结冰冰形是结冰风洞试验中关注的要素之一,针对其冰形测量问题,提出了一种基于变分图像分割技术的非接触测量方法。基本思路是对翼型结冰俯视图像进行图像分割,将获得的曲线标定后与翼型数模结合,得到最终结冰冰形。为了排除结冰图像中不相干物体和因素的干扰,更精确获取结冰冰形,提出了选择分割的思想,构造了区域特征函数,建立了新的能量函数。采用所提基于变分分割模型的测量方法,对结冰风洞试验冰形进行了测量,并定量分析了方法的误差,测试了方法的抗噪声能力。结果表明该方法对结冰冰形的测量是准确可行的,且具有较高的精度,另外,其对噪声具有鲁棒性。所提方法可推广至其他与物体形状相关的测量中。
Ice shape is one of the key elements in an icing wind tunnel test.In order to measure ice shape efficiently,a noncontact measurement approach is proposed based on a variational segmentation model.The primary work of the proposed approach is to obtain shape curves by segmenting the overhead view image of airfoil.With exact transformation of the curves,the ice shape can be determined.In the segmentation procedure,a novel energy function is presented by using a constructed region characteristic function at the beginning.And then the selective segmentation which is for the purpose of removing the interference of unrelated objects or factors and getting a more accurate result can be implemented by minimizing the energy.The proposed approach has been successfully applied to ice shape measurement,which indicates that it is viable.And the quantitative result of error analysis demonstrates that the proposed measurement approach is highly accurate.Moreover,the measurement test on noisy image shows that it is robust to noise.The proposed approach can be easily applied to other fields which are related to shape measurement.