摘要
由于针织毛衫套口工序对操作工的视力和经验要求较高,使套口工序成为毛衫编织加工技术发展的瓶颈。为此,模拟手工套口中的关键环节:织物对齐,探索了一种结合计算视觉,控制织物对齐的软件系统,以实现套口工序的自动化和智能化。这个软件系统有3个创新点:使用直线检测绕过传统的双目计算框架,简化了系统设计的复杂性;去杂滤波机制保证了平行线的提取;聚类计数法能精准地识别刻度线数目。该系统在线运行能够智能地识别并判断两侧织物的对齐情况,从而证实视控化的套口机设计方案是可行的。
Due to the high requirements for operators' visual ability and experience, loop linking becomes the bottle neck of sweater knitting. In order to realize automatic and intelligent loop linking, the key process of web alignment in the manual operations is simulated, and a software system of controlling alignment based on computation vision is developed. There are three creative points in that software system: using line detection avoids the conventional binocular visual computation framework, simplifying the complexity of system designing; noise filtering mechanism guarantees the extraction of parallel lines; and number counter based on clustering recognizes precisely the number of graduation marks. The online running of the system can smartly recognize the web alignment at two sides, thus proving the feasibility of the design program of visual-control based linking machines.
出处
《纺织学报》
EI
CAS
CSCD
北大核心
2014年第1期117-121,共5页
Journal of Textile Research
基金
国家自然科学基金项目(NSF11001241)
浙江省教育厅基金项目(Y201121787)
关键词
套口机
直线检测
视控
聚类
linking machine
line detection
visual-control
clustering