关键词: 嵌入式, 名片识别, 二值化处理, 目标函数

Abstract: Binary processing is one of the most important steps in the automatic recognition of business card, its effect will directly affect the subsequent layout analysis, character segmentation and recognition. While the existing binarization algorithm time complexity is very high, and the lack of pertinence. Therefore, a new fast binarization algorithm for a business card is proposed in the paper. This method uses color model and through optimizing an weighted objective function of error sum squares to find the optimal threshold, and what’s more, a fast iterative algorithm based on it is designed. The experiments show that, compared with the present algorithms, this new algorithm not only reduces the time complexity and improves the processing efficiency in an embedded environment of a business card image binarization processing. And for those complex business card also plays a good pretreatment effect. Thus, can remove the useless pattern recognition results, purification space, and lays a good foundation for the subsequent page segmentation, etc.

Key words: embedded environment, business card recognition, binary process, objective function