion compensation by adopting the estimated affine transformation parameters among the images to obtain the images after registration. The image registration method has stronger robustness for error matching, is simpler to implement and has better real-time.

Description

technical field [0001] The invention belongs to the field of computer image processing and relates to an image registration method. Background technique [0002] Image registration is the spatial alignment of two or more images of the same scene. It is widely used in medicine, remote sensing image analysis, image fusion, image retrieval, target recognition and other fields. At the same time, it is also the bottleneck of improving accuracy and effectiveness in the fields of multi-sensor image fusion, remote sensing image mosaicking, target change detection, and 3D reconstruction, and is a necessary preliminary work. Image registration methods are divided into registration based on image grayscale and registration based on image features. Among them, the image registration method based on grayscale information generally does not require complex preprocessing of the image, but uses some statistical information of the grayscale of the image itself to measure the similarity of ...

AI Technical Summary

Problems solved by technology

[0003] Although the above two methods have their own advantages and have achieved good results, both methods also have their own shortcomings. The registration method based on image grayscale is relatively slow in registration speed; The algorithm of the registration method is complex, and often due to incomplete feature extraction, the matching rate is low, and it is sensitive to mismatching. With the increase of the mismatching rate, the registration effect will decrease significantly.

This largely limits the application of the above two methods in practical engineering.

[0004] Since the existing grayscale-based and feature-based image registration methods are sensitive to mismatching, high requirements are placed on the accuracy of feature point registration. However, in many practical projects, such as aerial remote sensing aerial image registration Because the flight attitude of the aircraft is seriously affected by the airflow, there is a large distortion between adjacent frame images, and it is inevitable to mis-match when performing point registration. Large deviations will occur in image registration using traditional methods

Images