Phase noise filtering and phase unwrapping method based on unscented Kalman filter

This paper presents a new phase unwrapping algorithm based on the unscented Kalman filter（UKF） for synthetic aperture radar（SAR） interferometry.This method is the result of combining an UKF with path-following strategy and an omni-directional local phase slope estimator.This technique performs simultaneously noise filtering and phase unwrapping along the high-quality region to the low-quality region,which is also able to avoid going directly through the noisy regions.In addition,phase slope is estimated directly from the sample frequency spectrum of the complex interferogram,by which the underestimation of phase slope is overcome.Simulation and real data processing results validate the effectiveness of the proposed method,and show a significant improvement with respect to the extended Kalman filtering（EKF） algorithm and some conventional phase unwrapping algorithms in some situations.

THE COHERENCE COEFFICIENT MAP AND RESIDUE GUIDED LEAST SQUARE PHASE UNWRAPPING ALGORITHM

The generation of high-resolution DEM from interferometric SAR has resulted in the need for accurate and efficient methods of 2-dimensional phase unwrapping. In this paper, we give a brief description of the mathematical base of phase unwrapping, and a detailed description of the unweighted and weighted least square phase unwrapping algorithm.Then our algorithm combining with the weighted least square phase unwrapping guided by the branch-cuts derived from Goldstein’ s algorithm and coherence coefficient map derived from the INSAR data is provided. In our experiment we write subroutines of the Goldstein’s branch-cut algorithm,unweighted and weighted least square phase unwrapping algorithm as well as our algorithm,and construct a small experiment system to resolve the phase unwrapping problem. Finally we test our algorithm on some INSAR data. The result shows that our approach can obtain unwrapped phase correctly and efficiently.

Enhanced multi-baseline unscented Kalman filtering phase unwrapping algorithm

This paper presents an enhanced multi-baseline phase unwrapping algorithm by combining an unscented Kalman filter with an enhanced joint phase gradient estimator based on the amended matrix pencil model, and an optimal path-following strategy based on phase quality estimate function. The enhanced joint phase gradient estimator can accurately and effectively extract the phase gradient information of wrapped pixels from noisy interferograms, which greatly increases the performances of the proposed method. The optimal path-following strategy ensures that the proposed algorithm simultaneously performs noise suppression and phase unwrapping along the pixels with high-reliance to the pixels with low-reliance. Accordingly, the proposed algorithm can be predicted to obtain better results, with respect to some other algorithms, as will be demonstrated by the results obtained from synthetic data.

Quality Guided Phase Unwrapping for Fringe Projection Profilometry

Fringe projection technique is a non-contact, full field 3-D shape measurement method. The object depth information is recorded in one or several deformed fringe patterns. The phase-shifting algorithm or the Fourier transform method can be used to extract the wrapped phase data. A phase unwrapping process is then applied to retrieve a continuous phase distribution, which represents the surface profile of the test object. In this paper, a quality-guided phase unwrapping approach is incorporated and two novel phase quality evaluation methods are proposed to facilitate the phase unwrapping process.

Multi-baseline extended particle filtering phase unwrapping algorithm based on amended matrix pencil model and quantized path-following strategy

This paper proposes a new multi-baseline extended particle filtering phase unwrapping algorithm which combines an extended particle filter with an amended matrix pencil model and a quantized path-following strategy. The contributions to multibaseline synthetic aperture radar(SAR) interferometry are as follows: a new recursive multi-baseline phase unwrapping model based on an extended particle filter is built, and the amended matrix pencil model is used to acquire phase gradient information with a higher precision and lower computational cost, and the quantized path-following strategy is introduced to guide the proposed phase unwrapping procedure to efficiently unwrap wrapped phase image along the paths routed by a phase derivative variance map.

