摘要
基于横纵投影的版面理解递归算法具有简单易行的优点,对无嵌入式矩形版面可以做出比较理想的分析和识别.但是,这种算法也有效率不高、时间复杂度大等缺点.本文分析了这种典型的二分算法,提出了版面理解的多叉树模型,以及在此基础上的基于横纵投影的递归多分法,从而使这种算法得到改进,提高了运算效率.
Recursive dichotomy algorithm based on projection of horizontal/vertical orientation is a practical bur, inefficient algorithm for document understanding. When applied to the nonembeded-document images, it can get satisfied results. To improve the effciency and reduce the time-complexity, this article analyzes the dichotomy algorithm and provides a polytomy one based on polytree model.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
1997年第2期118-126,共9页
Pattern Recognition and Artificial Intelligence
关键词
版面理解
版面分析
版面识别
投影
递归多分法
Document Analysis, Document Recognition, Document Understanding, Projection, Recursive Polytomy
