Gloved Finger Recognition by Curve Matching

Abstract

Finger recognition is an important basis of natural human-computer interaction, but existing researches are mainly on bare-hand finger recognition. Due to the fact that people need to wear gloves because of cold weather, working environment, etc, this paper presents a method to recognize gloved fingers by curve matching. This method firstly extracts contour curve from image edge. After fitting the contour curve, it calculates curvature integration of the fitted curve and then assesses similarity between the curvature integration and finger template's. Finally, it recognizes fingers and positions fingertips according to the assessment. And the finger template which is used to match contour curve can be chosen at random without training. The experiment shows that this method is effective in recognizing fingers and positioning fingertips under the condition of scale variations, rotation variations and affine transformations, whether the subjects wear gloves or not.