Research The Technology Of Hand Gesture Feature Extraction And Recognition

In nearly half a century, the rapid development of computer technology has been applied, human-computer interaction technology(Human Computer Interaction, HCI) has become an integral part of our daily life. Vision-based gesture recognition is a branch of human-computer interaction, and then it becomes a hot gesture recognition experts and research scholars. Because of the natural gesture has spatial and temporal differences and diversity, so Vision-based gesture recognition has many practical value and theoretical significance, it is a research topic which is very interesting and challenging.This paper studies about the static gesture recognition in the complex background,including the aspects of gesture detection segmentation, feature extraction, and gesture recognition gesture.aiming at improving the accuracy of the static gesture recognition.The main research work of this paper includes the following aspects:First, gesture segmentation section: This section includes two aspects: preprocessing and gesture segmentation. Preprocessing part is the image size is normalized so that all the pictures have uniform size standard for feature extraction and classification study. In the aspect of gesture segmentation, we select YCb’Cr ’Ellipse Skin color space model. Color in YCb’Cr ’color space has good clustering, and this region which the skin is clustering does not change with the brightness changing. Experiments show that the method can be split gesture accurately.Second, gesture feature extraction section: The paper explore about gradient histogram(HOG) feature, and make the appropriate improvements. Being point at the difference for each block in the gradient has different influence in the classification,accumulate the error, gesture recognition results would be affected.. Therefore it abandoned the equally weighted value of the classic HOG concept, this paper improved HOG features, conformed W-HOG feature, reduced the accumulated error effectively. In the same time, Extracted W-HOG feature in the standard static gesture library, in order to verify its validity.Third, Gesture recognition section: In this paper, using the support vector machine(SVM) in the gesture recognition, support vector machine has certain advantages in solving nonlinear problems and the smaller number of samples and other issues. We choose Bochumn Gestures static gesture database with database-based experiments, and taking a group of static gestures in the natural background as an auxiliary database. Experimental data show that: the algorithm has good recognition rate in the static gesture recognition.