Research On Real-time Palm Localization And Hand Gesture Recognition Based On Monocular Vision

Hand gesture recognition is one of the most popular researches in the field ofhuman-computer interaction, and it performs great efforts in constructing intelligentinteraction. However, gesture has the characters of the variability, diversity, ambiguityand spatiotemporal differences. Moreover, the recognition results are easily affected bylight conditions, temperature and other disturbances. Therefore, there are still manydifficulties in gesture recognition based on computer vision. Aiming at these problems,a real-time palm localization and hand gesture recognition method based on monocularvision is presented in this paper which can overcome the shortage that the traditionalhand gesture recognition is easily affected by illumination and temperature changes, andthe stability, practicability and accuracy of gesture recognition system are enhanced.The main work is as follows:Firstly, the research background and significance of this paper are introduced, andthe present research states of hand gesture recognition are summarized. Aiming at theexisting problems of gesture recognition at present, the solution of real-time palmlocalization and hand gesture recognition based on monocular camera is proposed.Secondly, denoising processing and mathematical morphology algorithms arepresented, and the common methods of image contour tracking detection, motiongesture detection and skin color detection and segmentation are summarized andanalyzed, the corresponding experiments are also carried out so that gesture contourtracking detection and segmentation are realized.Thirdly, the basic contents of the convex hull algorithm are introduced, and theconvex hull detection experiment and the convexity defect detection experiment of handgesture are also carried out. On this basis, palm center localization based on pointincrement method is proposed and the depth point which comes from convexity defectdetection experiment is served as the point set in point incremental method, then thepalm center localization experiment is also carried out so that the center position ofpalm is determined accurately and effectively.Finally, the framework design, hardware and software environment and runningprocess of the system are introduced, and a real-time palm localization and hand gesturerecognition method based on monocular camera is also presented, then the experimentaldesign schemes of the system are described, and at the same time, different experimentsare designed to verify the stability, practicality and feasibility of the proposedalgorithms in this paper.