Algorithm of Fingertip Detection and Its Improvement Based on Kinect
Chenqi Li, Rui Zhang, Zhen Liu, Chengcheng Hang, Zhijun Li · 2017
Fingertip feature is the most important feature of gesture recognition, and fingertip detection is the core of the interactive system. Because of high requirement of fingertip information for video gesture recognition technology, a method of fingertip detection is proposed based on information of depth image and human skeleton. Firstly, the Kinect is chosen to carry out the collection of the depth image, then the hand is positioned, the image is preprocessed, the eight neighborhood tracking algorithm and a series of other algorithms are used to split the gestures. Then the shortcomings of traditional K-curvature fingertip detection algorithm are analyzed, and it combines with fast convex hull algorithm for fingertip recognition. Finally, an algorithm called "scan fitting method" is proposed to solve the problem of fingertip close detection.