Registration of Lithium Battery x-ray Images Based on an Improved RANSAC Algorithm

Abstract

Image registration and stitching are required in the defect detection of lithium batteries. However, existing image registration methods will suffer from a large number of mismatches caused by similar textures and structures in lithium battery x-ray images. In order to address the problem, we propose an improved RANSAC (random sample consensus, RANSAC) algorithm, which is optimized based on the structure characteristics of lithium battery. When solving the relationship model, this method calculates the matching quality of the matching pair based on the longitudinal pixel distance of the matching pair, and then eliminates the wrong matching point pairs according to the matching quality of the matching pair, thereby reducing the number of mismatched pairs. Experiments show that the algorithm proposed in this paper can eliminate obvious mismatched pairs, and the registration accuracy of lithium battery X-ray digital images is improved by 20.5% on average.