提出了一种基于显微图像的微小尺寸零件复杂边缘识别算法。该算法通过对不同加工方法的边缘过渡区进行提取,建立边缘过渡区的多项式回归模型,然后从多项式模型中根据加工特性和零件的实际应用,利用求导的方法确定边缘点精确位置。实验结果表明:精研磨零级1 mm标准量块的测量误差为0.395 μm。通过对精研磨、冲压、线切割3种常见微细加工方法建模分析得出:加工特性对微小型结构件边缘区域影响较大,边缘精确识别时应考虑加工特性的影响。该算法在思想上考虑了实际加工特性的影响,算法上加入了统计学方法,通过建立过渡区数学模型使边缘检测结果达到亚像素级;而且识别的单像素宽,其连续的边缘能进一步满足精密测量的需要。

In order to realize the edge recognition for micro-accessory, a new complex edge recognition algorithm for micro-accessory is proposed based on the process characteristics of micro-accessory. By extracting the edge transitional region of different micro-accessories,and the polynomial regression mathematical models of transitional region are established.Then, the accurate position is obtained by process characteristics from the mathematics model. The experimental results indicate that the measurement error of 1 mm standard gauge block is 0.395 μm. After analysing the model for micro-fabrication,it comes to a conclusion that the process characteristics have greater influences on the edge of micro-accessory, so the edge recognition of micro-accessory should consider the influences of process characteristics. Proposed algorithm has considered the effects by processing characteristics and has induced a statistical scheme,so that the edge recognition result can reach to the sub-pixel level by establishing the mathematics model of transitional region. The algorithm also can offer a wide single-pixel and continuous edge,which can meet the precision measurement requirements.