可见光热反射成像测温中图像配准技术研究

Weak Signal Measurement in Thermoreflectance Temperature Testing Based on Stochastic Resonance

为了消除被测件与测温仪器相对位移对光热反射成像测温结果的影响,研究了亚像素量级的图像配准算法,配合以基于压电陶瓷的三轴纳米位移台,实现了对位移的实时修正。采用傅里叶变换、频域卷积及sinc函数近似等方法完成了能够实现亚像素级图像配准的算法,实时计算被测件与测温仪器的相对位移;搭建了一套光热反射成像测温实验装置,采用了三轴压电平台实时调节被测件的位置;试验结果显示,当被测件温度从20℃升高至50℃时,被测件在水平方向上发生了明显的位移,采用实验装置能有效修正该位置偏移。对典型的GaN HEMTs器件进行了测试,测试结果与国外检测机构的结果基本一致,证明了图像配准方法的有效性。

To eliminate the impact of relative drift between DUT(device under test)and test instruments on thermoreflectance thermography,the image registration algorithm is studied,cooperated with a piezoelectric three axis nano-scale position stage,real time correction of drift is realized.Sub-pixel image registration algorithm is realized by Fourier transform,convolution in frequency domain and sinc function approximation,it can calculate the relative drift in real time;a test setup for thermoreflectance thermography is build,the position of the DUT is adjusted in real time by the three axis nano-scale position stage;the results show that,when the temperature of DUT rises from 20℃to 50℃,there is an obvious displacement in the horizontal direction of the DUT,and this displacement can be correction by the test setup.A typical GaN HEMTs device is tested,the results are consistent with those measured by foreign testing institutions,and the image registration method is proved to be effective.