Image group registration method fused with image grayscale features and structured representation

Abstract

An image group registration method that fuses grayscale features and structured representation, including the following steps: 1) Obtain the original image from the machine, obtain the structured representation of the image, and combine this structured representation with the original image As input data input; 2) Continuously perform affine transformation on the input image, and obtain the similarity measure of the transformed image, so that the transformed image has the optimal similarity; 3) Apply the image with the optimal similarity after affine transformation B-spline transformation is continuously performed, and the similarity measure of the image after B-spline transformation is calculated to optimize the similarity of the image after B-spline transformation, thereby obtaining the final precise registration image. The present invention provides an image group registration method that effectively increases effective information for registration, relatively reduces useless information, and has high configuration accuracy by integrating grayscale features and structured representation.

Description

technical field [0001] The invention relates to the field of image registration, in particular to a method for group registration of images. Background technique [0002] Image registration is to spatially calibrate two or more images of the same scene acquired by different sensors or the same sensor at different times, so that the information expressed in space by the corresponding pixels of the reference image and the floating image after registration is consistent. [0003] Registration methods can be divided into pairwise registration and group registration according to the number of input images. Pairwise image registration generally specifies a reference image and a floating image; group registration generally does not specify a reference image, and multiple images are registered at the same time. In the registration method, according to the image features used, it can be mainly divided into registration methods based on image geometric features, registration methods ...

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the disadvantages that the input image contains a large amount of image information, the registration accuracy is low, and a suitable reference image cannot be selected during registration, the present invention provides a method that effectively increases registration effective information, relatively reduces useless information, and configuration accuracy Higher Image Group Registration Method Fused with Grayscale Features and Structured Representation

Images