Enhancing Character Recognition by Optimizing Focal Length

This paper proposes to use zooming function of digital cameras for robust character recognition and to control focal length for improving the performance even in dim light conditions. For this purpose, we determine the optimum focal length vs. camera-object distance through several recognition experiments. Recognition performance for insufficient illuminated images can be enhanced by adopting the optimum focal length. Experiments involving multiple fonts show that our method improves the recognition rates over a wide range of camera-object distances and illumination variations.

• Type: Journal
• Year: 2002
• Where: MVA
• Authors: Takahiko Shimmura, Yoshinao Shiraki, Naoki Mukawa, Minako Sawaki, Norihiro Hagita
• Keywords: Computer Vision, Focal Length, Focal Length Vs, MVA 2002, Optimum Focal Length