改进Niblack算法及其在不均匀光照条件下的应用

Improvement of Niblack Algorithm and Its Application in Uneven Illumination

文本二值化是光学字符识别的关键技术,但在光照不均的情况下,采用传统全局阈值二值化在图像过亮或暗区域情况下会造成大量文字信息丢失,因此通常采用局部阈值二值化方法。Niblack二值化是一种经典的局部阈值法,能够提取全部文字信息,缺点是存在大量伪影,且运算效率低,但优点是方法简单,易于实现。算法。该算法结合像素点空间八邻域灰度信息,能自适应调整阈值,逐点进行二值化处理,从而基本消除了伪影,并使用积分图法使运算时间从30s降低到3s,同时运用形态学腐蚀操作对笔画进行增强。实验结果表明,与传统Niblack、Sauvola等算法相比,在光照不均条件下,该方法图像噪声少、速度快,笔画更清晰,且更易于识别。

Text binarization is the key technology of optical character recognition,but in the case of uneven illumination,a large amount of text information is lost in an image that is too bright and dark in the use of traditional global threshold binarization. So the lo. cal threshold binarization is usually used. Niblack binarization is a classical local threshold method which can extract all text informa. tion,but there are a lot of artifacts,and the operation efficiency is low. The algorithm is simple and easy to implement. If its shortcom. ing is solved,it is easy to be widely used. In order to solve the problem of Niblack algorithm,a Niblack algorithm based on neighbor. hood information is proposed in this paper. The algorithm can adaptively adjust the threshold value and binary processing point by point,which can eliminate the artifacts,and reduce the operation time from 30 seconds to 3 seconds by using the integral graph meth. od. The morphological corrosion operation was used to enhance the stroke. The experimental results show that compared with the cur. rent Niblack-Sauvola algorithm,the proposed method has the advantages of less noise,faster speed,clearer strokes and easier recog. nition under the condition of uneven illumination.