Method for detection and optimisation of vein boundaries on the backof the hand based on U-Net and Canny algorithms

Hand dorsal vein boundary detection is an important research direction in the field of smart healthcare. This paper proposes a hand dorsal vein boundary detection algorithm that integrates deep learning and Canny edge detection, uses the U-Net model for global semantic segmentation of hand dorsal veins, and optimises the accuracy of the boundary by combining with the classical edge detection algorithm. Experiments show that the method exhibits good robustness in complex backgrounds, and the mIoU (Mean Intersection over Union) of boundary detection reaches 80.5%. In addition, the innovative evaluation index for vein edge detection proposed in this paper, the 𝐵𝑜𝑢𝑛𝑑𝑎𝑟𝑦𝑚𝐼𝑜𝑈, focuses on the overlapping region between the real boundary and the predicted boundary, effectively avoiding the interference of the traditional IoU evaluation on the irrelevant region, and providing a more accurate performance evaluation method for vein boundary detection.