Research on Machine Vision Image Mosaic Algorithm for Multi-workpiece Cutting Platform

The cutter head parts of the tunnel boring machine (TBM) are of different sizes, and the cutting platform is of large size. The panoramic view of the cutting platform that meets the accuracy requirements through image stitching is a prerequisite for realizing the visual guidance of autonomous cutting. An image stitching algorithm is studied to reduce the complexity of the algorithm on the basis of ensuring the stitching precision. The ORB algorithm is selected to implement feature detection, and a series of strategies are adopted on this basis. The ROI region is extracted, and only the images in the region of interest are matched. Then, the feature detection, description and matching are performed by the ORB feature detection algorithm and Hamming distance. The feature points are purified by the RANSAC algorithm, and the transformation matrix H is calculated. Finally, the image fusion is carried out. The experimental results show that the ROI-based ORB image stitching algorithm satisfies the accuracy requirements of subsequent visual guidance.