Robust Image Registration Method Research Articles

RDLR: A Robust Deep Learning-Based Image Registration Method for Pediatric Retinal Images.

Retinal diseases stand as a primary cause of childhood blindness. Analyzing the progression of these diseases requires close attention to lesion morphology and spatial information. Standard image registration methods fail to accurately reconstruct pediatric fundus images containing significant distortion and blurring. To address this challenge, we proposed a robust deep learning-based image registration method (RDLR). The method consisted of two modules: registration module (RM) and panoramic view module (PVM). RM effectively integrated global and local feature information and learned prior information related to the orientation of images. PVM was capable of reconstructing spatial information in panoramic images. Furthermore, as the registration model was trained on over 280,000 pediatric fundus images, we introduced a registration annotation automatic generation process coupled with a quality control module to ensure the reliability of training data. We compared the performance of RDLR to the other methods, including conventional registration pipeline (CRP), voxel morph (WM), generalizable image matcher (GIM), and self-supervised techniques (SS). RDLR achieved significantly higher registration accuracy (average Dice score of 0.948) than the other methods (ranging from 0.491 to 0.802). The resulting panoramic retinal maps reconstructed by RDLR also demonstrated substantially higher fidelity (average Dice score of 0.960) compared to the other methods (ranging from 0.720 to 0.783). Overall, the proposed method addressed key challenges in pediatric retinal imaging, providing an effective solution to enhance disease diagnosis. Our source code is available at https://github.com/wuwusky/RobustDeepLeraningRegistration .

Image registration method for mobile-device-based multispectral optical diagnostics for buildings.

The recent advances in mobile device hardware and software introduced new opportunities to perform more advanced computation and measurements for buildings. Multispectral optical imaging is a widely recognized technique for building diagnostics, as it is non-destructive, quick, and provides rich spatial-spectral information about the different surfaces for decisions related to building energy performance. However, such implementations require an accurate image fusion. The purpose of this study is to introduce a practical and robust image registration method for mobile-device-based multispectral imaging analysis for building diagnostics. Considering the complicated building geometries including walls, floors, ceilings, roofs, windows, and doors, a new approach based on planar homography is chosen for determining the feature points that are required to perform image fusion between different images. The results obtained are compared against the other available methods, which show that the current method provides multispectral images enhanced with accurate qualitative information, as long as the constraints are satisfied. The results are compared with the state-of-the-art methods. The possible impact of the multispectral imaging to the future of building diagnostics is also discussed.

Robust image registration using subspace method in Radon transform domain

PurposeThe purpose of this paper is to represent a robust image registration method to align noisy and deformed images in their Radon transform domain. Due to the limitation of imaging mechanism, the images are often highly noisy. Even worse, the objects in images have structural differences from time to time.Design/methodology/approachTo eliminate these degressions, the proposed method is equipped with subspace-based power spectrum analysis algorithm for rotation estimation and a new global median filter least square algorithm for displacement computation.FindingsExperiments on strongly noisy and degenerated images show that the proposed method exhibits better accuracy and robustness than phase correlation-based method. In addition, the method can also be applied to multi-modal registration, where the results are comparable to mutual information method but spending much less time.Originality/valueA robust image registration method is proposed, which has better performance than traditional methods.

Multiple Image Features-Based Retinal Image Registration Using Global and Local Geometric Structure Constraints

Retinal image registration is a key step in treating hypertension, diabetes and various retinal global diseases. In current methods of retinal image registration, they generally suffer from a lack of reliable features, missing true correspondences and geometric distortion. To address above problem, we propose a robust non-rigid retinal image registration method using multi-image features and dual constraints (i.e. the global and local geometric structure constraints). Our method contains the following main contributions. (i) A finite mixture model based on multi-feature is constructed for handling different types of image features. (ii) A combination of three features is substituted into the mixture model to improve the complementarities of different features. (iii) Dual constraints are proposed for ensuring the stability of the global and local structures of feature sets in the process of spatial transformation and updating. The performance of our method is evaluated by four main types of retinal images, which shows our method outperforms five state-of-the-art methods in most scenarios, especially when the retinal image has a large angle change.

