Automatic text extraction from color image
Wenping Liu, Hui Su, Chang Y. Chi · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2000
An effective and fast method to extract text from color background is proposed in this paper. Under the assumption that the text strings with high contrast are usually important, a modified sobel operator is utilized to transform the color image into an edge binary image and an edge-based smear processing is also used to speed up the color image processing time greatly. Combining the feature- based text candidates identification with the moment preserving region segmentation, we can effectively extract text from scanned true-color image varying in text font, size and color complexity. The extracted texts are finally input to the Optical Character Recognition system. Experimental results demonstrate the high speed performance and feasibility of this method.