Design And Implementation Of Intelligent Acupoint Positioning Moxibustion System

Moxibustion is one of the external treatment methods of traditional Chinese medicine(TCM)in our country.It uses moxa sticks to burn at the acupoints of the human body to generate warm stimulation to treat diseases.The warming and tonic effect of moxibustion plays an important role in promoting the recovery of chronic diseases.In traditional Chinese medicine,acupoints are the stimulation points and reaction points of moxibustion to treat diseases.According to the distribution of the meridians in various parts of the human body,the accuracy of acupoint positioning is the key to the effectiveness.The clinical realization of acupoint selection in traditional Chinese medicine requires long-term study and experience accumulation of the physician.The grasp of the distance depends on the subjective perception of the doctor,and finally the location of the acupoints in the target area is explored and determined with the cooperation of the hand.This research focuses on the key engineering and technical issues of the intelligent moxibustion system,and aims to realize autonomous moxibustion.Taking acupoint positioning technology as the starting point,and comprehensively using multiple technologies such as robotic arm control,machine vision,and machine learning,the intelligent acupoint positioning moxibustion therapy system is designed.The research work of automatic acupoint positioning method based on deep learning,control of moxibustion robot arm,and software design of upper computer are completed.Finally,the automatic identification of acupoints is realized,and the extremity of the moxibustion is driven by the robot arm to move to the position of the acupoints and automatically realize gentle moxibustion therapy.The main work of this thesis includes:(1)According to the specific realization process of moxibustion,combined with the theory of TCM meridians,an intelligent acupoint positioning moxibustion therapy system is designed with the moxibustion mechanical arm as the actuator.Firstly,the construction of the coordinate system of acupoints and the physical characteristics of moxibustion are analyzed.Based on this,the requirements of various functions of the system are analyzed.According to the actual requirements,the system hardware and communication scheme are selected and the system platform is constructed.(2)Aiming at the problem of autonomous identification of human acupoints,a real-time acupoint localization method based on monocular camera is designed.Through data analysis,the arm is selected as the bone-length measurement.For the problem of the deflection direction of the limb,the front and background segmentation are performed based on the color space conversion to obtain the deflection angle of the limb.For the positioning of the wrist line and the cubital crease,the hand width feature is introduced,and the position of the wrist line and cubital crease is obtained by combining the corner point detection.A lightweight human key point detection network is used to obtain reference acupoints,the coordinates of the acupoints with partial calendar are calculated combined with the pixel length and deflection direction of cun.Then the accurate and real-time positioning of acupoints can be realized finally.(3)Aiming at the realization of moxibustion techniques,a moxibustion manipulator is designed as the actuator.The main controller,distance sensor and other hardware of the moxibustion manipulator are selected and the circuit design of the manipulator system is realized.The kinematics analysis of the moxibustion manipulator is realized by the D-H parameter method.The vision system with "eyes out" and the nine-point calibration method are used to realize the transformation from the image coordinates to the coordinates of the working plane of the manipulator,so as to complete the hand-eye calibration of the manipulator.(4)The host computer interface is designed based on PyQt5 and realizes the corresponding functions.The MySQL database is used to store the system information.Based on needs,the physical parameters of the moxibustion process are set to control the robotic arm through the interface.Finally,the overall test of the system is carried out,and the acupoint positioning experiment and the acupoint tracking experiment of the robotic arm are carried out respectively.The experimental results show that the acupoint positioning method has high accuracy and real-time performance,and the stability of the robotic arm control meets the requirements of the system.