Practical Algorithm Design for the FFT-based Robust Image Registration Method

Practical Algorithm Design for the FFT-based Robust Image Registration Method

Efficient and Robust Direct Image Registration Based on Joint Geometric and Photometric Lie Algebra.

This paper considers the joint geometric and photometric image registration problem. The inverse compositional (IC) algorithm and the efficient second-order minimization (ESM) algorithm are two typical efficient methods applied to the geometric registration problem. Their efficiency stems from the utilization of the group structure of geometric transformations. To allow for photometric variations, the dual IC algorithm (DIC) proposed by Bartoli performs joint geometric and photometric image registration by extending the IC algorithm. The group structures of both geometric and photometric transformations are exploited. Despite the robustness to large photometric variations, DIC is vulnerable to large geometric deformations. The ESM algorithm is extended by Silveira et al. to address photometric variations. In their approach, the photometric transformations are modeled in Euclidean space. Their approach is robust to relatively large geometric and photometric transformations; however, it is not efficient for large photometric variations. We propose a new efficient and robust image registration method by exploiting the non-Euclidean Lie group structure of joint geometric and photometric transformations for both grayscale and color images. The image registration is formulated as a nonlinear least squares problem. In our method, the geometric and photometric transformations are jointly parameterized by their corresponding Lie algebras. Based on this parameterization approach, the second-order approximation strategy of ESM is employed to optimize the joint geometric and photometric parameters. The error function in the nonlinear least squares problem is approximated by a second-order Taylor expansion with respect to joint geometric and photometric parameters without computing the Hessian matrix. For further efficiency, independent convergence criteria for geometric and photometric parameters are used in the iterative optimization process. The superiority of our proposed method over the previous methods, in terms of efficiency, accuracy, and robustness, is demonstrated through extensive experiments on synthetic and real data.

Geo-Positioning Accuracy Improvement of Multi-Mode GF-3 Satellite SAR Imagery Based on Error Sources Analysis.

The GaoFen-3 (GF-3) satellite is the only synthetic aperture radar (SAR) satellite in the High-Resolution Earth Observation System Project, which is the first C-band full-polarization SAR satellite in China. In this paper, we proposed some error sources-based weight strategies to improve the geometric performance of multi-mode GF-3 satellite SAR images without using ground control points (GCPs). To get enough tie points, a robust SAR image registration method and the SAR-features from accelerated segment test (SAR-FAST) method is used to achieve the image registration and tie point extraction. Then, the original position of these tie points in object-space is calculated with the help of the space intersection method. With the dataset clustered by the density-based spatial clustering of applications with noise (DBSCAN) algorithm, we undertake the block adjustment with a bias-compensated rational function model (RFM) aided to improve the geometric performance of these multi-mode GF-3 satellite SAR images. Different weight strategies are proposed to develop the normal equation matrix according to the error sources analysis of GF-3 satellite SAR images, and the preconditioned conjugate gradient (PCG) method is utilized to solve the normal equation. The experimental results indicate that our proposed method can improve the geometric positioning accuracy of GF-3 satellite SAR images within 2 pixels.

A robust image registration method based on total variation regularization under complex illumination changes

Background and objectiveImage registration is one of the fundamental and essential tasks for medical imaging and remote sensing applications. One of the most common challenges in this area is the presence of complex spatially varying intensity distortion in the images. The widely used similarity metrics, such as MI (Mutual Information), CC (Correlation Coefficient), SSD (Sum of Square Difference), SAD (Sum of Absolute Difference) and CR (Correlation Ratio), are not robust against this kind of distortion because stationarity assumption and the pixel-wise independence cannot be obeyed and captured by these metrics. MethodsIn this paper, we propose a new intensity-based method for simultaneous image registration and intensity correction. We assume that the registered moving image can be reconstructed by the reference image through a linear function that consists of multiplicative and additive coefficients. We also assume that the illumination changes in the images are spatially smooth in each region, so we use weighted Total Variation as a regularization term to estimate the aforesaid multiplicative and additive coefficients. Using weighted Total Variation leads to reduce the smoothness-effect on the coefficients across the edges and causes low level segmentation on the coefficients. For minimizing the reconstruction error, as a dissimilarity term, we use l1norm which is more robust against illumination change and non-Gaussian noises than the l2 norm. Primal-Dual method is used for solving the optimization problem. ResultsThe proposed method is applied to simulated as well as real-world data consisting of clinically 4-D Computed Tomography, retina, Digital Subtraction Angiography (DSA), and iris image pairs. Then, the comparisons are made to MI, CC, SSD, SAD and RC qualitatively and sometimes quantitatively. ConclusionsThe experiment results are demonstrating that the proposed method produces more accurate registration results than conventional methods.

