【作者】 张强；

【导师】 洪伟；

【作者基本信息】 吉林大学 ， 工程硕士（专业学位）， 2018， 硕士

【摘要】 目前,伴随着科学和经济的不断发展,双目立体视觉技术凭借其设备成本低,精度高,非接触等优点越来越广泛地应用于工业检测、航空航天、生物医学等各个领域。本文针对目前工业生产以及工业检测中所出现的问题对双目立体视觉的关键技术进行了研究,并提出了一种基于双目立体视觉的桶箱与桶盖识别和定位方法。本文主要完成了以下工作:第一,本文首先针对在检测时桶盖的表面涂层会反光导致特征点的检测不准的问题,设计出了一种基于双立方插值法的光照补偿算法,该算法有利于提高后续左右两幅图像在图像处理算法中参数或阈值选取的适用性和提取左右两幅图像特征点的准确性。第二,对本文中的目标工件进行了分析,主要针对工件的表面以及工件的外形识别特征。本文根据桶箱特征点的特性设计了基于连通区域的特征点轮廓提取方法,根据桶盖特征点的特性设计了基于Hough圆和Hough直线变换的特征点提取方法。第三,深入地研究了图像处理技术。根据原始图像的特性,经多种算法对比研究后,确定了合适的图像处理方案,包括桶箱和桶盖图像的滤波,对桶盖进行的边缘检测以及对桶箱进行的二值化。最后利用三维重建原理完成了对特征点的三维坐标计算。第四,根据本文的应用环境自行研发了双目立体视觉硬件系统,并根据检测环境完成了双目立体视觉系统的标定,对采集到的图像进行了立体校正。最后,本文开发了一套基于OPENCV与MFC相结合应用软件,成功实现了对二维特征点的求取以及三维坐标的计算,并且通过实验对三维坐标的准确性和稳定性进行了验证。实验结果表明,可以通过双目立体视觉完成对工件特征点的三维坐标计算,并且具有较高的准确性和稳定性。更多还原

【Abstract】 At present,with the continuous development of science and economy,binocular stereo vision technology is increasingly used in various fields such as industrial inspection,aerospace,and biomedicine due to its low equipment cost,high precision,and non-contact advantages.In this paper,the key technologies of binocular stereo vision are studied in the current problems in industrial production and industrial inspection,and a method for bin box and bin lid identification and positioning based on binocular stereo vision is proposed.This mainly completes the following work:First of all,in this paper,we first address the problem that the detection of feature points is inaccurate due to the coating on the surface of the object,designed a light compensation algorithm based on the bicubic interpolation method.The algorithm is helpful to improve the applicability of the parameters of the subsequent two images in the image processing algorithm or the threshold selection and the accuracy of extracting the left and right image feature points.Second,analyze the target artifacts in this paper,mainly for the surface of the workpiece and the shape recognition features of the workpiece.In this paper,based on the characteristics of the barrel box feature points,a feature point contour extraction method based on connected area is designed and a feature point extraction method based on Hough circle and linear transformation is designed according to the characteristics of the barrel cover feature points.Third,in-depth study of image processing technology.According to the characteristics of the original image,after a comparative study of various algorithms,a suitable image processing scheme was determined.Including the filter of the barrel box and the lid image,the edge detection of the lid and binarization of the barrel.Finally,the three-dimensional coordinates of the feature points are calculated using the principle of three-dimensional reconstruction.Fourth,according to the application environment of this paper,we have developed a binocular stereo vision hardware system.According to the detection environment,the binocular stereo vision system is calibrated and stereo images are corrected for the collected images.Finally,this paper developed a set of application software based on OPENCV and MFC,successfully achieved the calculation of two-dimensional feature points and the calculation of three-dimensional coordinates,verified the accuracy and stability of three-dimensional coordinates through experiments.The experimental results show that the three-dimensional coordinates of the workpiece feature points can be calculated by binocular stereo vision,and have high accuracy and stability.更多还原