A Robust Table Detection Method for Distortion in Image Acquired from Camera
Toshiya Nakaigawa, Yoshiki Mashiyama, Yasue Mitsukura, Nozomu Hamada · 2019
In this paper, the robust table detection method for distortion is proposed in image acquired by camera. Images acquired by the camera contain distortion due to the curvature of the paper and it makes difficult to detect the table in images. In order to address this issue, a method of dividing the frame line of the table in the vertical direction and the horizontal direction and detecting the frame lines by curve approximation in each direction is proposed. Dividing the frame lines in each direction, it is possible to simplify the multiple curve detection. As a result, the lines detection accuracy in the curved surface image was 99.4%. In addition, similar results were obtained for curved images rotated by 90°. This method can cope with distortion in both the vertical and horizontal directions. From these results, it was confirmed the effectiveness of the proposed method.