| Surgical navigation system which applies the technology of image processingand three dimensional positioning enables a surgeon to look through the anatomy ofthe patient, to see the instrument guided into the anatomy and to plan the path of theinstrument, so that the safety and the accuracy of operation is improved. Opticalpositioning subsystem is a key part of the surgical navigation system it is directlyrelated to the accuracy of the surgical navigation system and the success of the surgery.The positioning system mainly uses the sensors of spatial location to determine theanatomy of the surgical area or the spatial coordinates of the external markers and totrack the instruments in real-time. Therefore, the relative technology of thepositioning subsystem has become one of the most cutting-edge research areas.For the active optical positioning system, its positioning accuracy is high and itcan track the object steadily, but the cost of it is very expensive. The purpose of thethesis is to develop a optical positioning system of low-cost and high-precision basedon binocular vision. According to these, the key technical details of the opticalpositioning system and some key issues related are deeply studied in this thesis viaexperiments. Preliminary results have been obtained.A low-cost, three-dimensional positioning system has been built based onbinocular vision. The system uses the visible light LED as the markers of the surgicalinstrument and two CCD cameras to build the binocular vision system. According tothe principle of the camera calibration, the matlab calibration toolbox and opencvlibrary functions have been used in the program to finish the task of calibration.A weighted method of calculating the center of mass based on gray valuedistribution of the image is proposed. As the lights emitted by the markers of thesurgical instrument is uneven. The gray level distribution around the marked points inthe images is uneven, and because of these, error has occurred while we arecalculating the center of mass of the object area. In order to solve the problem, we usea weighted method which based on gray value distribution of the image to calculatethe he center of mass of the object area. The result of the experiment shows that themethod is robust.The improved stereo matching algorithm based on the spatial order of the featurepoints has been realized. According to characteristics of the optical positioning systemof surgical navigation and the constraint of the order of the space, the algorithmmatches the feature points in two images. Results of the experiments show that the algorithm can match the feature points of the two images correctly and the matchingresults are not disturbed by the position and the direction of the surgical instrument.Based on the research of the key technologies, the thesis chooses the coordinatesystem of the left camera as the world coordinate system, calculates the projectionmatrixes of the left and right camera, and then reconstructs the feature points andanalysis for the error of the result at last. By doing Argumentative experiments on theconstructed prototype system of surgical navigation Optical Positioning System. Theresults show that the system is built of low-cost and it can run reliably and itsaccuracy of positioning can satisfied the basic requirements of the surgical navigationOptical Positioning System. The completion of the project provides favorable help tolet the project be successfully applied in medical image navigation in the future. |
PDF Full Text Request