An iterative point correspondence algorithm for automatic image registration: An application to dental subtraction radiography

Many image registration algorithms rely on the use of homologous control points on the two input image sets to be registered. In reality, the interactive identification of the control points on both images is tedious, difficult, and often a source of error. We propose a two-step algorithm to automatically identify homologous regions that are used as a priori information during the image registration procedure. First, a number of small control volumes having distinct anatomical features are identified on the model image in a somewhat arbitrary fashion. Instead of attempting to find their correspondences in the reference image through user interaction, in the proposed method, each of the control regions is mapped to the corresponding part of the reference image by using an automated image registration algorithm. A normalized cross-correlation (NCC) function or mutual information was used as the auto-mapping metric and a limited memory Broyden-Fletcher-Goldfarb-Shanno algorithm (L-BFGS) was employed to optimize the function to find the optimal mapping. For rigid registration, the transformation parameters of the system are obtained by averaging that derived from the individual control volumes. In our deformable calculation, the mapped control volumes are treated as the nodes or control points with known positions on the two images. If the number of control volumes is not enough to cover the whole image to be registered, additional nodes are placed on the model image and then located on the reference image in a manner similar to the conventional BSpline deformable calculation. For deformable registration, the established correspondence by the auto-mapped control volumes provides valuable guidance for the registration calculation and greatly reduces the dimensionality of the problem. The performance of the two-step registrations was applied to three rigid registration cases (two PET-CT registrations and a brain MRI-CT registration) and one deformable registration of inhale and exhale phases of a lung 4D CT. Algorithm convergence was confirmed by starting the registration calculations from a large number of initial transformation parameters. An accuracy of approximately 2 mm was achieved for both deformable and rigid registration. The proposed image registration method greatly reduces the complexity involved in the determination of homologous control points and allows us to minimize the subjectivity and uncertainty associated with the current manual interactive approach. Patient studies have indicated that the two-step registration technique is fast, reliable, and provides a valuable tool to facilitate both rigid and nonrigid image registrations.

In order to overcome the difficulty of automatic registration between the images of different rotations and resolutions, this paper presents a new image registration algorithm. In the algorithm, Harris-Laplace detector is applied to extract image features; a method named scale-space projection based on scale-invariance of Harris-Laplace is proposed; and a feature space on the projection information is constructed in the direction of the main feature orientation, which makes feature space rotation-invariant and resolution-invariant. In addition, the two steps are taken to accomplish feature matching: Euclidean distance is used to match two features coarsely, and then Random Sample Consensus (RANSAC) is applied to identify the correct matches, which implements feature matching accurately and efficiently. Experiments show that the algorithm in this study can accomplish image registration automatically and accurately between images even with large differences in resolutions and rotations.

Improved Rapid Corner Detection Algorithm in Medical Image

Image registration is a fundamental problem for applications in remote sensing. In this paper, a new coarse-to-fine registration framework is proposed. In coarse registration step, Quantum Particle Swarm Optimization (QPSO) is used as optimizer to find best rigid parameters. The similarity measure is the Mutual Information (MI) of whole images. This method is valid under various displacements. In fine registration step, Harris detector is implemented to extract feature points in reference image, and template window is used to obtain corresponding points in sensed image. The parameters of the best affine transformation are estimated using the corresponding feature points. Analysis and experiments show our method leads to highly automatic registration, and is able to handle large displacements between remote sensing images fast and robustly.

This chapter covers a general class of image registration algorithms that apply numerical optimization to similarity measures relating to cumulative functions of image intensities. An example of these algorithms is an algorithm minimizing the least-squares difference in image intensities due to an iterative gradient-descent approach. Algorithms in this class, which work well in 2D and 3D, can be applied simultaneously to multiple bands in an image pair and images with significant radiometric differences to accurately recover subpixel transformations. The algorithms discussed differ in the specific similarity measure, the numerical method used for optimization, and the actual computation used. The similarity measure can vary from a measure that uses a radiometric function to account for nonlinear image intensity differences in the least-squares equations, to one that is based on mutual information, which accounts for image intensity differences not accounted for by a standard functional model. The numerical methods considered are basic recursive descent, a method based on Levenberg-Marquardt's technique, and Spall's algorithm. This chapter relates to the above registration algorithms and classifies them by their various elements. It also analyzes the image classes for which variants of these algorithms apply best.

