The Gesture Recognition Based On Computer Vision

With the development of artificial intelligence,interaction between human and computer become more and more varied, especially become one of important elements in modern scienence and technology products. Especially as the standard of people’s life is enhanced, the interactive mode of electronic products have become increasing demanding, even the interaction quality has become the primary measure to the standard electronic product, therefore the design of a good interaction has become the research direction.However, now the human-computer interaction give the priority to contact type, such as a keypad, mouse, handwriting board etc, this method need hands in direct contact with the device of receive information, this tied the manpower freedom and flexibility of information receiving devices, so the non-contact way of human-computer interaction style has become a hot research topic, for axample, speech recognition developed in recent years, image recognition graphics.Gesture is a form of information transmission mode, because of its convenient and easy to come ture without additional devices, so the hand gesture recognition has become a hot issure of non-contact interaction. By hand as an input device, reduce the use of the equipment such as mouse,keyboard, greatly increases the human-computer interaction. This paper studied the gesture position based on the analysis of domestic and foreign gesture recognition. The main work is as follows:This paper simple analysis the implementation of gesture recognition system is divided into four parts:gestures positioning, gesture feature extraction, hand tracking and gesture recognition. Briefly introduces the commonly used algorithm and the advantages and disadvantages of each stage. Aiming at the shortcoming of the skin detection or motion detection respectively in gesture positioning, put forward using the two ways together to gesture segmentation, skin color has good clustering, First by motion detection to extract the target, it can remove the static scene, and then the motion area extracted using skin color detection,reducing the range of skin color detection, also lowing the calculated quantity, make up for the deficiency of the single using skin color detection or motion detection, to pick up the gesture from the background effectively, and using eight neighborhood search method to extract the contour gesture, and based on this architecture gesture reccognition system.This paper mainly is divided into six parts:Introduction; Overview of gesture recognition each part algorithm; Gesture pretreatment; gesture contour extraction; Feature extraction and hand tracking; Gesture tracking and gesture recognition system experiment; Summary and prospect. With the help of computer vision library OpenCV, acquisition gesture sequence through the camera, preprocess the image,using Hu moment to extract gesture feature and then tracking, combined with Markov model, finally get the gesture recognition system.