The Research On The Positioning System Of Surgical Instruments Based On Binocular Vision

With the continuous developments of computer vision and stereotactic technology,surgical navigation systems have been widely used in neurosurgery,otorhinolaryngology,plastic surgery,and other medical fields.The surgical instrument positioning system is an essential part of the surgical navigation system.Its positioning accuracy and computing performance directly affect the use of the surgical navigation system.Therefore,research in this area is very active.The positioning methods of surgical instruments are mainly divided into four categories:mechanical positioning,ultrasonic positioning,electromagnetic positioning and optical positioning.Among them,the Optical positioning method based on binocular vision principle has become the most popular surgical instrument positioning system due to its comprehensive advantages in application scenarios,positioning accuracy,stability and cost.After analyzing the related researches of optical surgical instrument positioning systems in recent years and the popular optical surgical instrument positioning products,the following problems were found.In order to solve these problems,this thesis proposes related algorithms and implements an optical surgical instrument positioning system using three passive marker balls.The specific work is as follows:Aiming at the problem of low accuracy of marker coordinates extraction and positioning failure caused by occlusion,this thesis proposes a fast circle detection and center positioning algorithm based on chord,referred to as CDBC(Circle Detection and Center Location Based on chord).Different from most circle detection algorithms,this algorithm judges circle and the center based on the geometric property that the vertical line of the chord of the circle must pass through the center of the circle,rather than using gradient information,thus avoiding the problems that the gradient in the discrete image cannot be calculated accurately.The CDBC algorithm first extracts high-quality edges of the marker ball through image processing,and then calculates candidate circle centers based on geometric properties,and proposes a scale-adaptive convergence coefficient method to eliminate misdetection and correct errors.CDBC algorithm also divides the connected edges caused by the marker ball’s mutual occlusion based on the watershed segmentation algorithm.In this thesis,the synthetic image experiments and the real image experiments are carried out,which prove that the CDBC algorithm has stable circle detection ability and has better center positioning accuracy and calculation speed than the excellent circle detection algorithms in recent years.Aiming at the problem of inaccurate stereo matching of marker points and the restrictions on the geometric relationship between the markers,this thesis proposes a marker matching algorithm based on tracking and area matching,referred to as MTAM(Marker Tracking and Matching).MTAM uses the correlation filtering algorithm to track the marker ball,which can correctly correspond to the center coordinates of the circle.Then,MTAM adopts the matching area delineation method based on the disparity variance to narrow the search range of matching corresponding points and avoid false matches.Finally,MTAM matches the corresponding points based on the improved AD-Census matching cost calculation function.Experimental data shows that the MTAM algorithm can correctly correspond the center coordinates with the marker ball during the continuous positioning process,accurately match the marked corresponding points,and does not impose any restrictions on the geometric relationships of the marks.Finally,this thesis combines the theory and the two proposed algorithms to implement the SI-Tracker,an optical surgical instrument positioning system based on three passive marker balls,and sets up different experiments under various conditions to verify its positioning accuracy and operating speed.The experimental results show that SI-Tracker achieves the same positioning accuracy and operating speed as the international leading products at a very low cost,and meets the requirements of surgical navigation systems.