Real time handwritten digit recognition on mobile devices

Handwritten digit recognition (HDR) is an important problem in computer vision and has been studied for many years. However, real time HDR on mobile devices still faces the challenge because of hardware limitation and algorithm computation complexity. To address this problem, we propose a method which adopts a non-overlapped pyramid histogram of oriented gradients (PHOG) for HDR in this paper. Each cell in every layer of the PHOG can be calculated from the cells in the lower layers except the first layer. Our experimental results show that the recognition speed using the proposed non-overlapped PHOG is about 4 times faster than the half-overlapped PHOG and keep a relative high accuracy with a feature that only has hundreds of dimensions. To the best of our knowledge, it is the first attempt to achieve a real time HDR application on mobile devices.