Disclosed is a depth map-based heuristic finger detection method. The method comprises: acquiring, from a depth map of a user, a hand connected domain; calculating a central point of the hand connected domain; performing calculation to obtain multiple distant points having maximum 3D geodesic distances from the central point in the hand connected domain; detecting a fingertip and a finger region from the multiple obtained distant points; and outputting a position of the fingertip and the finger region. The invention employs a 3D geodesic distance to calculate and detect a fingertip of a user without extracting an outline of a hand region, thereby increasing robustness of gesture detection, and reducing a detection error rate. The method has high finger detection accuracy and a high calculation speed.
展开▼