摘要
针对工程图中的字符特别是粘连字符的提取与分割这一工程图识别与理解中的难点,提出并实现了一种新的基于光栅直线识别的字符串整体提取算法,能比较准确地提取图纸中的粘连字符串,并对提取后的字符串进行空间位置分解组合,从而确定字符串的分组与方向,实现较高层次上的图文分割。
A new technique for extracting character strings in engineering drawings based on raster line recognition is put forward. The connected regions in an image are acquired using the growth Al- gorithm of the connected segment through the run-length coding. An integral line recognizing algo- rithm based on connected segments matching is given to extract lines of arbitrary orientation. The ad- hered character strings are acquired through a search for the profile of lines. A character spatial syn- thesis algorithm is used to merge the split regions into characters and to group the dispersive charac- ters into definite strings . The mutually adhered characters are segmented into single ones to be sent to perform the recognition.
出处
《华中理工大学学报》
CSCD
北大核心
1997年第4期43-46,共4页
Journal of Huazhong University of Science and Technology
关键词
字符提取
字符分割
图像识别
工程图
机械图
run-length coding (RLC) : character extraction : character segmentation
image recogni- tion
vectorization