Research And Implementation On The Photo Album Browser Based On Gestures Identification

The fast development of science and technology, has brought great changes to the people’s life, Techniques are becoming more and more popular with the improvement of the intelligentialize degree.Gesture is such a natural and intuitive way of human-Computer Interaction that gesture recognition technology will have much development in the area of intelligence in the future. This paper designs and implements a photo album browser system based on the visual gesture recognition technology, successfully achieving the real-time operating of the album browser through the transformation and move of gestures.Gesture recognition process based on computer vision can be roughly divided into four parts, including gesture segmentation, gesture modeling, gesture feature extraction and gesture recognition. The following three are the major work of this paper on the basis of consulting a large number of reference materials:Firstly, this paper puts forward a kind of skin detection algorithm of brightness adaptive elliptical model based on YCbCr color space, and realizes the segmentation of the different gestures in video sequences. After analyzing and comparing the nature of a variety of color spaces, this paper combines YCbCr color space and elliptical model to detect the color of skin. At the same time, through a large number of experiments, in view of the correlation of brightness and chrominance information, we use the adaptive threshold value judgment to adaptively adjust the ellipse center position and long and short half shaft where skin color maps in the Cb-Cr plane. Experiments show that the algorithm adopted in this paper can well detect skin color area of different skin color groups, better than nonlinear segmentation threshold value method, especially in the area where the luminance signal is stronger, which better reflects the accuracy of the algorithm in brightness self-adaptive detection.Secondly, in gesture recognition, we apply LBP local texture feature to gesture recognition, analyze the distribution characteristics of LBP operator and gesture feature, and put forward a recognition algorithm combining block and the whole based on the LBP local texture feature. We firstly divide gestures into different sub-blocks, extract and analyze LBP textural features image of every sub-block, then combine it with the overall LBP features to constitute characteristic vector of gestures. At last we use SVM classifier to do classification training of different gestures. The experiment results indicate that the method can realize the recognition of different gestures quickly and exactly, and identification accuracy is high.Finally, this paper designs and realizes the photo album browser system based on the gesture recognition. We successfully realize real-time operation of browsing reading, joining-together display, close-up grab, free scaling and window shifting functions in all the pictures of photo album through the gesture transformation and movement, and verify that the skin detection algorithm and LBP local texture characteristics identification algorithm put forward in this paper is effective.