基于人眼视觉的对不良照明图像的二值化方法(英文)

Novel Adaptive Thresholding Technique for Poor Lighting Document Image Based on Human Vision Characteristic

提出一种新的基于视觉特性的自适应阈值分割方法.利用人眼视觉对比敏感度特性把图像块分成两类,分别借助OTSU算法和模拟人眼识别过程的多尺度模糊隶属方法实现对这两类具有不同灰度特性图像块的自动阈值分割.实验结果表明,使用该方法能够有效的克服不良照明的影响,还原图像的原本特征信息,二值化效果较好.

An algorithm of adaptive threshold partition based on human visual characteristic was proposed. All the image blocks were divided into two types by classification algorithm based on human visual contrast sensitivity. Then blocks with different types were binarized by OTSU and Multi-scale fuzzy membership grade algorithm respectively. The algorithm was realized with VC＋＋ language. Experimental results show that this proposed method is well suitable for poor quality document image binarization.