Finger Recognition And Tracking System Based On Single Optical Camera

We have witnessed a rapid evolution of computing, communication and displaytechnology, but the human-computer interface appears to fall behind. However,traditional interface devices provide an attractive alternative to the cumbersomeinterface devices for human-computer interaction that are typical today. At present,the new human-computer interaction methods have been researched widely byscholars in that it does not require the complicated hardware to implement it.Therefore, the finger detection based on the optical camera has been the hotspot ofresearch and it has extensive application prospect and research value.In this thesis, we have achieved the finger detection and hand tracking system.This thesis is composed of three different parts: hand extraction, fingertipsrecognition and hand track. The hand extraction process provides the target to thefingertip recognition and hand tracking process, the tracking process determines thesearch area for hand extraction in the next frame. The fingertips recognition justplays a role after the hand extraction. These three parts integrate to implementtracking hands and detecting fingertips in the Sequential frames.During the research and achieve procedure, the research focuses are handextraction and fingertips detection. They specially take advantage of hand traits toidentify the position of fingertips and hand area. Hand extraction which relies onthe homogeneous distribution of skin color to extract hand area and adopts the skinmode that established in the YCbCr color space to eliminate the interference ofbrightness accomplished a favorable outcome in the complicated background andstrong light circumstance. The fingertips recognition method adopts the curvaturewhich is stable in the tips to locate its position. Meanwhile, we also used the trait offingertips that is situated on the convex hull of hand contour to get rid of theinterferences, because of two-dimension projection of fingertips distribute on theperiphery of hand area. The tracking method utilized in this paper is the well-knownCamshift scheme, which is based on the Color Histogram to distinguish the targetand to track hand motion. This peculiarity is fit for tracking objective which hasuniform color distribution, especially for hands. At last, in the actual circumstance, the program has been tested in differentconditions. It can be conclude that the program is operating smoothly in respectivesetting which like strong light environment, different hand shape and with skincolor disturbance. After analyzing the result, we have a consequence of thisprogram that the program works well and sufficed the demand of design.