Multi-step image registration algorithm based on affine transformation and template matching

Abstract

The invention discloses a multi-step image registration algorithm based on affine transformation and template matching. The algorithm comprises the following steps: obtaining a schematic diagram and a real shot image of an object; extracting a detection target area of the real-shot image, and completing first-step coarse registration of the real-shot image and the schematic diagram by using affine transformation according to the outer contour features of the target area and the schematic diagram; registration of the registration area is completed; extracting a registration point position, and matching the registration point position with the schematic diagram; according to the corresponding relation of the matching points, obtaining a transformation matrix of affine transformation of the real shot image and the schematic diagram, and completing affine transformation; once template matching is carried out around the corresponding position of the real shot image, compensation of the area registration error is completed, and finally the registration work of each area is completed. The invention belongs to the technical field of machine vision defect detection, and particularly provides a multi-step image registration algorithm based on affine transformation and template matching, which can allow a to-be-detected product to have a larger error in an image and has higher image registration precision.

Description

technical field [0001] The invention belongs to the technical field of machine vision defect detection, and specifically refers to a multi-step image registration algorithm based on affine transformation and template matching. Background technique [0002] With the improvement of living standards and the development of manufacturing technology, people have put forward higher requirements for product quality. 3C parts, processed parts, textiles, pharmaceutical packaging, food packaging and other industries all have product appearance quality inspection problems. Usually people pay attention to the quality of product appearance, mainly including: (1) products whose value depends on the quality of product appearance, such as printing, packaging, handicrafts, etc.; (2) products whose surface defects directly affect the use of products and deep processing, will It will bring great losses to users and deep processing customers. Therefore, it is very necessary to control the qualit...

AI Technical Summary

Problems solved by technology

[0004] For the above methods, the advantage of the algorithm (1) geometric registration method is that the processing speed is fast, and only one registration area is needed, but the disadvantage is also obvious, that is, the registration accuracy is low; the advantage of the algorithm (2) two-point registration method is that The processing speed is fast and the positioning accuracy is high, but the disadvantage is that two registration points are required, and the requirements for registration points are high