The Research Of Gesture Recognition Techonlogy Based On FPGA

The way of human-machine interaction has gradually tend to becoming a simple, humanistic, the technology of hand gesture recognition as an interactive mode has become a important research subject.Television is an essential household appliance in people’s lives, watching TV occupies an important position in all kinds of leisure activities. So,the topic of how to be more convenient and effective control of television is hotUsing freehand gestures to replace the traditional remote control to controlling the television has important research significance and application prospect. Using FPGA to process image has become a developing trend,the paper uses FPGA to realize the arithmetic of gesture gecognition in order to improving the speed of image processing and overall system performance.Firstly,the algorthm of gesture technology is studied in this paper,the process of identification including pretreatment, gesture segmentation,feature extraction and recognition.The paper studies the skin color model with good segmentation effect though the analysis of the segmentation algorithm,and it establishes the skin color threshold model which has good skin color clustering feature in YCbCr space.It further studies the characteristic parameters of the Hu moment and ratio of area and perimeter of the gesture outline.The paper studies the algorithm of template matching in matching recognition part.Secondly, the hardware circuit of the gesture recognition system is designed.In order to converting gesture image to digital signals the paper designs the video decoding circuit. Image storage circuit is designed ping-pong structure using two SRAM memories,and it realizes the seamless between data cache.VGA display circuit is builded by encoding chip,and it can display the image collected by CCD on VGA.Finally,the paper realizes the algortithm of gesture recognition by the FPGA. In reprocessing part, fast median filtering module is designed using pipelining structure.ROM look-up table is established to segment the hand gesture. The paper designs the minimum filter and maximum filter to realizing the operation of corrosion and expansion. Sobel edge detection module is designed to get the gestures contours.The paper improves the traditional feature extraction method which is seven moment invariants,it puts forward the ratio of area and perimeter of gesture outline and the first fout Hu moment as feature vectors and realizes by FPGA.lt has good recognition effect.rn the matching recognition part,it realizes the minimum Euclidean distance in the multidimensional space. This system successfully completed the gesture recognition based on FPGA and the recognition results are verified.