Hand Gesture Recognition Using Deep Neural Network and Its Implementation in Augmented Reality

In this paper, we study the augmented reality and its application in intelligent human-computer interaction. The complex structure in the visual background is a major challenge for gesture segmentation. First, we use a robust skin color segmentation method to preprocess the input image. Second, visual texture features are analyzed and modeled for hand region recognition. Third, finger landmarks are annotated and configured by active shape model. Finally, the hand gestures are recognized and used for enhanced interaction. Experimental results show that the proposed gesture recognition system is robust against various background changes and illumination changes.