中文摘要：

与仅仅承受弹性变形时不同，当金属试件承受塑性变形或发生疲劳损伤时，内部微观晶粒滑移位错等会导致表面粗糙化。此时其镜面反射能力减弱，漫反射能力增强。据此提出了一种检测试件弹塑性边界和塑性变形程度的新方法。承受均匀拉伸的金属试件被普通光源沿斜向照明，载荷按预定方式逐渐增加。用CCD摄像机沿试件法线方向记录板材试件表面的亮度变化，这些亮度分布摄像机可按一定频率采集以实时图像形式记录下来。我们提出了下述简单直观的方法来分析图像数据。根据实际情况适当选择一个闽值，定义图像中灰度高于闽值的点为亮点，取参数BI为亮点个数的倒数。试验发现在弹性段参数BI基本维持定值；在试件发生屈服时由于表面粗糙化以致漫反射能力增强，图像的亮点数会显著增多，BI参数发生陡然下降的变化，在整个塑性段参数呈单调下降趋势。所以参数BI可以作为描述这一过程的特性参数。由BI的变化可以判断试件塑性变形的发生，也可以作为塑性变形程度的度量。典型的实验结果验证了这一方法，所得结果与用相关搜索方法得到的结果相比较，吻合得相当好。由于这种方法使用白光照明，对环境要求不高，操作过程简单，有望在工程实际中获得广泛的应用。本文还讨论了影响判别精度的闽值选取问题，通过对于所得数据的分析得出闽值选取可以按照误差理论中置信区间的方法来设置，并且置信区间较小时能得到比较好的结果。

英文摘要：

Roughness on the metal specimen polished surface will be induced when it yields plastic deform or fatigue damage for its interior microscopic crystal grain slip dislocation, and this is very different when it yields elastic deform. Its specula reflection weakened while scattering reflection reinforced. This paper presented a new technique to determine the elastic-plastic boundary and plastic strains on the surface of metal specimen. The metal specimen subjected uniform tension is projected by Slant incident white-light; loads increase as preseted. These distributions of the brightness can be recorded as real-time images by CCD camera. The following simple method we advanced is used to analysis the image data. According the practical situation, one threshold is chosen. The gray level which is greater that the threshold is defined the bright point. Define BI as the reciprocal value of the number of the bright points. In the experiments, BI keeps stable during the period of elasticity; while the material is yielding, the increase of the surface roughness leads to the stronger ability of the random reflections and the BI goes down sharply. BI keeps monotone decreasing during the period of plasticity. So BI Can be taken as a feature parameter. BI can not only be used as a judgment of the status hut also he the measuring of the plasticity. Typical experiments can verify the method. The conclusion matches the traditional correlation method very well. This method can he prospectively applied in the engineering with the merit of simple operation and low environmental condition. This paper also discusses a parameter which is very important to the accuracy of this method. Based on the analysis of the data, a method which is called confidence interval was developed to determine the threshold value. And also find that the better results can be obtained with the little of the confidence interval.