【作者】 马大国；

【导师】 金维洙； 王阿川；

【作者基本信息】 东北林业大学 ， 机械设计与理论， 2002， 硕士

【摘要】 原木分等一直是林业生产和加工等部门的一个关键环节，长期以来所使用的方法是人工检测。人工检测受人为因素影响很大，工作效率低，不能形成自动化的生产线。节子的类型、大小和多少是原木分等时的一项重要检测指标，所以本文以原木节子的检测作为研究对象。 本文首先对计算机视觉检测技术进行了概述，然后较为深入的阐述了图象处理和模式识别等技术，对现有的木材缺陷检测的主要方法进行了总结评述，阐明了目前视觉检测技术在木材加工中的应用及发展水平；本文运用微软公司的VisualC++6.0作为开发工具。这些都为本系统的设计做好了技术准备。针对本文的研究对象节子的特殊性，本文在节子的图象处理和模式识别过程中都提出了适合于本实验的新的技术和方法。通过对实验结果的分析，本文成功的设计了原木表面节子的计算机视觉检测系统，详细论证了系统检测原理，编制了系统软件，证明了系统设计的可行性和合理性。 原木表面节子的计算机视觉检测系统的设计是合理的，具有坚实的理论基础和可行性。本文的研究成果将为原木的自动分等系统的建立奠定了基础。更多还原

【Abstract】 Classifying log is always a key part in log producing departments and log processing departments. The manual measurement has been adopted for a long time,which is greatly affected by artificial factor resulting in low measuring precision and can’t be processed by automatic line. The kind dimension and quantity of knots are important detecting indexes in classifying log. So we choose knots as our research target in the thesis.Firstly,the thesis concluded the methods of detecting technology by computer vision in forest industry. Based on analyzing the development of image processing and pattern recognition in detail and on outlining the present methods,the thesis illuminated the present level of the computer vision technology used in wood processing;Visual C++6.0 of Microsoft Company is applied as a tool to develop image-processing module. And various feasible algorithms and methods were expounded in image processing and pattern recognition. The thesis successfully designed the software architecture of knots detecting system by computer vision. Through experiments,the detecting system’s feasibility and rationalization were proved.The detecting system of log’s superficial knots by computer vision is reasonable,and its detecting principles have a sound theoretical basis. The thesis’s achievements will provide theoretical and feasible support in classifying log.更多还原