【作者】 王武；

【导师】 叶明； 郑启旺；

【作者基本信息】 南京航空航天大学 ， 工程硕士（专业学位）， 2018， 硕士

【摘要】 近年来,智能手机行业迅猛发展,不同手机生产企业之间的竞争越发激烈,企业为了提升产品自身竞争力,对于手机壳体表面质量要求也越来越高。传统手机壳体表面缺陷检测方法采用人工检测法,检测效率低,成本高。机器视觉用于缺陷检测具有检测效率高、实时性好等优点。因此,手机壳体表面缺陷视觉检测已成为未来发展的重要方向。由于手机壳体表面的缺陷尺寸小,且壳体表面,尤其是其侧面往往由不同形状的平面和曲面构成,而这些平面和曲面的表面加工工艺也不尽相同,因此被检测区域的光照反射类型差异大,导致对缺陷的识别造成干扰,影响最终缺陷检测结果的正确性。针对上述问题,本文以目前手机壳体中常见的两种表面(高光表面、亚光表面)的划痕缺陷以及亚光表面孔洞边缘的磕碰缺陷为检测对象,开展了手机壳体表面缺陷视觉检测技术的研究。在此基础上,主要进行了缺陷检测系统总体功能设计、成像系统设计以及缺陷检测算法设计。其中成像系统的设计内容主要是针对被检测壳体表面的反射特性,设计了相应的照明方式,以达到突出缺陷同时抑制反射光干扰的目的,并通过实验验证了该照明方式的可行性。在缺陷检测算法方面,对于高光表面,采用了局部自适应阈值分割算法来提取划痕缺陷;针对亚光表面的细浅划痕缺陷易受噪声干扰的问题,利用划痕缺陷的线状特性,首先运用各向异性高斯导数滤波器分割出特定方向的划痕缺陷,再用不同方向的直线结构元素进行形态学闭运算以连接断续划痕,接着利用几何特征完成缺陷的提取;针对亚光表面孔洞边缘的磕碰缺陷,首先利用形态学算子完成对孔洞区域的提取,然后进行孔洞区域匹配,最后用基于边界分段傅里叶特征来实现缺陷的检测。基于上述研究,本文实现了手机壳体表面划痕缺陷以及孔洞边缘磕碰缺陷的检测。实验结果表明,本文设计的缺陷检测算法漏检率低于4%,误检率低于5%,能够实现长度为0.4mm以上划痕缺陷的检测,磕碰缺陷检测的最小检测长度能达到0.25mm,基本满足实际的检测需求。更多还原

【Abstract】 In recent years,with the rapid development of smart phone industry,the competition among different cell phone manufacturers is more and more intense.In order to enhance the competitiveness of products,the requirements for the surface quality of mobile phone shells are increasing.The traditional detection method of the surface defect of the shell of mobile phone adopts the artificial detection method,which has low detection efficiency and high cost.Machine vision for defect detection has many advantages,such as high detection efficiency and good real-time performance.Therefore,visual inspection of mobile phone shell surface defects has become an important direction of future development.Because of the defect size mobile phone shell,and the shell surface,especially its side is composed of flat and curved surfaces of different shapes,and surface processing technology of these flat and curved surfaces are not the same,so the detected region light reflection type differences,leading to the identification of the defects caused by interference,affects the correctness of the final defect detection results.Aiming at the above problems,based on the current mobile phone shell in two common surface(specular surface,matt surface)scratch defects and matt surface edge bump defects as detection object,the research of visual inspection technology for surface defect of cell phone shell is carried out.On this basis,the design of the imaging system of the defect detection system and the design of the defect detection algorithm are mainly carried out.The design content of the imaging system is mainly aimed at the reflection characteristics of the surface of the detected shell,and the corresponding lighting modes are designed to achieve prominent defects and suppress the interference of reflected light.The feasibility of the illumination method is verified by experiments.In the defect detection algorithm,the specular surface,using local adaptive threshold segmentation algorithm to extract the scratch defect.For the problem of thin and shallow scratch defect on subsurface easily disturbed by noise,using the linear characteristics of scratch defects,the anisotropic Gauss filter is first used to extract the scratching defects in a specific direction,then the linear structural elements in different directions are used to carry out morphological closed operation to connect the intermittent scratch,and then the geometric feature is applied to complete the defect extraction.Aiming at the defects of hole edge bump of matt surface,the morphological operators are used to extract the hole area,the hole area matching is carried out,and then the defect detection is realized by Fourier characteristics of segment boundary.Based on the above research,this paper realizes the detection of the mobile phone shell surface of scratch defects and hole edge bump defects.The experimental results show that the missing detection rate of the defect detection algorithm is below 4% and the false detection rate is less than 5%,the detection of scratches with length of 0.4mm can be realized,the minimum detection length of the bump defect detection can reach 0.25 mm,basically meet the actual needs of the detection.更多还原