Investigation on Three Types of Residues in Phase-unwrapping

In phase unwrapping, how to control the residues and cut lines is critical to unwrap ill-state wrapped phase map successfully. Some new concepts in phase unwrapping field are proposed. Firstly, the residues in three types, i.e., singular residues, normal residues and virtual residues, are classified, which leads to some rational unwrapping paths. Secondly, some differences between cut-line and curved line are analyzed. And finally, experiment is carried out to compare our new enhanced Goldstein algorithm with the original Goldstein algorithm.

InSAR Kalman Filter Phase Unwrapping Algorithm Based on SRTM DEM

PU （phase unwrapping） is the key step and important problem in DEM （digital elevation model） extraction and the measurement of surface deformation of InSAR （Interferometric synthetic aperture radar）. The CKFPUA （conventional Kalman filter phase unwrapping algorithm） can obtain reliable results in the flat terrain areas, but it caused error transmission not making the accurate inversion of surface deformation information in the steep terrain. Considering this situation, so it needs to introduce topographic information for guiding phase unwrapping. Here the 90 m resolution DEM data have been used and it is obtained by SRTM （shuttle radar topography mission） measured jointly by NASA （National Aeronautics and Space Administration） and NIMA （National Imaging Mapping Agency） of U.S. Department of Defense. This paper presents a SD-KFPUA （Kalman filter phase unwrapping algorithm） based on SRTM DEM. With SRTM DEM directing InSAR image to implement phase unwrapping, the speed and accuracy are improved. By analyzing with the conventional Kalman filter phase unwrapping algorithms, it is shown that the proposed method can achieve good results in particular to improve unwrapping accuracy in the low coherence region.

Phase Unwrapping in Magnetic Resonance Elastography

Phase Unwrapping (PU) is an ill-posed problem in Magnetic Resonance Elastography (MRE). The phase information is not usable until the phases are retrieved by using PU algorithms. In this present study, we attempt to determine the ideal PU method for MRE using both phantom and volunteer psoas major (PM) muscle images. All the MRE experiments were carried out in Philips MRI (Achieva 3.0 T, Best, The Netherlands). A multi-echo gradient-echo MRE pulse sequence was employed and the four PU methods were considered based on their easy user platform. They are namely, Minimum Discontinuity (MD), Laplacian-Based Estimate (LBE), Region Growing (RG) and Dilate-Erode (DE) Propagate. Phantom images were successfully unwrapped by all four methods, whereas MD and LBE could only unwrap PM muscle images properly. RG and DE failed to unwrap the PM muscle images.

UTILIZING PHASE RETARD INTEGRATION METHOD OF FLOW BIREFRINGENCE TO ANALYSE THE FLOW WITH SYMMETRIC PLANE

The formulas which are suitable to birefringent medium with symmetric plane are derived by means of phase retard integration.We have adopted this concept to the axisymmetric problems and deduced some useful formulas for these cases.As a practical application,the strain rate analysis of flow in a diverging or a converging vessel is illustrated at the end of this paper.

DEFORMATION MEASUREMENT USING DUAL-FREQUENCY PROJECTION GRATING PHASE-SHIFT PROFILOMETRY

2π phase ambiguity problem is very important in phase measurement when a deformed object has a large out of plane displacement. The dual-frequency projection grating phaseshifting profilometry （PSP） can be used to solve such an issue. In the measurement, two properchosen frequency gratings are utilized to synthesize an equivalent wavelength grating which ensures the computed phase in a principal phase range. Thus, the error caused by the phase unwrapping process with the conventional phase reconstruct algorithm can be eliminated. Finally, experimental result of a specimen with large plastic deformation is given to prove that the proposed method is effective to handle the phase discontinuity.

A new method of weight choice in InSAR least squares unwrapping

The de-coherence phenomena such as Low-SNR radar signal, shadows and layover caused by topography, etc. , causing phase data discontinuity, makes the result of unwrapping phase inaccuracy or even completely wrong. Based on the analysis of influencing factors to weight choice, this thesis develops a new method to choose the weights based on the measure of the confidence in the frequency domain. Experiments show that it could overcome the defect of sub-estimate to the slope of least squares method very well, which has a better rationale, stability and performance.

