机器视觉技术在固体材料形貌检测中的应用

Application of machine vision technique in profile detection for solid material

提出了一种基于图像球面化算法的机器视觉技术,建立了球面、椭球面和马鞍面三种模型,并将该技术应用于固体材料微纳米尺度形貌检测中。首先通过数学推演方式建立了图像球面化模型,然后以直线图像为标准图对各种模型做投影成像模拟,根据模拟成像结果分析不同模型的成像特征,并解释固体材料像旋转现象的产生机理。光干涉仪的实际测量结果表明,固体材料的实际形貌与模拟结论相符。该方法只需观察固体材料上反射的投影物体的像即可完成形变检测,测量过程简单易行,且对测试环境要求较低,适合于工业检测。

A novel image spherizing algorithm-based machine vision technique for micro scale profile detection of solid materials is proposed. Hemisphere, ellipsoid and saddle-shape models are built. Firstly, a rigorous mathematical deduction on image spherizing models is described in detail, subsequently a straight line image is chosen as standard picture and the imaging simulated projection test for deformed solid material is performed. According to the simulated results, the characteristics of standard picture which projected on different models are analyzed and the mechanism of image twisting phenomenon is described. The experimental results obtained by interferometer show that the actual profile characteristics of deformed solid material is consistent with the theoretical analysis. In this method, rapid profile detection for solid material could be completed only by observing the reflected image of solid material even in harsh environment, which is appropriate for industrial inspection.