SU‐E‐J‐122: The CBCT Dose Calculation Using a Patient Specific CBCT Number to Mass Density Conversion Curve Based On a Novel Image Registration and Organ Mapping Method in Head‐And‐Neck Radiation Therapy

Purpose:To develop a CBCT HU correction method using a patient specific HU to mass density conversion curve based on a novel image registration and organ mapping method for head‐and‐neck radiation therapy.Methods:There are three steps to generate a patient specific CBCT HU to mass density conversion curve. First, we developed a novel robust image registration method based on sparseness analysis to register the planning CT (PCT) and the CBCT. Second, a novel organ mapping method was developed to transfer the organs at risk (OAR) contours from the PCT to the CBCT and corresponding mean HU values of each OAR were measured in both the PCT and CBCT volumes. Third, a set of PCT and CBCT HU to mass density conversion curves were created based on the mean HU values of OARs and the corresponding mass density of the OAR in the PCT. Then, we compared our proposed conversion curve with the traditional Catphan phantom based CBCT HU to mass density calibration curve. Both curves were input into the treatment planning system (TPS) for dose calculation. Last, the PTV and OAR doses, DVH and dose distributions of CBCT plans are compared to the original treatment plan.Results:One head‐and‐neck cases which contained a pair of PCT and CBCT was used. The dose differences between the PCT and CBCT plans using the proposed method are −1.33% for the mean PTV, 0.06% for PTV D95%, and −0.56% for the left neck. The dose differences between plans of PCT and CBCT corrected using the CATPhan based method are −4.39% for mean PTV, 4.07% for PTV D95%, and −2.01% for the left neck.Conclusion:The proposed CBCT HU correction method achieves better agreement with the original treatment plan compared to the traditional CATPhan based calibration method.

Key technology of D-InSAR at X-band for monitoring land subsidence in mining area and its application

In theory, land subsidence measurement results with high accuracy can be obtained by using the Differential Interferometry Synthetic Aperture Radar (D-InSAR) at X-band. In practice, however, the measuring accuracy of D-InSAR at X-band has been seriously affected by some factors, e.g., decorrelation and high deformation gradient. In this work, the monitoring capability of D-InSAR for coal-mining subsidence is evaluated by using SAR data acquired by TerrraSAR-X system. The SAR image registration method for low coherence image pairs, the denoising phase filter for high noise level interferogram and atmospheric effects mitigation method are the key technical aspects which directly influence the measurement results of D-InSAR at X-band. Thus, a robust image registration method, an improved phase filter method and an atmospheric effects mitigation method are proposed in this paper. The proposed image registration method successfully achieves InSAR coregistration, while the amplitude cross-correlation cannot properly coregister low coherence SAR image pairs. Moreover, the time complexity of the proposed image registration method is obviously slighter than that of the Singular Value Decomposition (SVD) method. The comparing experiment results and the unwrapping phase results show that the improved Goldstein filter is more effective than the original Goldstein filter in noise elimination. The atmospheric influence correction experiment results show that the land subsidence areas with atmospheric influence correction are more clarified than that of without atmospheric influence correction. In summary, the presented methods directly improved the measurement results of D-InSAR at X-band.

Robust image registration of biological microscopic images.

