Three-dimensional real-time positioning compensation method for surgery

A three-dimensional real-time positioning compensation method for surgery, comprising:

• obtaining a surgical image, wherein the surgical image comprises a plurality of marking devices, the marking device has a regular polyhedron, the regular polyhedron has at least four geometric faces, the geometric face has a mark, the mark comprises a frame and a graphic, and the graphic is located inside the frame and is used for identification to obtain a unique identification code;
• inputting the surgical image into an object detection model to detect first bounding box information of each of the plurality of marking devices, second bounding box information of the frame of each marking device, and an identification code and third bounding box information represented by the graphic of each marking device;
• obtaining four corner points of the corresponding frame according to the second bounding box information, and performing a first projection transformation calculation according to the first bounding box information of each of the plurality of marking devices, the four corner points of the frame and corresponding mechanical corner points, to obtain an initial posture of each of the plurality of marking devices in a world coordinate system;
• performing brightness normalization on the surgical image to generate a normalized image;
• respectively generating a plurality of texture images having equal ratio of length to width but having different resolution for all marks of the plurality of marking devices;
• respectively performing a fuzzy logic operation on the different resolutions of the plurality of texture images with the same identification code according to the identification code represented by the graphic of each marking device, and using the texture image having the closest resolution as a standard image;
• selecting at least N sampling points other than the four corner points from a frame or a graphic of the standard image according to the second bounding box information and the third bounding box information, performing a second projection transformation calculation according to the at least N sampling points and corresponding mechanical corner points, and performing a calculation based on a camera parameter, to obtain a reference point corresponding to each of the at least N sampling points in the normalized image, wherein N is a positive integer;
• obtaining an optimization point within an allowable range by using the reference point as an origin, wherein there is a minimum brightness error between the optimization point and the reference point;
• calculating an error between each reference point and the corresponding optimization point based on a minimization error function, and correcting the initial posture according to the error to generate a compensation posture;
• and
• respectively drawing and displaying a compensation posture of each of the plurality of marking devices in the world coordinate system on a display screen in real time.