CN-121982079-A - Image registration method and device based on ORB algorithm and electronic equipment

CN121982079ACN 121982079 ACN121982079 ACN 121982079ACN-121982079-A

Abstract

An ORB algorithm-based image registration method, an ORB algorithm-based image registration device and electronic equipment relate to the field of image processing and can improve the accuracy of image registration. The method comprises the steps of obtaining a reference image and an image to be registered, wherein the reference image is an image with an overlapping area with the image to be registered, respectively extracting feature points of the reference image and the image to be registered by adopting an ORB algorithm, extracting edge points in the neighborhood of the feature points by adopting a Canny operator, constructing feature descriptors of the feature points according to the edge points in the neighborhood of the feature points, matching the feature points of the reference image with the feature points of the image to be registered through the feature descriptors, determining matching point pairs, calculating a transformation matrix between the reference image and the image to be registered through the matching point pairs, and registering the image to be registered based on the transformation matrix.

Inventors

ZHOU HUIXIN
SUN RUIYANG
DUAN NIAN
WANG BINGJIAN
QIN HANLIN
LAI RUI
GAO MENGYANG
XIANG PEI
YANG TIANFANG
QI SHUXIA
SONG JIANGLUQI
LI HUAN
WANG JIAYI
LI QIANYOU

Assignees

西安电子科技大学
中国石油大学(北京)克拉玛依校区

Dates

Publication Date: 20260505
Application Date: 20251230

Claims (10)

1. An ORB algorithm based image registration method, comprising: acquiring a reference image and an image to be registered, wherein the reference image is an image with an overlapping area with the image to be registered; Respectively extracting feature points of the reference image and the image to be registered by adopting an ORB algorithm; Extracting edge points in the feature point neighborhood by adopting a Canny operator, and constructing feature descriptors of feature points according to the edge points in the feature point neighborhood; matching the feature points of the reference image with the feature points of the image to be registered through the feature descriptors, and determining matching point pairs; and calculating a transformation matrix between the reference image and the image to be registered through the matching point pairs, and registering the image to be registered based on the transformation matrix.
2. The ORB algorithm based image registration method of claim 1 wherein the constructing feature descriptors of feature points from edge points within the feature point neighborhood comprises: Determining a feature description area of the feature points, wherein the feature description area comprises a plurality of pairs of pixel areas; Rotating according to the main direction of the feature points, calculating the average gray scale of each pair of pixel areas, comparing the average gray scale, and determining a binary value according to the comparison result to obtain a first feature descriptor; determining a preset sector area in the feature description area, rotating clockwise with a preset step length, respectively comparing the number of feature points and the number of edge points included in adjacent sectors, determining one-bit binary according to a comparison result, and acquiring a second feature descriptor and a third feature descriptor; And combining the first feature descriptor, the second feature descriptor and the third feature descriptor to obtain feature descriptors of feature points.
3. The ORB algorithm based image registration method of claim 2 wherein the feature description area is a circular area centered around the feature point, the circular area is divided into four quadrants, 64 pairs of pixel areas of a predetermined size are randomly selected in each quadrant, and the first feature descriptor is 256 bits.
4. The ORB algorithm based image registration method of claim 3 wherein said determining a one-bit binary according to the comparison result comprises: When the average gray level of the first pixel area is smaller than that of the second pixel area, the binary bit is 1, and when the average gray level of the first pixel area is not smaller than that of the second pixel area, the binary bit is 0, wherein the first pixel area and the second pixel area are any pair of pixel areas.
5. The ORB algorithm based image registration method of claim 1 wherein the matching the feature points of the reference image with the feature points of the image to be registered by the feature descriptors, determining the matching point pairs, comprises: calculating the hamming distance between the feature points of the reference image and the feature points of the image to be registered through the feature descriptors of the feature points, and determining the nearest hamming distance and the next nearest hamming distance of the feature points of the reference image; if the ratio of the nearest hamming distance to the next nearest hamming distance is smaller than a preset threshold value, determining the feature point corresponding to the nearest hamming distance as a matching point pair.
6. The ORB algorithm based image registration method of claim 1 further comprising: dividing a reference image through a quadtree to obtain a plurality of subareas; Calculating the geometrical slope of the connection line between the matching point pairs of the feature points in each sub-region, extracting a plurality of feature points according to the sequence from small to large of the Hamming distance, calculating the average value of the slope, and if the geometrical slope and the average value of the slope of the feature points do not meet Deleting the matching point pair of the characteristic point; If the hamming distance between the matching point pairs of the feature points is smaller than omega 2 times of the maximum hamming distance of the feature points, the matching point pairs of the feature points are used as strong matching points, and the strong matching points omega 1 <ω 2 are reserved.
7. The ORB algorithm based image registration method of claim 6 wherein deleting the matched pairs of feature points further comprises: And adopting a grid-based motion statistical algorithm to reject mismatching points in the matching point pairs.
8. The ORB algorithm based image registration method of claim 2 wherein the predetermined sector is a 60 arc area within the circular area, the predetermined step size is 10, and the second feature descriptor is 72 bits in total with the third feature descriptor.
9. An ORB algorithm based image registration apparatus comprising: The image acquisition module is used for acquiring a reference image and an image to be registered, wherein the reference image is an image with an overlapping area with the image to be registered; the feature point extraction module is used for respectively extracting feature points of the reference image and the image to be registered by adopting an ORB algorithm; The edge extraction module is used for extracting edge points in the feature point neighborhood by adopting a Canny operator and constructing feature descriptors of feature points according to the edge points in the feature point neighborhood; The matching module is used for matching the characteristic points of the reference image with the characteristic points of the image to be registered through the characteristic descriptors and determining matching point pairs; And the transformation matrix calculation module is used for calculating a transformation matrix between the reference image and the image to be registered through the matching point pairs, and registering the image to be registered based on the transformation matrix.
10. An electronic device comprising a processor and a memory, the memory having stored therein one or more computer programs, the one or more computer programs comprising instructions, which when executed by the electronic device, cause the electronic device to perform the ORB algorithm based image registration method of any of claims 1-8.

