高级检索

基于直线连续性的页面倾斜检测与校正

Page Skew Detection and Correction Based on Line Continuity

  • 摘要: 在文档扫描过程中,输入的文档图像不可避免地会发生倾斜现象,而布局分析及字符识别算法对页面倾斜都十分敏感,因此倾斜检测和校正是文档分析预处理的重要环节.文中提出了一个基于直线连续性的倾斜检测方法.它将字符连通区包围盒底边中心点作为特征点,利用文本行中特征点与基线的关系,计算出基线的方向,即为页面倾斜方向.接着,介绍了一种基于偏移值的倾斜校正方法.实验证明,该算法速度快,准确度高.

     

    Abstract: During document scanning,the input document images inevitably introduce skew, while the algorithms for layout analysis and character recognition are every sensitive to page skew.For detecting the page skew,center of the lowest row pixels within each bounding box of connected component is taken as eigen-point.The baseline direction that verifies the page skew can be calculated according to the eigen-points in a text line.Then a correction method based on horizontal and vertical offset is provided.Experiments show that this approach is fast and accurate.

     

/

返回文章
返回