Research Of Opencv-based Gesture Recognition Technology

Gesture recognition of computer vision is a harmonious and natural human-computerinteraction, which has been an important research topic in human-computer interaction fieldand has a large mount of application. With the rapid development of computer science, imageprocessing technology and artificial intelligence etc. makes the gesture recognition developrapidly especially in the current rapid development of virtual reality technology. But thecharacteristic of the hand and the gesture often need to limit the environment for gesturerecognition, which makes it hard to be popular in people’s daily life. How to reduce therestrictive conditions for accurate recognition of gestures and ensure the accuracy, efficiencyand stability is a key for study. This paper mainly in the OpenCV database as the foundationto study the key technology of gesture recognition.Firstly, this paper reviews the development and research status of hand gesturerecognition technology, and compares the main research methods in this field. The preprocessof image is introduced briefly, such as image filtering, image binarization and image,morphology etc.. Beside, the key technology of the method of gesture recognition is studied.The main work of the hand gesture segmentation is: the method of segmentation in differentcolor space is summarized, such as the color space of HSV and YCrCb, and then the colordistance between two picture is defined. In addition, further study the color model in the HSVcolor space, and recombination of H S V color space. The effect of skin color segmentation isbetter in this color space. In addition, proposed a dynamic threshold segmentation algorithm.We can be an effective segmentation of the gesture through the above process.Secondly, the main work of gesture features extraction is the feature extraction methodof geometric features is studied include of topological features, sift features and fourierdescriptors; method of different recognition of static hand gestures and dynamic gestures isproposed for gesture recognition algorithm. In addition, the fingertip detection methods are studied, and get better results. And using of gesture area and perimeter of the contourproduces a new feature. We can get a set of feature vectors which reflects the gesture featuresthrough screening and analysis. The method of different recognition of static hand gesturesand dynamic gestures is put forward for gesture recognition. And according to the matchingalgorithm, gesture using the template is recognized in this paper. Experiments show that, thisfeature can improve the gesture recognition rate effectively.Lastly, Completed the design of hand gesture recognition based on computer vision byusing the development platform of VS2010software and OpenCV2.4.7Library.