Reserach Of Scoliosis Based On Kinect Camera

In recent years,with the increasing incidence of scoliosis in adolescents,scoliosis detection equipment has become an important branch of medical detection and has been greatly developed.The detection of scoliosis,especially the detection of mild scoliosis,is an important part of the detection of spinal lesions.However,at present,the majority of hospitals in China have scoliosis tests using X-ray photographs through specialized medical staff.This method is inefficient,expensive and harmful to the body.Equipment developers in foreign countries have applied image recognition and machine vision technology to the detection of scoliosis.The technology concentrates on the knowledge of optics,mechanics,algorithms,etc.,and has great research value,However,the spine detection technology based on Kinect camera has not been reported yet.This thesis mainly studies the overall design of the scoliosis detection system based on Kinect camera,and extracts and processes the back image data of the human body.The main research contents of this thesis are as follows:Firstly,based on the main characteristics of the detection system,referring to the current identification method of scoliosis in medical detection,designed the modules of the scoliosis detection system,which are roughly divided into hardware and software modules.The software part includes the scoliosis detection module,the camera photographing module,the motion control module,the hardware part includes the camera selection,the photographing environment,the platform motion control module,and finally,the two modules are joint debugging to complete the design of the entire system.Secondly,this thesis studies various aspects of human body back image processing,including high-precision depth image extraction algorithm,human back feature point extraction algorithm,spine midline extraction algorithm and Cobb angle detection algorithm.The high-precision depth image extraction mainly uses the quickhull convex hull algorithm,Delaunay triangulation and linear triangular interpolation processing algorithm to extract high-precision data,and the human back feature extraction is processed by traditional image processing algorithms.Including image gray-scale algorithm,ROI extraction,depth-first search and contour search algorithm.The spine midline extraction algorithm and the spine midline fitting algorithm are discussed in the spine line extraction method,and the Cobb angle calculation method is discussed.Finally,we use the Windows system-based VS2012 tool for software development,and use C++,OpenCV,PCL point cloud library to implement software implementation of each part.According to the functional requirements of the scoliosis detection project,the scoliosis detection system with image acquisition,depth image processing,color image processing,motion control and other modules is completed..According to the test results of this system,the scoliosis detection system designed in this thesis can detect the scoliosis of people with Cobb angle between 0 and 20 degrees within 3s,and mark people's back feature points,The final result error is within a reasonable range(±1°~ ±4°).This scheme can meet the needs of medical testing,and has a certain practical significance.