高斯预滤波对数字体图像相关测量的影响  

Effect of Gaussian Prefiltering on Digital Volume Correlation Measurement

作　　者：邹翔 潘兵[1] 王延珺 刘晨[2] 许蔚[3] Zou Xiang;Pan Bing;Wang Yanjun;Liu Chen;Xu Wei(Institute of Solid Mechanics,Beihang University,Beijing 100191,China;Institute of Chemical Materials,China Academy of Engineering Physics,Mianyang,Sichuan 621999,China;School of Architecture,Kunming University of Science and Technology,Kunming,Yunnan 650500,China)

机构地区：[1]北京航空航天大学固体力学研究所,北京100191 [2]中国工程物理研究院化工材料研究所,四川绵阳621999 [3]昆明理工大学建筑工程学院,云南昆明650500

出　　处：《光学学报》2021年第15期132-142,共11页Acta Optica Sinica

基　　金：国家自然科学基金(11925202,11872009,11762009);北京航空航天大学博士研究生卓越学术基金。

摘　　要：图像噪声会严重影响数字体图像相关(DVC)方法的测量精度,因此,通过数值模拟平移实验、真实重扫描实验和压缩实验研究了图像噪声及高斯预滤波对DVC位移测量结果的影响。实验结果表明,高斯预滤波能显著降低图像的噪声水平,从而减小由噪声引起的系统误差,但对插值误差的影响较小,原因是高斯预滤波会降低图像的灰度梯度。因此,高斯预滤波可以显著提高经典前向累加牛顿-拉弗森算法的精度,但对反向组合高斯-牛顿算法的精度影响不大。此外,高斯预滤波处理可提升2种算法中各计算点的相关系数,得到匹配度更高的子体块。Image noise will seriously affect the measurement accuracy of the digital volume correlation(DVC)method.Therefore,this paper studies the effect of image noise and Gaussian prefiltering on the DVC displacements measurement results through numerical simulation translation experiments,real rescan and compression experiments.The experimental results show that,Gaussian prefiltering can significantly reduce the noise level of the images,and thus mitigates the noise-induced bias error,but has little effect on the interpolation error.The reason is that the Gaussian prefiltering will reduce the gray gradient of the image.Therefore,Gaussian prefiltering can significantly improve the accuracy of the classic forward additive Newton-Raphson algorithm.But it has little effect on the accuracy of the inverse compositional Gauss-Newton algorithm.Further,the Gaussian prefiltering is used to reduce image noise,which can improve the correlation coefficient of each calculation point in the 2 algorithms,and obtain a sub-volume with a higher degree of matching.

关 键 词：图像处理 数字体图像相关方法 图像噪声 高斯预滤波 位移测量 

分 类 号：O348.1[理学—固体力学]