Target Following System Based On Embedded Mobile Vision

According to the statistics of our country's health organization,there are many deaf-mutes all over the country.Deaf-mutes can use sign language to communicate,but when they communicate with people who do not understand sign language,there is a lot of inconvenience.Based on the above problems,this paper proposes a glove-based gesture recognition system based on data glove.The inertial sensor and bending sensor were used to collect the gesture infor-mation of sign language gesture,the gesture recognition algorithm was used to recognize the sign language movement.The hardware part was composed of main control module,attitude module,power module and wireless module based on data glove of the sign language ges-ture recognition system.Among them,the attitude module was composed of a flexible bending sensor and an inertial sensor.The relationship between the re-sistance value and the curvature of the flexible bending sensor realized the detec-tion of the positional information of each finger joint,the inertial sensors to cap-ture spatial status and trajectory information during gesture movement process.The main control module transmited the acquired gesture posture information to the upper computer through the wireless module,finger position information and space gesture information of the entire gesture motion were intercepted according to a differential value method.Then,the number of extracted physical signs was based on the second knuckle defined by the joint characteristics of the fingers,metacarpophalangeal joint,and the three-axis acceleration and the three-axis magnetic angle in space attitude information.According to the difference in the feature quantity of each sign language action,the sign language action was iden-tified by a 4-layer decision tree classifier.Based on the above,developed the up-per computer of the sign language gesture recognition system using Processing Software Based on JAVA Language.The test platform was set up and 30 experimenters were selected to test the six kinds of defined sign language actions in this paper.The test results shows that the average recognition rate of sign language gesture recognition system reached 87%based on data glove,verify the effectiveness of the system,the recognition rate increased by 5.9%compared to a single inertial sensor sign lan-guage gesture recognition system.