Application of Deep Learning-Based Chessboard Marker Recognition Algorithm in Surgical Positioning

Abstract

The chessboard grid, as common positioning markers in computer-assisted navigation systems, is crucial for the speed and accuracy of the navigation system. In computer vision, people typically use a chessboard detection function OpenCV: findChessboardCorners to detect the corners of the chessboard grid for camera calibration and pose estimation. However, the applicability of this function is limited, as it may not return accurate results in cases of occlusion, low image quality, small chessboard grid areas, or multiple targets. Additionally, as image resolution increases, the algorithm's processing time significantly increases. These limitations make it challenging to meet the requirements of surgical positioning. To address issues such as small chessboard markers, insufficient clarity, and the presence of multiple chessboard grids simultaneously in surgical scenarios, this paper proposes a two-stage chessboard detection algorithm based on deep learning. This algorithm first segments the approximate location of chessboard grids and then identifies the positions of the grid points. Compared to traditional chessboard detection algorithms, this approach significantly improves the accuracy and efficiency of chessboard detection.