Abstract: Aim To propose a method of realizing hand tracking in virtual reality systems. Methods Based on computer vision techniques, reflective markers were placed on the moving hand to simplify the recognition calculation and improve the reliability of the device. Images captured by a CCD camera were analyzed by the R,G,B value of the pixels in real time. The positions of markers were calculated. The relationship of 3D position and 2D image of hand was analyzed and the positions of the palm and the fingers were automatically calculated. Results and Conclusion The method overcomes the drawbacks of traditional data gloves which are cumbersome and fragile. Ordinary visible light source is used instead of infrared light sources in other studies to reduce the cost of the device.