Cell Image Stitching Algorithm Based on Retinex And Improved ORB
Nanyan Liu, Zhongpin Li · 2021
Due to the limited shooting range of microscopic imaging equipment, as well as the complicated factors such as the shooting environment and the cell tissue itself, the microscopic cell image is not clear in detail and the brightness is not equal. An algorithm based on Retinex and improved ORB for microscopic cell image registration and stitching is proposed. Firstly, the multi-scale Retinex algorithm is improved by homomorphic high-low-pass filtering to enhance the image, and the image feature extraction area is specified to improve the efficiency of algorithm execution. Then the improved ORB algorithm can be used to extract feature points, which can reduce feature point redundancy while reducing feature extraction time. Finally, the Laplacian multi-resolution fusion algorithm based on the optimal seam algorithm was used to eliminate the large color difference and ghosting on both sides of the stitch. The experimental results show that the algorithm has achieved better results in the quality and clarity of microscopic cell image stitching, and the efficiency of the algorithm has been improved.