摘要
为将存贮在纸介质上的工程图纸信息转化为 CAD的数据格式 ,研制了工程图纸矢量化处理系统。该系统将工程图纸信息自动输入到计算机系统中 ,形成点阵图像 ,经过图像预处理、字符识别、图形矢量化等处理步骤转化为 CAD文件。解决的系统关键技术包括 :蓝图的消蓝去噪 ,提取粘连和非粘连字符的图文分离算法 ,基于复数域非线性指数自回归模型的神经网络字符识别算法 ,快速细化处理 ,自适应节点修正算法 ,跨节点矢量跟踪算法 ,基于神经网络的箭头识别 ,图形的分类整合等。系统在 Windows 95操作系统上运行 。
Engineering drawing vectorization and processing system (EDVPS) was developed to convert paper drawings to CAD format. EDVPS can automatically scan a paper drawing into the computer as a raster image and transfer the scanned digitized drawing to a commercially available CAD system using scanned image preprocessing, text and symbol recognition, and graph vectorization. The key EDVPS technologies include noise removal from the raster image, extraction of characters using a novel text/graphics separation algorithm, complex nonlinear exponential autoregressive model for character and symbol recognition using neural networks, rapid thinning approach, adaptive node regulation algorithm, cross node tracing algorithm to obtain accurate vectorization of cross lines, neural network based arrow recognition, and classification and mergence of vectors. EDVPS was developed in the Windows 95 environment and has been used by many users.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2000年第4期35-38,共4页
Journal of Tsinghua University(Science and Technology)
基金
北京市科委重点基金项目 !(952850400)
关键词
工程图纸
矢量化
字符识别
细化
engineering drawings
vectorization
character recognition
thinning