Image registration of biological data is challenging as complex deformation problems are common. Possible deformation effects can be caused in individual data preparation processes, involving morphological deformations, stain variations, stain artifacts, rotation, translation, and missing tissues. The combining deformation effects tend to make existing automatic registration methods perform poor. In our experiments on serial histopathological images, the six state of the art image registration techniques, including TrakEM2, SURF + affine transformation, UnwarpJ, bUnwarpJ, CLAHE + bUnwarpJ and BrainAligner, achieve no greater than 70% averaged accuracies, while the proposed method achieves 91.49% averaged accuracy. The proposed method has also been demonstrated to be significantly better in alignment of laser scanning microscope brain images and serial ssTEM images than the benchmark automatic approaches (p < 0.001). The contribution of this study is to introduce a fully automatic, robust and fast image registration method for 2D image registration.

A robust aerial image registration method using Gaussian mixture models

Aerial image registration is one of the bases in many aerospace applications, such as aerial reconnaissance and aerial mapping. In this paper, we propose a novel aerial image registration algorithm which is based on Gaussian mixture models. First of all, considering the characters of the aerial images, the work uses a shape feature detector which computes the boundaries of regions with nearly the same gray-value to extract invariant feature. Then, a Gaussian mixture models (GMM) based image registration model is built and solved to estimate the transformation matrix between two aerial images. Furthermore, the proposed method is applied on real aerial images, and the results demonstrate the improved performance of the proposed algorithm.

Multi-contour registration based on feature points correspondence and two-stage gene expression programming

Abstract Image registration is a fundamental task in 3D reconstruction from an image sequence. Although this topic has been studied for decades, a general, robust, and automatic image registration method is rare, and most existing image registration methods are designed for a particular application. In this paper, image registration is treated as a formula discovery problem. A novel contour registration pipeline was proposed based on a foot-point-based feature point correspondence algorithm and a two-stage evolutionary algorithm. Our proposal has three objectives. First, we introduce a novel feature point extraction method that uses estimation of the curvature and the support region for every contour in the floating image. Second, we approximate the reference contour using a Gaussian mixture model (GMM) continuous optimization algorithm followed by an order-preserved foot-point detection method used to extract the feature points that correspond to the feature points of the floating contours. Third, we propose a hybrid evolutionary algorithm used to identify the registration formula for the reference image and the floating image. The hybrid evolutionary algorithm is a two-stage algorithm based on gene expression programming (GEP) and the improved cooperative particle swarm optimizer (CPSO). The optimal or near-optimal structure is accomplished using the GEP algorithm, and the parameters embedded in the structure are optimized by an opposition based learning (OBL)-based cooperative particle swarm optimizer (CPSO). Compared with other non-rigid registration methods, the developed registration pipeline produces competitive results with high accuracy.

A hybrid approach based on MEP and CSP for contour registration

Image registration is a fundamental task in 3D reconstruction from an image sequence. Although has been studied for decades, it is still rare to find a general, robust and automatic image registration method, and most existing image registration methods are designed for particular application. In this paper, image registration is boil down to a formula discovery problem to match feature points, we develop a new feature-based algorithm for contour registration automatically based on a hybrid approach combining Multi Expression Programming (MEP) with Clonal Selection Principle (CSP). Firstly, the image contours are extracted by fast global minimization of the active contour model, the feature point pairs which are used to establish training set of the hybrid approach are obtained using invariable moments. Secondly, the registration equations are acquired automatically by the proposed hybrid approach and the contours are then registered based on the equations. Experiments show that the proposed approach can be successfully applied to register the image pairs in which the contour information is well preserved.

Robust subpixel image registration of different-exposed images

Proposed is a new robust image registration method that can perform subpixel image registration for different-exposed images with highly photometric distortion. In this method, phase congruency is used to eliminate the effects of illumination change and photometric distortion for image registration using the phase correlation technique. Considering the presence of non-overlapped regions owing to photometric distortion, evolutionary programming is applied to search for the accurate translation parameters. Experimental results show that the proposed method is efficient and robust against illumination change and photometric distortion.