Description

Image registration method and device based on ORB algorithm and electronic equipment Technical Field The present application relates to the field of image processing, and in particular, to an ORB algorithm-based image registration method, apparatus, and electronic device. Background In the existing image registration technology, the ORB algorithm can extract image characteristic information more perfectly, and is a mainstream characteristic point matching scheme. However, the RBRIEF descriptor of the ORB algorithm generates binary codes by randomly selecting neighborhood pixel pairs, and does not consider feature point distribution and edge geometric information in feature point neighborhood, so that in similar texture areas, subtle differences are difficult to distinguish, and ambiguous matches are easy to generate. Disclosure of Invention The application provides an ORB algorithm-based image registration method, an ORB algorithm-based image registration device and electronic equipment, which can improve the accuracy of image registration. In a first aspect, the present application provides an image registration method based on an ORB algorithm, including: acquiring a reference image and an image to be registered, wherein the reference image is an image with an overlapping area with the image to be registered; Respectively extracting feature points of the reference image and the image to be registered by adopting an ORB algorithm; Extracting edge points in the feature point neighborhood by adopting a Canny operator, and constructing feature descriptors of feature points according to the edge points in the feature point neighborhood; matching the feature points of the reference image with the feature points of the image to be registered through the feature descriptors, and determining matching point pairs; and calculating a transformation matrix between the reference image and the image to be registered through the matching point pairs, and registering the image to be registered based on the transformation matrix. According to the ORB algorithm-based image registration method in the embodiment, feature extraction is performed in a feature extraction layer by adopting a mode of combining an ORB algorithm and a Canny operator, and compared with the existing ORB method, edge point information in the field of feature points is increased, so that interference of noise on edge detection can be effectively suppressed. In the aspect of feature description, the improved feature descriptor is adopted to fuse the feature point distribution and the edge point distribution in the feature point field, so that the feature expression of the gray level contrast area is balanced, the distinguishing capability of similar textures and repeated structures is enhanced, and the probability of feature ambiguity matching can be reduced. In a second aspect, the present application provides an image registration apparatus based on an ORB algorithm, comprising: The image acquisition module is used for acquiring a reference image and an image to be registered, wherein the reference image is an image with an overlapping area with the image to be registered; the feature point extraction module is used for respectively extracting feature points of the reference image and the image to be registered by adopting an ORB algorithm; The edge extraction module is used for extracting edge points in the feature point neighborhood by adopting a Canny operator and constructing feature descriptors of feature points according to the edge points in the feature point neighborhood; The matching module is used for matching the characteristic points of the reference image with the characteristic points of the image to be registered through the characteristic descriptors and determining matching point pairs; And the transformation matrix calculation module is used for calculating a transformation matrix between the reference image and the image to be registered through the matching point pairs, and registering the image to be registered based on the transformation matrix. In a third aspect, the present application provides an electronic device comprising a memory, one or more processors. Wherein the memory has stored therein one or more computer programs, the computer programs comprising instructions, which when executed by the processor, cause the electronic device to perform the ORB algorithm based image registration method as in the first aspect. In a fourth aspect, the application provides a computer readable storage medium having instructions stored therein which, when executed on an electronic device, cause the electronic device to perform the ORB algorithm based image registration method as in the first aspect. In a fifth aspect, the application provides a computer program product which, when run on an electronic device, causes the electronic device to perform the ORB algorithm based image registration method as described in the first aspect. It may