A consideration of effective feature extraction method for Nail Camera System

A text reader for mobile phone has a problem that the camera position is matched to a line in documents hardly. In this paper, we propose “Nail Camera System” by using character recognition to solve this problem. This system has an advantage that the position of the camera is matched to a line easily because the user attaches a small camera to user's finger. Additionally, a line can be traced by simple and natural motion of the finger. However, obtained image from this system is often distorted, because users' motion and angle of the finger is not stable. Therefore, in our method, four directional feature fields are used as a feature extraction method for character recognition. The effectiveness of our proposed approach is described with experimental results.