Video Background Modeling In Hand Gesture Recognition

The traditional human-computer interaction (HCI) is based on mechanical devices such as keyboards, mouses, joysticks or gamepads. In recent years there has been a growing interest in methods based on computational vision due to its ability to recognise human gestures in a natural way. For any kind of computer vision processing it is important (and is often the first step) to separate out objects from the image sequences that we are viewing. This process is called background modeling. The results of the background modeling relates directly to the tracking or detection accuracy in the subsequent process. Thus the study of robust background modeling algorithm is very important.In this paper, background modeling and object extraction algorithm is studied for hand gestures interactive application. At first, the existing methods of background modeling and updating is analyzed. We focus on the Gaussian Mixture Model (GMM) and the Stochastic Approximation for background modeling, and improvement is carried out to achieve better results.It is not the best model in practice that each pixel is modeled a fixed number of Gaussian distributions for the original GMM algorithm. In this paper, an effective adaptive background updating method is presented by choosing the number of components for each pixel in an on-line procedure. And the threshold value is also given a flexible selection to decide the background or foreground. The algorithm can automatically adapt to the scene.The stochastic approximation method is based on the Robbins-Monro algorithm and mixtures of uniform distributions and multivariate Gaussians with full covariance matrices is used in the model. In this paper, the background mutation is mainly considered. Based on the same test video, a contrast experiment is carried out with the GMM algorithm. The results show that the stochastic approximation algorithm is superior to the GMM algorithm even in complex scene.These works lay a good foundation for gesture tracking, recognition and analysis, and it can be used to build the intelligent systems based on gesture interactions in the future.