基于自适应参数估计多元非线性函数拟合的散斑图像亚像素级配准关键技术研究

Digital Speckle Correlation Method(DSCM) is a common method to image registration employed in experimental mechanics, geodesic survey and guidance fields, however, when the image quality is relatively low or deformed in large scale, the matching resolution could hardly reach the sub-pixel level. Combining the pre-studies over the DSCM and the future trend of the image registration techniques, this project is aimed to make detailed research of the several core issues related to DSCM, such as the modelling, mode recognition and interpolation theory, and propose the according solutions as well, to be more specifically, these solutions are mainly deliberated as: (1) with the multivariate nonlinear function as the research target, proposing the novel interpolation model for the speckle image sub-pixel registration; (2) by using the low-resolution image registration mesh data to estimate the key parameters of the high-resolution fitting function, an adaptive parameter estimated multivariate nonlinear measure function interpolation is proposed; (3) by studying the speckle image registration system, providing the reliable feedback data of adaptive estimation for the measurement function. This project is aimed to promote the development of the image registration techniques and results in the substantial improvement of the registration resolution without changing the image hardware acquisition.

数字散斑相关方法(DSCM)是在实验力学、大地测量、制导等领域中常用的一种图像配准方法，然而当目标图像质量较差或有较大形变时，其配准精度很难达到亚像素级要求。本项目拟结合申请人在DSCM方面的前期研究工作以及图像配准技术的发展趋势，分别从识别模型的构建、模式识别和插值理论等方面对DSCM的几个重要问题进行研究并提出解决方法，主要体现为：(1)提出散斑图像亚像素级配准插值函数的新模型，以多元非线性函数为研究目标；(2)提出多元非线性亚像素级配准拟合函数关键参数的自适应估计方法，利用容易计算的低精度配准测度网格数据，用数理统计的方法估计高精度测度函数关键参数；(3)散斑图像配准精度评价体系研究，为测度函数参数自适应估计提供可信的反馈数据。本项目的研究成果将有力推动图像配准技术的研究，在不改变图像采集硬件的情况下使配准精度获得明显且有效的提高。

数字散斑相关方法(DSCM)是在实验力学、大地测量、制导等领域中常用的一种图像配准方法，然而当目标图像质量较差或有较大形变时，其配准精度很难达到亚像素级要求。本项目组成员分别从识别模型的构建、模式识别和插值理论等方面对DSCM的几个重要问题进行研究并提出解决方法，主要研究内容和成果有：.(1)提出了散斑图像亚像素级配准插值函数的新模型，最终确定为特定的多元非线性函数；(2)提出多元非线性亚像素级配准拟合函数关键参数的自适应估计方法，利用容易计算的低精度配准测度网格数据，用数理统计的方法估计高精度测度函数关键参数；(3)完成散斑图像配准精度评价体系研究，为测度函数参数自适应估计提供可信的反馈数据。本项目的研究成果将有力推动图像配准技术的研究，在不改变图像采集硬件的情况下使配准精度获得明显且有效的提高。