CONF chen-iciap01/IDIAP Text Enhancement with Asymmetric Filter for Video OCR Chen, Datong Shearer, Kim Bourlard, Hervé EXTERNAL https://publications.idiap.ch/attachments/reports/2001/rr01-19.pdf PUBLIC https://publications.idiap.ch/index.php/publications/showcite/chen-01-19 Related documents Proceedings of the 11th International Conference on Image Analysis and Processing 2001 Palermo, Italy September 2001 192-198 Published in Int. Conf. Image Analysis and Processing, Palermo Italy, Sep. 26-28, 2001, IEEE Computer Society. Stripes are common sub-structures of text characters, and the scale of these stripes varies little within a word. This scale consistency thus provides us with a useful feature for text detection and segmentation. In this paper a new form of filter is derived from the Gabor filter, and it is shown this filter can efficiently estimate the scales of these stripes. The contrast of text in video can then be increased by enhancing the edges of only those stripes found to correspond to a suitable scale. More specifically the algorithm presented here enhances the stripes in three pre-selected scale ranges. The resulting enhancement yields much better performance from the binarization process, which is the step required before character recognition. REPORT chen-01-19/IDIAP Text Enhancement with Asymmetric Filter for Video OCR Chen, Datong Shearer, Kim Bourlard, Hervé EXTERNAL https://publications.idiap.ch/attachments/reports/2001/rr01-19.pdf PUBLIC Idiap-RR-19-2001 2001 IDIAP September 2001 Published in Int. Conf. Image Analysis and Processing, Palermo Italy, Sep. 26-28, 2001, IEEE Computer Society. Stripes are common sub-structures of text characters, and the scale of these stripes varies little within a word. This scale consistency thus provides us with a useful feature for text detection and segmentation. In this paper a new form of filter is derived from the Gabor filter, and it is shown this filter can efficiently estimate the scales of these stripes. The contrast of text in video can then be increased by enhancing the edges of only those stripes found to correspond to a suitable scale. More specifically the algorithm presented here enhances the stripes in three pre-selected scale ranges. The resulting enhancement yields much better performance from the binarization process, which is the step required before character recognition.