Influence of Color-to-Gray Conversion on the Performance of Document Image Binarization: Toward a Novel Optimization Problem

摘要: This paper presents a novel preprocessing method of color-to-gray document image conversion. In contrast to the conventional methods designed for natural images that aim preserve between different classes in converted gray image, proposed conversion reduces as much possible (i.e., intensity variance) within text class. It is based on learning linear filter from predefined data set and background pixels that: 1) when applied pixels, minimizes output response 2) maximizes response, while minimizing variance Our (called learning-based color-to-gray) conceived be used binarization. A 46 historical created evaluate subjectively objectively method. The demonstrates drastically its effectiveness impact performance state-of-the-art binarization methods. Four other Web-based sets are scalability