A new method of weighted choice in InSAR Least Squares unwrapping

The decorelation phenomena such as Low-SNR radar signal, shadows and layover caused by topography etc, causes phase data discontinuous and makes the result of unwrapping phase inaccurate or completely wrong. Based on the analysis of influencing factors to the weight selection, this paper develops a new method to choose the weights based on the measurement of confidence in frequency domain. Results show that it is more precise and robust than other methods, and can make up for the defect of sub-estimate to the slope of least squares method.

Deep learning spatial phase unwrapping:a comparative review

Phase unwrapping is an indispensable step for many optical imaging and metrology techniques.The rapid development of deep learning has brought ideas to phase unwrapping.In the past four years,various phase dataset generation methods and deep-learning-involved spatial phase unwrapping methods have emerged quickly.However,these methods were proposed and analyzed individually,using different strategies,neural networks,and datasets,and applied to different scenarios.It is thus necessary to do a detailed comparison of these deep-learning-involved methods and the traditional methods in the same context.We first divide the phase dataset generation methods into random matrix enlargement,Gauss matrix superposition,and Zernike polynomials superposition,and then divide the deep-learning-involved phase unwrapping methods into deep-learning-performed regression,deep-learning-performed wrap count,and deep-learningassisted denoising.For the phase dataset generation methods,the richness of the datasets and the generalization capabilities of the trained networks are compared in detail.In addition,the deep-learning-involved methods are analyzed and compared with the traditional methods in ideal,noisy,discontinuous,and aliasing cases.Finally,we give suggestions on the best methods for different situations and propose the potential development direction for the dataset generation method,neural network structure,generalization ability enhancement,and neural network training strategy for the deep-learning-involved spatial phase unwrapping methods.

Photoelastic Stress Analysis by Use of Hybrid Technique and Fringe Phase Shifting Method

Photoelastic fringe patterns for stress analysis are investigated by use of hybrid technique and fringe phase shift method. The first one is a hybrid method which combines the conformal mapping technique and measured data away from the edge of a geometric discontinuity. Photoelastic data are hybridized with complex variable/mapping techniques to calculate photoelastic stress-field around a circular hole or an elliptical hole in plates under uniaxial tensile loading. This method determines full-field stresses in perforated finite tensile plates containing either a circular or an elliptical hole. The second one is a fringe phase shift method to separate isochromatics and isoclinics from photoelastic fringes of a circular disk under diametric compression by use of phase shift method. Digitally determined isochromatics and isoclinics are agreed well with those of manual measurements.

融合长短基线的干涉图解缠错误自动改正

合成孔径雷达干涉测量(interferometric synthetic aperture radar,InSAR)技术因具有全天时、全天候、大范围的独特优势,已被广泛应用于地球科学研究、自然灾害监测和环境管理等领域。然而,受地表失相干、干涉条纹密集等因素影响,InSAR数据处理关键的步骤——干涉图相位解缠常出现错误,进而影响高程反演和地表形变监测精度。为此,文章提出了一种融合长短基线干涉相位闭合环的解缠错误自动探测与改正方法,并利用湖南省-江西省高植被覆盖区的Sentinel-1雷达影像开展了算法验证。实验结果表明,该方法可有效地改正干涉相位解缠错误,这对InSAR技术提升高程反演和形变监测精度具有重要意义。