Wavelet-based image registration technique for high-resolution remote sensing images

Image registration is the process of geometrically aligning one image to another image of the same scene taken from different viewpoints at different times or by different sensors. It is an important image processing procedure in remote sensing and has been studied by remote sensing image processing professionals for several decades. Nevertheless, it is still difficult to find an accurate, robust, and automatic image registration method, and most existing image registration methods are designed for a particular application. High-resolution remote sensing images have made it more convenient for professionals to study the Earth; however, they also create new challenges when traditional processing methods are used. In terms of image registration, a number of problems exist in the registration of high-resolution images: (1) the increased relief displacements, introduced by increasing the spatial resolution and lowering the altitude of the sensors, cause obvious geometric distortion in local areas where elevation variation exists; (2) precisely locating control points in high-resolution images is not as simple as in moderate-resolution images; (3) a large number of control points are required for a precise registration, which is a tedious and time-consuming process; and (4) high data volume often affects the processing speed in the image registration. Thus, the demand for an image registration approach that can reduce the above problems is growing. This study proposes a new image registration technique, which is based on the combination of feature-based matching (FBM) and area-based matching (ABM). A wavelet-based feature extraction technique and a normalized cross-correlation matching and relaxation-based image matching techniques are employed in this new method. Two pairs of data sets, one pair of IKONOS panchromatic images from different times and the other pair of images consisting of an IKONOS panchromatic image and a QuickBird multispectral image, are used to evaluate the proposed image registration algorithm. The experimental results show that the proposed algorithm can select sufficient control points semi-automatically to reduce the local distortions caused by local height variation, resulting in improved image registration results.

A fast and robust image registration method based on an early consensus paradigm

This paper presents a fast and extremely robust feature-based method for planar registration of partly overlapping images; it uses a two-stage robust fitting approach comprising a fast estimation of a transformation hypothesis (that we show is highly likely to be correct) followed by a confirmation and refinement stage. The method is particularly suited for automatic stitching of oversize documents scanned in two or more parts. We show simulations, also supported by practical experiments, that prove both the robustness and computational efficiency of the approach.

Automated registration of laser Doppler perfusion images by an adaptive correlation approach: application to focal cerebral ischemia in the rat

Hemodynamic changes are extremely important in analyzing responses from a brain subjected to a stimulus or treatment. The Laser Doppler technique has emerged as an important tool in neuroscience research. This non-invasive method scans a low-power laser beam in a raster pattern over a tissue surface to generate the time course of images in unit of relative flux changes. Laser Doppler imager (LDI) records cerebral perfusion not only in the temporal but also in the spatial domain. The traditional analysis of LD images has been focused on the region-of-interest (ROI) approach, in which the analytical accuracy in an experiment that necessitates a relative repositioning between the LDI and the scanned tissue area will be weakened due to the operator's subjective decision in data collecting. This report describes a robust image registration method designed to obviate this problem, which is based on the adaptive correlation approach. The assumption in mapping corresponding pixels in two images is to correlate the regions in which these pixels are centered. Based on this assumption, correlation coefficients are calculated between two regions by a method in which one region is moved around over the other in all possible combinations. To avoid ambiguity in distinguishing maximum correlation coefficients, an adaptive algorithm is adopted. Correspondences are then used to estimate the transformation by linear regression. We used a pair of phantom LD images to test this algorithm. A reliability test was also performed on each of the 15 sequential LD images derived from an actual experiment by imposing rotation and translation. The result shows that the calculated transformation parameters (rotation: θ ̂ =7.7±0.5°; translation: Δ x ̂ =2.8±0.3, Δ y ̂ =4.7±0.4) are very close to the prior-set parameters (rotation: θ=8°; translation: Δ x=3, Δ y=5). This result indicates that this approach is a valuable adjunct to LD perfusion monitoring. An original sequence of LD images that recorded cerebral perfusion through a cranial window before, during and after middle cerebral artery occlusion (MCAo) is presented, together with the registered image sequence. Cerebral perfusion data acquired in a pixel-based manner from different anatomic locations of the registered LD image sequence are also presented over the whole time-course of the experiment.