Estimating Characteristic Points of Human Body for Automatic Hand Pointing Gesture Recognition
The pointing gesture recognition system based on stereo vision and implemented using open source computer vision library (OpenCV) is presented. The system is able to detect pointing gestures and to determine the 3D pointing direction in real-time. To obtain input features for gesture recognition, at first, detect a person’s head and pointing hand, mark some feature points in detected regions of the head and pointing hand and then perform its tracking in 3D. Two cascades of boosted classifiers were used to detect the occurrence of the head and pointing hand. Results of experimental evaluation prove that the rate of correct detection is quite high, while tracking still needs improvements. Ill.4, bibl. 7 (in English; summaries in English, Russian and Lithuanian).
Copyright terms are indicated in the Republic of Lithuania Law on Copyright and Related Rights, Articles 4-37.