Research Gesture Interaction And Virtual Reality Model Enhancements

Augmented Reality (AR) is a kind of technology which places virtual objects that computer generates to the real scene accurately and strengthens the visual experience of human by means of a video display apparatus, with the development of virtual reality technology.In recent years, it has become a hot research topic. The software system based on AR technology is widely used in various fields such as education and scientific research, urban planning, most of which are based on the interactive technology of artificial markers. Although this interaction is more mature, but there are still a lot of inadequacies. The gesture is a very intuitive way, if we use manpower instead of artificial markers as a computer input to complete human-computer interaction, so the applications of software system based on AR will be wider and the immersion of human-computer interaction will also be improved.In this paper, the technology how manpower interacts with virtual objects in software system based on AR is researched, including gesture segmentation, fingertip detection and tracking, interactive way and a system based on AR that manpower interacts with virtual objects will be achieved. The main work of this paper as follows:(1) For the problem that the area of analogous skin color is difficult to identify on gesture segmentation, a method based on the combination of skin color detection and background subtraction is put forward in this paper. First, an improved neighborhood algorithm is used to smooth target image; then the YCrCb color space is selected to establish the Gaussian model of skin color distribution and the Gaussian mixture model is established for the background of image after skin detection. After the logic operation of the two methods, we could obtain a single area of the gesture. It is verified that this method can meet the system’s real-time segmentation requirements.(2) Combined with the characteristics of the system, a method based on contour curvature is used for fingertip detection, curve fitting is used for finger area, and the precise position of each fingertip is found out by taking the ellipse fitting. When tracking the detected fingertip real-time, we propose a matching algorithm by calculating the displacement of fingertips in two adjacent image and combining with the movement of the hand centroid to track fingertip real-time. After experimental verification, the algorithm has robustness on the detection and identification as well as real-time tracking fingertips. (3) Designed and implemented a human-computer interaction system based AR. The early realization of the algorithms is integrated into this system, natural hand instead of the traditional paper marker to achieve real-time interaction with the virtual model. Through several tests, we can see that the system implemented in this article can adapt to different environments and interact naturally.