Gesture Recognition System Based On FPGA

With the development of technology,human-computer interaction methods have become more friendly and natural.Smart home has also moved from concept to real life,but traditional control methods are still used in the field of smart home control.The control method based on naked hand gesture recognition is convenient,fast,and does not require the user to touch any control device.It is of great research significance to replace the existing control methods of home appliances.The gesture recognition system is designed,which uses field programmable gate array(FPGA)as the algorithm and control core.Realized real-time gesture recognition and wireless manipulation of peripheral devices.The algorithm design includes the image preprocessing stage to perform median filtering on the image.Then perform gesture segmentation and denoise the segmented gesture image through a morphological filtering algorithm.Features such as area perimeter ratio,effective area ratio,and fifth-order Hu moment are extracted by a feature extraction algorithm.Group according to the palm shape of the experimenter and train the corresponding sample library of each group.The hand-type adaptive algorithm is used to pair the user's hand shape with a suitable sample library.Then the effective area ratio feature is used for preliminary recognition of gestures.Finally,the final gesture is recognized using the European distance template matching method.Each module inside FPGA was Designed through Verilog HDL.Configure the camera through I2 C bus to output RGB format video stream data.Further processing is performed on the video stream data output by the camera through the data acquisition and decoding module.Obtain the video stream data that can be processed by subsequent algorithms,and then write the video stream data to the off-chip memory SDRAM through FIFO.Process video stream data through image pre-processing module,gesture segmentation module,gesture feature extraction module,hand-type adaptive module,and gesture recognition module to obtain gesture recognition results.Through experiments and tests,the designed gesture recognition system can recognize 9 kinds of gestures.The recognition rate can reach 95.31%.It can meet the requirements of control accuracy and real-time.