摘要
工程图纸扫描识别和字符笔画提取一直是模式识别中的难点问题．为克服细化算法逐象素处理的局部特性，已提出了一些整体算法．基于图段的整体识别算法突破了细化算法在象素层次做局部操作的局限，力图抓住图象的宏观特征进行整体识别；但它在处理小线素、曲线及字符图象时仍有较大局限性．本文采用扫描串（行程编码）作为基本处理单元，动态扫描整个图象数据，利用预先建立的信息头指导后续跟踪，得到了更精确的矢量化结果．本文提出的算法可适用于各种不同工程图纸的处理，用来提取汉字笔画也得到了满意的结果．
The recognition of scanned image of engineering drawings and the extraction of character strokes are very difficult in pattern recognition. To overcome the shortcoming of thinning algorithm, many global algorithms are devised. Among them the global algorithm based on graphical segment directly recognizes the image through the macro features of image. But it can’t process curve and character image. The new algorithm takes scan strip as the basic image process unit and scans the whole image dynamically. A parameter head having been set up is used to supervise the succeeding trace procedure, so more precise vectorization result is obtained. The satisfying result of processing many different kinds of image, including Chinese character image, indicates that the new algorithm is powerful and adaptive.
出处
《软件学报》
EI
CSCD
北大核心
1998年第6期426-431,共6页
Journal of Software
基金
国家杰出青年科学基金
关键词
工程图纸
扫描识别
矢量化算法
工程设计
Recognition of scanned image of engineering drawing, character stroke extraction, global recognition, scan strip.