We present a diffeomorphic image registration algorithm to learn spatial transformations between pairs of images to be registered using fully convolutional networks (FCNs) under a self‐supervised learning setting. Particularly, a deep neural network is trained to estimate diffeomorphic spatial transformations between pairs of images by maximizing an image‐wise similarity metric between fixed and warped moving images, similar to those adopted in conventional image registration algorithms. The network is implemented in a multi‐resolution image registration framework to optimize and learn spatial transformations at different image resolutions jointly and incrementally with deep self‐supervision in order to better handle large deformation between images. A spatial Gaussian smoothing kernel is integrated with the FCNs to yield sufficiently smooth deformation fields for diffeomorphic image registration. The spatial transformations learned at coarser resolutions are utilized to warp the moving image, which is subsequently used as input to the network for learning incremental transformations at finer resolutions. This procedure proceeds recursively to the full image resolution and the accumulated transformations serve as the final transformation to warp the moving image at the finest resolution. Experimental results for registering high‐resolution 3D structural brain magnetic resonance (MR) images have demonstrated that image registration networks trained by our method obtain robust, diffeomorphic image registration results within seconds with improved accuracy compared with state‐of‐the‐art image registration algorithms.

Image registration is a fundamental procedure in image processing that aligns two or more images of the same scene taken from different times, different viewpoints, or even different sensors. It is reasonable to orientate two images by matching corresponding pixels or regions that are considered identical. Based on this concept, this paper proposes a novel image registration method that applies the information theorem on intensity difference data. An entropy-based objective function is then developed according to the histogram of the intensity difference. The intensity difference represents the absolute gray-level difference of the corresponding pixels between the reference and sensed images over the overlapped region. The proposed registration method is to align the sensed image onto the reference image by minimizing the entropy of the intensity difference through iteratively updating the parameters of the similarity transformation. For performance evaluation, the proposed method is compared with the two exiting registration methods in terms of eight test image sets. The experiment is divided into two scenarios. One is to investigate the sensitivity (i.e., robustness) of the objective functions in these three different methods; the other is to verify the effectiveness of the proposed method. Through the experimental results, the proposed method is shown to be very effective in image registration and outperforms the other two methods over the test image sets.

Over the past few decades, fast<strong>-</strong>growth has occurred in the area of medical image acquisition devices, and physicians now rely on the utilization of medical images for the diagnosis, treatment plans, and surgical guidance. Researchers have classified medical images according to two structures: anatomical and functional structures. Due to this classification, the data obtained from two or more images of the same object frequently provide complementary and more abundant information through a process known as multimodal medical model registration. Image registration is spatially mapping the coordinate system of the two images obtained from a different viewpoint and utilizing various sensors. Several automatic multimodal medical image registration algorithms have been introduced based on types of medical images and their applications to increase the reliability, robustness, and accuracy. Due to the diversity in imaging and the different demands<strong> </strong>for applications, there is no single registration algorithm that can do that. This paper introduces a novel method for developing a multimodal medical image registration system that can select the most accepted registration algorithm from a group of registration algorithms for a variety of input datasets. The method described here is based on a machine learning technique that selects the most promising candidate. Several experiments have been conducted, and the results reveal that the novel approach leads to considerably faster reliability, accuracy, and more robustness registration algorithm selection.

Dental radiographs are used to diagnose and treat dental diseases. For accurate diagnosis, geometrical registration of radiographs is required. In order to compare two dental radiographs prepared from the same part of teeth in different time intervals. It is necessary to register the locality of images, so that by removing deviations between the two images and finding a mapping function, local transitions, rotational and scale differences and even visual differences are corrected. In this paper, an X-ray dental image registration algorithm based on template matching and similarity measure is proposed. The proposed registration process is carried out in three stages. The first stage includes the automatic selection of a set of control points of the sensed image based on its features. In the second stage, displacement vectors of these control points are determined by comparing a template and normalized cross correlation (NCC) as a similarity measure. Full displacement vector field is then performed using these displacement vectors and the affine transformation function, and the final correction is carried out by changing the geometric shape of the reference image. The obtained results of the proposed algorithm show effectiveness of this algorithm in dental radiograph image registration.

