Hand Gesture Recognition Research In Modern Human-computer Interaction

Nowadays, the computer technology keep changing continuously, the way inwhich people interact with the computer has also been more easier, human-centered.Researchers has been trying to find an interacting method that can simply people’slife to the utmost. Hand recognition is one of the most popular human-computerinteracting methods. Hand is vivid, easy to understand, and also an important mediaused in people’s communication situation. The existing human-computerinteraction method is not humanized. When using a tool or a certain product, peopleoften apt to hope that the tool or product can simplify their life. Any tool producedfor people is aimed at serving for people,what is to say, it is human-centered. Ifpeople like to use it, it has its markets and its existing value. Similarly, thehuman-computer interaction method will also develop under the human-centeredidea. Hand gesture is one of the most common-used interaction methods in people’slife, compared with which the existing human-computer methods are really dull.Camera shooting-based hand gesture identification system is researched in thispaper. Firstly, Directshow is used to extract each frame from the videos and theframe differentiation method is used to detect moving hands. Together with Sobeltransformation, human color evaluation function is built to get the boundaryinformation. This way, the hand segmentation accuracy is enhanced. After that, theprojection method and mode match are combined to trace the hand, and the handtrajectory is stored. Finally, the rules defined originally are used to judge whetherthe trajectory is reasonable, additionally, the PPT software are adopted to show thehand identification results. For the use of human-color evaluation function, theinterference effect caused by the outside environment is decreased. Detecting andtracing moving hands before cameras is realized in the system, when the detectedtrace meets the defined trajectory features, the corresponding gesture is identified successfully.This paper combined the human-color evaluation function with the Sobeltransformation method, with this combination method, the image edges areextracted, the experimental results show that compared with just using the Sobeltransformation method,the detecting result of the method proposed in this paper ismore accurate.The programming work is done on the VC++6.0platform, hand locationcatching is realized and also the function of tracing moving hands. To test theexperimental effect, the method is used to realize turning PPT forward andbackward automatically. The experiment result shows that the success rate is ashigh as90%.