Research On Vision Based Real-time Dynamic Gesture Segmentation Method

In recent years,with the rapid development of artificial intelligence,virtual reality and other intelligent technologies,gesture-based gesture recognition has become an important part of human-computer interaction and research hotspot due to its simple,natural,intuitive and non-invasive features.Gesture segmentation as the initial part of the gesture recognition system is the basis and important part of gesture recognition.The accuracy and speed of gesture image segmentation directly determines the recognition effect of gesture recognition system and the performance of the whole interactive system.Therefore,the research of gesture segmentation method has important theoretical significance and practical application value.This article revolves around related techniques and key content such as vision-based hand gesture image preprocessing,color space,skin color model,clustering algorithm,and target tracking algorithm,aiming at color distortion of gesture images caused by changes in lighting conditions and interference of facial information in gesture images.Problems such as white balance color correction and face information removal are separately processed.Based on the static segmentation of gesture images,a real-time dynamic gesture segmentation method for video sequences is developed by combining theoretical derivation and experimental verification.The specific research content is as follows:1.The white balance algorithm is used to correct the color of the gesture image,and the facial area is determined and removed.The quality of the gesture image is improved effectively.Through the white balance color correction,the skin color information of the gesture and image is closer to the real skin color,which provides favorable conditions for thresholding segmentation based on skin color.In view of facial information interference,Haar-like features are used to recognize facial regions and remove facial regions,thus effectively reducing the interference of facial regions on gesture segmentation.Experimental results show that the accuracy and accuracy of gesture segmentation can be effectivelyimproved by image preprocessing.2.A static gesture segmentation method based on a mixture of Gaussian skin color model and K-means clustering algorithm is proposed.Based on the analysis of commonly used color space,a Gaussian skin color model was constructed.The similarity between pixels and skin color was calculated using a two-dimensional Gaussian distribution function to obtain the probability that any pixel in the image belongs to the skin color,and then the likelihood map of the gesture was obtained.Combining YCbCr color space chromaticity and brightness separation has the characteristics of significant clustering and K-means clustering efficient segmentation to solve the problem of inaccurate segmentation of skin color segmentation under illumination changes.The experimental results show that this method can effectively avoid the influence of factors such as illumination changes and improve the accuracy and adaptability of static gesture segmentation under weak light.3.A dynamic gesture segmentation method based on elliptical skin color model and three-frame difference algorithm is proposed.Firstly,a three-frame difference algorithm is used to segment the motion region,and the gesture region is tentatively determined.Then,the initially confirmed gesture region is secondarily segmented using the skin color segmentation of the elliptical boundary model to remove the non-skin color target region and processed by the grayscale normalization.For the second segmentation,bitwise AND operation and morphological processing are used to extract the gesture target region.At the same time,improvements and solutions have been proposed for the problems of large-area color patch area interference,removal of non-target skin color areas,and gesture loss due to brief pauses.The experimental results show that the improved algorithm can effectively solve the problem of large area skin color background,facial and other non-target skin color regions being misdiagnosed and illumination changes to the interference of gestures.It is suitable for real-time segmentation of dynamic gestures under visual conditions.