MCJ-UNet:一种双/多通道联合InSAR相位解缠网络

干涉合成孔径雷达(InSAR)可实现地表高程的高效获取,在地形测绘中应用广泛。双/多通道InSAR技术可借助不同通道(基线、频点)的高程模糊度差异,解决相位欠采样问题,完成高程陡变区域的干涉相位解缠,实现InSAR技术在测绘困难区域的有效应用。该文即面向高效高精度相位解缠需求,利用深度学习这一有力工具,结合不同通道的相位特征及相互约束关系,提出了一种双/多通道联合干涉相位解缠网络:Multi-Channel-Joint-UNet(MCJ-UNet)。该网络的构建以双通道(双频、双基线)InSAR为基本观测构型,并可实现向多通道构型的扩展,其构建的核心思路主要包括3点:首先,将干涉相位解缠中的模糊数估计问题转化为语义分割问题,并采用UNet网络完成分割处理;其次,引入挤压激励模块(SE)动态调整信息权重,以增强网络不同通道对其所需信息的感知能力;最后,利用多通道联合约束下的相位残差优化损失函数,实现网络调谐。此外,为避免语义分割结果的边缘细节误差对解缠效果的影响,该文还提出了一种基于多通道联合约束的解缠误差自修正方法,以保证解缠质量。模拟地形仿真数据、真实地形仿真数据以及TerraSAR-X实测数据验证了所提方法的有效性。

改进的三频四步相移面结构光三维重建算法

针对面结构光中传统三频四步相移算法投影图片过多、效率较慢这一问题,提出了改进的三频四步相移面结构光重建算法。改进的算法需要3个高频率的正弦条纹图,但所需投影的条纹数量从传统算法的12张减少到了7张,最高频率的条纹图4张,次高频和最低条纹图分别为2张和1张,投影效率比传统算法提高了41.7%。介绍了改进算法的原理,将最高频率4张图所计算的包裹相位用来三维重建,其余两个频率条纹图计算的包裹相位用来进行相位展开,理论上该方法与传统算法精度一致。最后,通过实验验证了该算法的重建精度以及对具有复杂表面物体的三维重建能力。实验结果表明改进的面结构光算法在提高投影效率的同时与传统算法保持着一致的精度。

基于改进U-Net网络的相位解包裹技术研究

提出了一种结合深度学习的空间相位解包裹方法,采用基于改进U-Net网络的编码器-解码器架构,同时加入包含双向长短期记忆网络(BILSTM)的CBiLSTM模块,并且结合注意力机制,避免了典型卷积神经网络学习全局空间依赖关系的固有缺陷的同时增强了深度学习模型对相位解包裹任务中的关键信息的关注能力。通过大量的模拟数据,验证了文中方法在严重噪声(SNR=0)、不连续条件和混叠条件下的鲁棒性,在以上三种情况下,同其他深度学习网络模型进行对比,文中所提出的网络模型的归一化均方根误差(NRMSE)分别为0.75%、1.81%和1.68%;结构相似性指数(SSIM)分别为0.98、0.92和0.94;峰值信噪比(PSNR)分别为40.87、32.56、37.38;同时计算时间显著减少,适合应用到需要快速准确的空间相位解包裹任务中去。通过实际测量数据,验证了文中提出网络模型的可行性。该研究将双向长短期记忆网络(BILSTM)和注意力机制同时引入光学相位解包裹问题中,为解决复杂相位场的解包裹提供了新的思路和方案。

基于相位质量融合估计与信息滤波的相位解缠方法

L波段差分干涉SAR卫星(陆探一号,LT-1)是我国首个L频段全极化民用SAR卫星星座,该组卫星将长期服务于自然资源监测监管,DEM生产是其主要任务之一。众所周知,相位解缠是影响DEM反演精度的关键因素之一,然而,高噪声及大梯度变化区域解缠结果精度低一直是相位解缠的瓶颈难题。对此,本文提出了一种相位质量融合估计策略引导的迭代信息滤波与解缠方法。首先,对干涉图像素的相干性、相位梯度、噪声分量和干涉条纹边缘进行估计与拟合,进而指导解缠路径布设,提高相位解缠结果的准确性;然后,采用中心差分信息滤波器实现解缠相位的准实时状态估计,进而提高解缠方法的稳健性;最后,通过TerraSAR-X/TanDEM-X双星及LT-1A/B双星SAR数据进行试验验证。结果表明,本文方法精度较统计费用流算法提高了约65.3%;较无迹信息滤波解缠算法提高了约13.6%。此外,试验结果不仅验证了本文方法的有效性,而且证明LT-1双星系统可达到高精度地形重建的预期效果,能为我国后续多云多雨区的DEM产品生产提供高精度的数据源。