Research And Implementation Of Image Recognition System For Lower Limb Rehabilitation Robot

Traditional human-computer interaction methods such as mouse,keyboard,and screen touch have been widely used in rehabilitation medical systems,which have problems of numerous keys and complicated operations.Therefore,in order to make the rehabilitation system more user-friendly and easy to operate,it is extremely urgent to find a more harmonious and natural new human-computer interaction method.Based on the rehabilitation training function of the YSU-I sitting and lying multi-joint lower limb rehabilitation robot,this article has developed the YSU-II intelligent rehabilitation system for the lower limb rehabilitation robot,which is a new type of human-machine interaction application example.The system uses a paper form as an input method and sets up rehabilitation training tasks by filling in multiple control command parameters.In addition,the system also includes voice recognition interactive functions to improve the practicality of the system.This topic takes the YSU-II intelligent interactive system as the research and development goal,in which the key technology of image recognition involved in this system is studied in depth.First,based on the background of intelligent rehabilitation robot and human-computer interaction,this paper analyzes the deficiencies of YSU-I lower limb rehabilitation robot in terms of human-computer interaction,and proposes a new dual-modal interaction mode that combines image recognition technology and speech technology.Secondly,in the analysis of common methods of image preprocessing and image feature extraction,based on the functional requirements of image correction,table contour detection and text image information recognition,the processing effects of their algorithms are compared,and the Hough suitable for this system is selected Transformation method image correction and table contour detection algorithm based on morphology.Third,the investigation of handwritten Chinese and handwritten digit recognition methods currently used in text information images on the market,because the recognition methods have different recognition accuracy for different types of target information,combined with optimized convolutional neural network algorithm and optical character recognition technology Build text information recognition engine.Finally,the functional structure of the YSU-II intelligent interactive system and the image recognition system is designed,and the two types of instruction input modes,form input and voice interaction,are compared and experimented to show the motion control effect of the interactive interface of the image recognition system and the rehabilitation robot task training mode.