Gesture Recognition Based On Depth Information And Color Image

In order to provide a more natural human-computer interaction, people have thought touse gestures to interact with machines. The gesture is commonly used means ofcommunication in daily life, more in line with human behavior and habits, so the study ofgesture recognition has become a hot research problem of human-computer interaction.Predecessors in the process of gesture recognition, paid a great energy, and providevaluable practical experience, leaving a large amount of research literature, but also accessto good achievement. This article has made a research on the basis of previous work.First,we rendered the reference gesture images, and then we captured to be recognizedgesture images from kinect, finally, we matched the two kinds images to obtain ourresults.There are two main problems in the poress of gesture recognition. First of all, how toposition the gesture area. Second, how to realize a rotation invariance and scale invariancetemplate matching algorithm. We conbine depth information and skin color method toresolve the first problem. Through the gesture area adjustment method, we obtain a scaleinvariance feature for template matching. To identify gestures direction, we canreduce the number of template matching. At last, we obtained good experimental results.