摘要： 表格的识别近年来受到了越来越多的关注，已取得一些行之有效的成果。其中的部分算法可抗任意角度的倾斜，然而其运行效率却都低于投影类方法。而投影方法虽然效率很高，但不能处理严重倾斜的表格。文章给出一种新的检测倾斜表格框线的途径。算法利用数字影像中直线段的结构特征高效地检测出表格的一条边框线，并由此获得所需的角度值，进而沿着角度所指定的方向迅速搜索出边框线。整个检测方法的效率类似于投影法，同时可抗任意角度的旋转。

关键词: 直线检测, 表格框线检测, 表格识别, 图像分析, 表格倾角检测

Abstract: More and more attention is paid to form extraction in recent years, and a number of efficient results are obtained. Some of the presented algorithms can be used to process a form image with an arbitrary skew angle, but they are not the most efficient methods. A projection approach can detect form frame-lines more efficiently but fail to detect any line when the form slants severely. A new idea is given here to measure this sort of skew angle rapidly, it first detects one of the form’s frame lines by using the structural characteristics of digital lines, and then calculates the skew angle’s value. Along the directions determined by the angle, the form frame-lines are searched out. The method is both efficient and anti-skew.

Key words: line detection, form frame-line detection, form recognition, image analysis, form skew angle measurement

中图分类号: 

• TN911.73