Rapid image matching method based on circular projection

Abstract

The invention discloses a rapid image matching method based on circular projection. The method comprises steps of firstly, utilizing isotropy and projection characteristics of a circle; converting thetwo-dimensional grayscale image into a one-dimensional rotation invariance representation form, calculating a subgraph feature vector by using a mixed circle projection combining original mean projection and variance projection, and calculating a matching similarity for the mixed circle projection vector calculated by the template and the search subgraph by using a normalized correlation coefficient; secondly, accelerating by using an image pyramid technology, and determining a target position by using top layer local clustering and layer-by-layer candidate point screening; and finally, for the obtained matching target, calculating the rotation angle of the matching target by using annular movement and angle histogram estimation, and completing the matching of the target. The rotation estimation of the method is superior to that of other similar methods, and an accurate and robust result can be obtained in rotation, offset and illumination change scene images. Various experimental results show that the method is suitable for image matching in industrial scenes.

Description

technical field [0001] The invention belongs to the technical field of computer vision and is used for matching objects in images, in particular to a fast image matching method based on circular projection. Background technique [0002] Template matching is the process of finding the best matching position in the image to be matched based on the known template. It is an important content in digital image processing and has been widely used in industrial alignment, target detection and recognition, tracking, etc. In recent years, there have been some effective algorithms in template matching research (such as literature [1]). However, most of the existing template matching processes convolve the template and the scene image, and calculate the similarity between the template and the scene image to determine the position (such as literature [2]). Due to the large amount of calculation of correlation, low-cost correlation algorithms are needed for real-time processing, and a la...

AI Technical Summary

Problems solved by technology

[0021] The problem to be solved by the present invention is: the traditional image matching algorithm based on grayscale has a large amount of calculation and is not resistant to rotation; the existing matching method based on circular projection is resistant to rotation, but the calculation amount is large, the matching accuracy is not high, and it cannot be used for homogeneous regions. Recognition, in the field of machine vision, it is necessary to be compatible with real-time and efficiency, and at the same time, it is necessary to recognize rotating targets

Images