Medical Limbs And Trunk Location Approach Research Based On Kinect

In clinical medicine,head positioning is particularly important.For the problem of inaccurate positioning,poor real-time performance and low resolution ratio in head positioning,this paper carries out the research from the aspect of face detection,face feature extraction and rotation angle of the head,we achieve a new head positioning method to achieve real-time and accurate positioning of the head.First,we analyze the hardware structure and working principle of the Kinect sensor which we use in the positioning process.Compared with the current head location algorithm,we select a fast and accurate head positioning algorithms,complete the experiment,and observe the effect of head positioning.Second,for the 2D image when we use the Ada Boost algorithm for face detection,the problems that one face is detected many times or the face is wrongly detected will occur.In order to solve these problems,this paper uses the color and depth image obtained by Kinect sensor to achieve a new face detection method which combines depth data with Ada Boost algorithm.The Haar-like feature which is used to construct the cascade connection splitter of the depth image is extracted during the training phase,and the depth information is used to filter the sub-windows during the detection phase of the algorithm.The experiments result show that,when the test sampling number is 600,the success rate of proposed face detection algorithm can achieve 98.2%.Finally,AAM algorithm is used to extract features of eyes,mouth,nose tip,etc.,and the qualitative and quantitative experiment schemes for head positioning are designed by using the extracted feature points.In the design of qualitative experiment,using left and right eyes,left and right lips middle point to construct to realize the head positioning of this method and verify the accuracy of this method reaching an average of 97.2%.In the design of quantitative experiment,using the left and right eyes,left and right lips,nose points to complete the calculation of therotation angle of the head,and using the public head posture database Bi Wi Kinect Head Pose to verify the method of this experiment to verify the method,and the method of verifying this experiment has a certain accuracy.In order to show the advantage of the proposed head positioning method,we compare the proposed method with the current existed method.The experiment results show that the proposed method has higher resolution.