Research On Positioning Method Of Assisted Puncture Robot Based On Binocular Vision

The localization of puncture robot is one of the key technologies of robot-assisted puncture surgery.At present,the surgical navigation and positioning of the puncture robot is mainly based on optical positioning technology.Most of the most common optical positioning devices rely on visible light.The extraction of feature points is easy to be interfered by ambient light,and the amount of extraction is large,and the algorithm is complex,which is not conducive to accurate positioning in the puncture surgery.Therefore,this paper focuses on the near-infrared optical positioning method,and the main research work is as follows:1.The near-infrared optical positioning system is built independently according to the existing conditions of the laboratory.The near-infrared binocular camera is composed of near-infrared filters installed outside the ZED Mini binocular camera lens.By adding near-infrared filter,visible light can be effectively filtered and system stability can be increased.2.Design and manufacture a near infrared calibration template.In this paper,a near infrared calibration board composed of near infrared patch light emitting tubes is designed.For obtaining the geometric information of the calibration board,first,the pixel coordinates of the marked points on the calibration board are extracted by human-computer interaction method combined with grayscale center of gravity method,and then the three-dimensional coordinates are obtained by a visible light binocular positioning system.The final near infrared calibration board is obtained.The experimental verification shows that the error of the calibration board is 0.045 mm.3.The corresponding calibration method is designed based on the near-infrared calibration board.Under the visible light condition,proposed traditional methods combined with neural network-based calibration algorithms and optimized improvements.The inertia variable weight is introduced to improve the performance of the algorithm.The experimental results show that the calibration average re-projection error is 0.09 pixels;Secondly,based on the distortion model of the near-infrared camera,the initial parameter values are solved by direct linear transformation,and the final parameters are obtained by nonlinear optimization method.The average re-projection error is 0.7936 pixels,achieving accurate calibration of the near-infrared camera.4.Realize the identification and positioning of the marker points under the near-infrared binocular camera,identify the imaging characteristics of the near-infrared marker points,and propose a two-step binocular matching method to identify and match the probe of the simulated puncture robot and the near-infrared reflective ball on the simulated human head model,and conduct three-dimensional reconstruction.The experiment shows that the reconstruction measurement accuracy of this method reaches 0.79 mm,which can meet the requirements of surgical navigation and positioning.