A non-rigid medical image registration method based on improved linear elastic model

Remote sensing data includes images captured by satellite-based sensors that encode geospatial information. Image registration, an essential preprocessing step, involves aligning satellite images with geometrically precise reference images. Co-registering multi-temporal datasets to reference images is a prerequisite for reliable spatiotemporal analysis and related applications. In this paper, we present a novel satellite image registration method using Spatial Transformer Networks (STNs), where a localization network learns affine transformation parameters between the given input and reference image, and a grid generator-sampler produces the transformed image. The proposed model is capable of learning complex features/patterns despite the spatiotemporal changes. We evaluate our model on a custom multi-temporal ResourceSAT-2/2A LISS-3 dataset using two training approaches. Approach 1 involves applying the loss function after the grid generator-sampler stage, simulating practical scenarios where transformation parameters are unknown. Approach 2 applies the loss function after the parameter retrieval stage, leveraging known parameters from our custom dataset. In Approach 1, we evaluated four distinct models using various hyperparameter configurations and conducted a comparative analysis. The results, based on normalized cross-correlation (NCC) loss computed between the transformed and reference images using the Adam optimizer, were promising. In Approach 2, the NCC loss with the Adam optimizer was leveraged, but here it measures the error between the predicted and the known affine parameters. Our proposed method yields sub-pixel-level accurate image registration. We performed a comparative analysis of the results from Approaches 1 and 2 against existing methods to evaluate their relative performance.

From the past few years, Image registration has become an emerging and hot topic for researchers. Due to the global need for low computation, less time consuming, and good quality image mapping methods has caused an image registration technique alive in multiple application areas. Image registration is the method, in which the pixels or control points of one image are superimposed on the other image. The image which is superimposed is called reference image and the image upon which superimposition process is done is called target image. The input images are reference images and sensed images. By processing these images through image registration algorithms, the target images are created. During this process, the concentration is on various methods of mapping parameters. Basically image registration is of two types: Area based and Feature based. Area based works on the intensity of image and feature based is based on feature points or objects of image. Image registration has wide scope in medical field and research. This paper presents a review on basic image registration and its techniques.

