Local absolute binary patterns as image preprocessing for grip-pattern recognition in smart gun

In a biometric verification system of a smart gun, the rightful user is recognized based on his handpressure pattern. The main factor which affects the verification performance of this system is the variation between the probe image and the gallery image of a subject, in particular when the probe and the gallery images have been recorded with a few weeks in between. One of the major variations is in the pressure distribution of images. In this work, we propose a novel preprocessing technique, Local Absolute Binary Patterns, prior to grippattern classification. With respect to a certain pixel in an image, Local Absolute Binary Patterns processing quantifies how its neighboring pixels are fluctuating. It will be shown that this technique can both reduce the variation of pressure distribution, and extract information of the hand shape in the image. Therefore, a significant improvement of the verification result has been achieved.