摘要
在数字影像中,理想的直线段由呈阶梯状排列的游程所构成。论文利用这一特征提出了一种检测直线段的新方法,它以逐行或逐列搜索带有缝隙的游程作为基础,搜索时能根据已取得的部分直线段预测下一段游程的起始像素位置,检测速度较快,获取的直线段较长,所得结果非常适合于用作为基元以合并成完整的直线。该算法的原理还可用来快速确定表格框线的转角。
In digital image an ideal line consists of some run-lengths which are arranged as a flight of stairs.According to this feature a new method for line segment detection is presented here based on searching for run-lengths row by row or column by column.In the process of line segment detection acquired part is used to predict the position of the first pixel of the next run-length,moreover small fissure is allowed to be contained in every line segment.By comparison with several other algorithms ,this method is performed more quickly and the average length of its output line segments is longer.So that these output line segments are more suitable for merging into final lines.The principle of this algorithm has proved to be efficient for determining the angle of rotation of a form rapidly.
出处
《计算机工程与应用》
CSCD
北大核心
2004年第22期71-74,共4页
Computer Engineering and Applications
基金
河海大学青年创新基金资助项目
关键词
直线检测
计算机视觉
表格识别
图像分析
line detection,computer vision,form recognition,image analysis