Feature point matching aims to automatically establish point-to-point correspondences between two images acquired from the same scene but through two different viewpoints. This matching process is essential to many image processing and computer vision tasks, such as image registration, object detection and tracking, to name a few. However, the well-known SIFT feature descriptors together with the use of the nearest neighbor matching criterion oftentimes lead to many incorrect point-to-point correspondences. This would be incurred especially when the two images undergo a large viewpoint variation and/or contain a severely cluttered background. To improve the performance of feature point matching process, three contributions are made in this thesis: 1) mismatch removal; 2) common visual pattern discovery; and 3) feature histogram equalization (FHE). The first two are graph model-based approaches, and the last one is a novel technique on the enhancement of feature descriptors. More details are provided as follows. For certain image processing applications such as matching two images captured from disparate views, or the so-called wide-baseline image matching as demonstrated in this thesis, a sufficiently large number of point-to-point correspondences are needed in the first place. However, these corresponding pairs often contain mismatched ones. Consequently, the accuracy of the spatial transformation estimation between the two images under matching will be reduced to a certain degree. Therefore, how to identify and remove the mismatched corresponding pairs from the established ones is the main goal of our first contribution. For that, a bipartite graph model is exploited to the segmented image regions to establish all possible region-to-region correspondences. After that one-to-one region correspondences, called the coherent region pairs (CRPs), can be further identified by using the Hungarian method and the proposed region-to-region similarity measurement metric. The established CRPs will be utilized as the reference information in order to identify and remove those mismatched point-to-point correspondences. Extensive experimental results have demonstrated that our proposed mismatch removal method could reduce incorrect SIFT-based point-to-point correspondences for the application of wide-baseline image matching that normally requires a large targeted number of matching pairs. For the second contribution on common visual pattern discovery, the entire work still begins with the SIFT feature descriptors. However, the main objective is to yield the correct point-to-point matching pairs in the first place, while the actual number of such established matching pairs tends to be small. The key novelty of our approach lies in the use of the directed graph (or diagraph) model that has two link weights on each link, rather than one link weight inherited in the conventional graph model. The principle of pairwise spatial consistency is first exploited to generate the initial link weight for each link. Based on this link weight value, two relative link weights are then generated by considering the relativeness of neighboring vertices at each vertex of the link. For that, the n-ranking process and a novel link weight enhancement technique are proposed. Consequently, the resulted relative link weights are more robust to combat various adverse scenarios such as large viewpoint variations and indiscriminative feature descriptors. Based on the relative link weights generated at each assumed scale change factor, the strongly-associated subgraph can be extracted from the digraph by applying the non-cooperative game theory for handling non-symmetric adjacency matrix issue. All the vertices (i.e., point-to-point correspondences) belonging to the strongly-associated subgraph extracted from the diagraph established at an assumed scale change factor are collectively treated as one common visual pattern; hopefully, this set of vertices corresponds to one visual object. If this is not the case, our proposed topological splitting algorithm might be able to further discriminate them. Extensive experiments have been conducted on the simulated SIFT feature points for standing out the technical challenges, followed by performing evaluations on six thoughtfully chosen natural image pairs and Columbia dataset to demonstrate the efficacy and robustness of the proposed method. For the third contribution, feature contrast is first introduced as a measurement of the degree of self-similarity contained in an image pair. Note that, for an image pair with strong self-similarity, the established SIFT feature descriptors would be quite similar to each other, therefore the feature contrast is low. The FHE is then proposed to equalize SIFT feature descriptors by independently modifying the vector-component values of feature descriptors at each vector dimension; consequently, the feature contrast is effectively enhanced for the purpose of better discrimination of the feature descriptors. Extensive simulation results have clearly shown that our proposed FHE method could effectively improve the precision of SIFT-based point-to-point correspondences, especially to those image pairs containing a large amount of self-similar regions. As expected, if the image pairs contain only a few or even no self-similar regions, the performance gain would be marginal in this case.

A protocol for evaluation of similarity measures for non-rigid registration

Abstract. This paper looks into the latest achievement in the low-cost Unmanned Aerial Vehicle (UAV) technology in their capacity to map the semi-development areas. The objectives of this study are to establish a new methodology or a new algorithm in image registration during interior orientation process and to determine the accuracy of the photogrammetric products by using UAV images. Recently, UAV technology has been used in several applications such as mapping, agriculture and surveillance. The aim of this study is to scrutinize the usage of UAV to map the semi-development areas. The performance of the low cost UAV mapping study was established on a study area with two image processing methods so that the results could be comparable. A non-metric camera was attached at the bottom of UAV and it was used to capture images at both sites after it went through several calibration steps. Calibration processes were carried out to determine focal length, principal distance, radial lens distortion, tangential lens distortion and affinity. A new method in image registration for a non-metric camera is discussed in this paper as a part of new methodology of this study. This method used the UAV Global Positioning System (GPS) onboard to register the UAV image for interior orientation process. Check points were established randomly at both sites using rapid static Global Positioning System. Ground control points are used for exterior orientation process, and check point is used for accuracy assessment of photogrammetric product. All acquired images were processed in a photogrammetric software. Two methods of image registration were applied in this study, namely, GPS onboard registration and ground control point registration. Both registrations were processed by using photogrammetric software and the result is discussed. Two results were produced in this study, which are the digital orthophoto and the digital terrain model. These results were analyzed by using the root mean square errors and mean absolute error to determine the level of accuracy and the precision of photogrammetric products. It can be concluded that the new method of image registration by using the GPS onboard of the UAV produces medium accuracy result compared to the method that uses the ground control point. This new method can be used for the medium accuracy requirements. Unmanned Aerial Vehicle can be used for several applications, which requires a medium accuracy.