论文部分内容阅读
针对二值数字化机械工程图中字符的提取和分割问题，提出了一种新的图文自动分离算法———行程编码匹配法，能对原始点阵图像的噪声水平、图形与字符粘连以及字符的大小和书写方向等因素几乎不加限制地进行处理．在此算法的基础上，提出的一种字符分割算法可以成功地将字符与字符的粘连分割转化为图的边割集问题．经对以此算法研制成的图纸识别软件系统ＧＩＲＳ８０的初步测试，提取率在９５％以上
Aiming at the problem of character extraction and segmentation in binary digital mechanical engineering drawings, a new algorithm of image automatic separation --- stroke encoding matching is proposed, which can measure the noise level, the adhesion of graphics and characters and the characters The size and direction of writing and other factors to be processed almost without restrictions. On the basis of this algorithm, a character segmentation algorithm proposed can successfully transform the adhesion segmentation of characters and characters into the edge cut sets of graphs. After the preliminary test of GIRS8.0, a drawing recognition software system developed by this algorithm, the extraction rate is over 95%