Efficient estimation of character normal direction for Camera-based OCR
Kanta Kuramoto, Wataru Ohyama, Tetsushi Wakabayashi, Fumitaka Kimura · 2015
Handling characters the appearances of which are affected by three-dimensional (3D) rotation is a major challenge in Camera-Based Optical Character Recognition. Proper handling of these 3D rotated characters is expected to improve the performance of both detection and recognition of camera-captured characters. In this paper, we propose an efficient implementation of 3D rotation estimation for camera-captured characters. The proposed implementation requires small memory load and short computational time. We employ Linear Discriminant Function (LDF) instead of Modified Quadratic Discriminant Function (MQDF) for further memory reduction. The results of experimental evaluation using a large-scale alphanumeric character dataset showed that small number of dimensionality of original feature vector is sufficient for keeping accuracy of 3D rotation estimation and total amount of memory required for 3D rotation estimation is reduced from 141.0 MB to 6.6 MB. Moreover, we propose probabilistic rotation angle estimation and achieved improvement of estimation accuracy without increase of memory size.