摘要
同步相移干涉技术通过同时获取多幅相移干涉图,实现动态波面的相移干涉测量。为了保证测量精度,在实施相移算法前,需要对相移干涉图进行准确的位置配准。提出了一种结合加速稳健特征(SURF)提取算法和随机采样一致性(RANSAC)算法的多步干涉图位置配准方法,通过配准具有一定特征的干涉背景图,获得图像之间的变换关系并作用于相移干涉图,实现同步相移干涉图位置配准。仿真分析表明,该方法在光强不均的情况下对平移和旋转变换均具有较好的校准效果;在含有高斯白噪声的情况下,仍能进行配准获得变换矩阵,完成相位恢复。将此方法应用于四步同步相移显微干涉测量仪对标准平面进行测量,得到测量结果峰谷值为23.2520nm,均方根值为2.3149nm。
Simultaneous phase-shifting interferometry obtains multiple phase-shifting interferograms simultaneously to realize phase-shifting interferometry of the dynamic wavefront.In order to ensure the measurement accuracy,it is necessary to perform accurate position registration of the phase-shifting interferograms before implementation of the phase shifting algorithm.A position registration method based on the speed up robust features(SURF)algorithm and the random sample consensus(RANSAC)algorithm is proposed.Through the registration of background image with certain characteristics,the transform relation between corresponding images is obtained,which is used for the phase-shifting interferogram to realize position registration of the simultaneous phase-shifting interferograms.The simulation results show that the method is effective on the translation and rotation transforms in the case of nonuniform light intensity.The method can also obtain transform matrix by registration to achieve phase recovery under the white Gaussian noise.The method is applied to measure the standard plane with a four-step simultaneous phaseshifting micro-interferometer,and the peak-to-valley value and root mean square value are 23.2520 nm and2.3149 nm,respectively.
作者
靳京京
卢文龙
郭小庭
刘晓军
Jin Jingjing;Lu Wenlong;Guo Xiaoting;Liu Xiaojun(School of Mechanical Science & Engineering, Huazhong University of Science and Technology Wuhan, Hubei 430074, Chin)
出处
《光学学报》
EI
CAS
CSCD
北大核心
2017年第10期169-176,共8页
Acta Optica Sinica
基金
国家自然科学基金(51475192)
湖北省科研条件与资源研究开发项目(2015BCE057)
关键词
测量
同步相移干涉
位置配准
加速稳健特征提取算法
随机采样一致性算法
measurement
simultaneous phase-shifting interferometry
position registration
speed up robustfeatures algorithm
random sample consensus algorithm