Hand Gesture Recognizing Tracking System Based On Camshift

With the rapid development of computer technology, communication betweenpeople and computers is much more diversified. Human-computer interactions are alsointroduced continuously into new areas. Due to the flexibility and complexity ofgestures, the gesture recognition technology, based on vision, becomes the majorconcern of studies. It can be applied in many fields like machine vision, patternrecognition, augmented reality, etc.This paper presents an effective method of gesture segmentation and tracking,which can split two hands when they appear simultaneously. With both of handswearing different colors of the hand gloves，the method converts the color space ofhand images from RGB to HSV. In the HSV color space, hands images are split usingsemi-automatic predefined templates color. By using a method of combination ofgesture tracking and gesture recognition technology, Camshift is improved to solve theproblem of semi-automation and to realize multi-targets tracking. After successful splitand tracking, a set of gestures profiles is established, which enable the process ofreal-time gesture recognition.On VC platform the software is implemented, which achieves the followingfunctions: It captures video images from USB camera, transmits them to computers,and then processes them to recognize the hands, and finally achieves the interactionbetween human and computers.After series of test, the algorithm can minimize the interference while usingCamshift to track gestures. It can also track and recognize both hands. Both theefficiency of tracking and the recognition accuracy are high. And they are affectedscarcely by light.