基于微镜器件和复合传感器的高反射回转面缺陷检测新方法

Highly reflective precision rotary parts have been more and more widely used, but defects (such as dents, scratches, abrasions and foreign materials) over the highly reflective surface of this parts may change the service performance, life, and safety properties of the products. Therefore, defect inspection is very important for surface quality assurance. However, the current technologies of defect inspection, which are generally based on vision sensor, could not reach a balance among the accuracy, efficiency and universality of defect recognition, so that their application is still restricted. A novel quantitative inspection method for surface defects on highly reflective rotary parts based on digital micro-mirror device (DMD) and combined sensor, is proposed in this project. Main works include: Firstly, a new illumination system based on DMD, which can obtain the adaptive fringes, will be studied. The illumination system can improve the sensitivity for visual inspection of surface defects and the adaptive ability for rotary surface, which has the different curvature radii. Secondly, a novel defect extraction method based on image subtraction integrating with active contour model is proposed, which can rapidly, precisely extract boundaries of the defect regions for the obtained visual images. Thirdly, the quantitative assessment and classification of the defects extracted based on laser sensor will be studied. Ultimately, this novel inspection method for surface defects on highly reflective rotary parts will be established, which will settle the theoretical and technological foundation for further development of corresponding device of defect inspection for the surface of highly reflective rotary parts. This novel inspection method based on DMD and combined sensor has the outstanding advantage of the high accuracy, efficiency and universality of defect recognition, and will be hopeful to become an important complement of the current technologies of defect inspection for the highly reflective surface, which indicates that this project have significant research value.

高反射精密回转类零部件的应用越来越广泛，但该类零部件高反射回转面的凹坑、划痕、擦伤、异物等缺陷严重影响产品的使用性能、寿命和安全，缺陷的准确检测至关重要，但当前的检测技术大多基于视觉传感器的方法，在识别准确性、效率和通用性之间不能兼顾，应用仍受制约。本项目提出基于数字微镜器件(DMD)结合复合传感器(视觉-激光传感器)的高反射回转面缺陷定量检测新方法，研究包括：研制基于DMD的新型自适应条纹照明系统，提高缺陷视觉检测的灵敏度及对曲率不同回转面的自适应能力；提出基于图像差影和主动轮廓模型的方法，实现图像缺陷区域边界快速准确提取；研究基于激光传感器的缺陷定量评价及分类方法，最终建立高反射回转面缺陷定量检测新方法，为进一步研制相应的缺陷检测设备奠定理论和技术基础。这种新的缺陷检测方法具有识别准确性高、效率高和通用性强等突出优点，有望成为现有高反射表面缺陷检测方法的重要补充，具有重要的研究价值。

高反射精密回转类零部件在其生产过程中可能会产生凹坑、划痕、擦伤、异物等表面缺陷，会严重影响产品使用性能、寿命和安全。因此，高反射回转面缺陷检测技术成为亟待解决的工程难题。本项目在前期研究基础上进行进一步的探索，首先设计了检测光照方法和缺陷提取算法，然后进行缺陷定量评价及分类方法设计，最终建立一种准确度高、性能优越的高反射回转面缺陷检测方法。.本项目主要由四部分构成：（1）缺陷检测光照方法设计：高反射回转面出现的强反光亮斑和易反射周围物体的像淹没了缺陷信息，以及不同曲率表面对光照的影响，因而，高质量清晰的缺陷图像很难采集。本项目设计了基于微镜器件的自适应条纹照明系统，适应不同曲率的高反射回转面缺陷视觉检测的图像采集。（2）表面缺陷视觉检测平台搭建：在自适应条纹照明的基础上，结合视觉传感器和机械系统等，搭建了高反射回转面缺陷检测图像的采集平台，该平台可实现缺陷图像快速采集、存储及处理等。（3）缺陷提取算法设计：本项目提出了基于阈值自适应调整的缺陷提取算法，该算法在构造兼顾空域和值域信息的滤波方式的基础上，对原始图像进行滤波预处理，其目的是保护缺陷目标边缘信息；其次，以高斯函数的一阶导数构建Canny最优边缘检测器，完成对图像的行列微分滤波及梯度非极大值抑制操作，以达到图像初步提取；然后，利用最大类间方差法提供的优化阈值，并结合形态学操作，实现缺陷目标最终提取。结果表明该方法能较好实现高反射回转面缺陷边界提取。（4）缺陷参数定量评价及分类方法设计：本项目在提取表面缺陷边界之后，利用高精度光学传感器对提取的缺陷区域进行精密测量，实现缺陷区域三维可视化表示，并进而可实现缺陷几何参数计算及分类。.本项目研究完成了计划的研究内容，重点研究了检测光照方法，以及缺陷提取算法，实现了对高反射回转面缺陷的定量检测，并提高了其缺陷识别的准确率，效率和通用性，有一定的科学价值和工程实用价值。