基于形态学和霍夫变换的文档图像倾斜检测  

Skew Detection of Document Images Using Mathematical Morphology and Hough Transform

作　　者：吴欣[1] 张志伟[2] 

机构地区：[1]南京理工大学计算机科学与技术学院,江苏南京210094 [2]中国科学技术大学工程科学学院,安徽合肥230027

出　　处：《南京理工大学学报》2009年第2期178-182,共5页Journal of Nanjing University of Science and Technology

摘　　要：为了解决各种文档倾斜方法难以同时兼顾检测精确度和处理速度,该文提出了一种基于形态学和霍夫变换的文档倾斜检测方法。首先用形态学方法对文档图像进行行程平滑、消除噪声和边缘提取,然后用霍夫变换检测出文档的倾角。通过改进形态学运算,加快了检测速度。实验结果表明,该方法对A4大小的文档图像的平均检测用时为1.78 s,最大误差为0.2°。To solve the problems that the existent methods detecting skew of documents have difficulties in precision, accuracy or speed, a method based on mathematical morphology and the Hough transform is presented. Morphology method is used to smooth the document image, eliminate the pixel noise and detect the edge of the text row, and the Hough Transform is applied to detect the skew angle. In order to improve the performance, fast morphology image transforms are implemented. The detection time for A4 paper is 1.78 s, and the biggest error is 0.2°. The test results show that this method is precise, accurate, fast and robust.

关 键 词：倾斜检测 文档图像处理 数学形态学 霍夫变换 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]