摘要
对工程图识别与理解研究现状进行深入和系统的综述。工程图识别与理解的核心在于利用知识表示、图形匹配、符号识别、几何推理、语义提取等技术,自动获取工程图中各种显式描述(如几何图元、工程符号、工程对象等)及隐式信息(如设计语义等)。分别对工程图矢量化、工程符号识别及工程对象识别进行了系统分析。在此基础上对工程图识别性能评测、基准库等进行总结。最后对现有研究的难点及进一步研究方向进行了展望。
A comprehensive survey on the automatic recognition and understanding methods for engineering drawings is proposed. Target of interpretation of engineering drawings lies on automatic extraction of explicit descriptions （e.g., graphical primitives, engineering symbols, engineering entities） and implicit semantics behind the drawings. Vectorization, recognition of engineering symbols and engineering entities are analyzed systematically. Based on these, performance evaluations and benchmarks are also proposed. Finally existing difficulties and future research are discussed.
出处
《图学学报》
CSCD
北大核心
2012年第5期1-12,共12页
Journal of Graphics
基金
国家自然科学基金资助项目(61272218
61021062)
973资助项目(2010CB327903)
教育部新世纪优秀人才支持计划资助项目(NCET-11-0232)
关键词
工程图
理解
识别
知识表示
矢量化
engineering drawings
understanding
recognition
knowledge representation
vectorization
