Given Pair Of Images Research Articles

DeCoGAN: MVCT image denoising via coupled generative adversarial network

Objective. In helical tomotherapy, image-guided radiotherapy employs megavoltage computed tomography (MVCT) for precise targeting. However, the high voltage of megavoltage radiation introduces substantial noise, significantly compromising MVCT image clarity. This study aims to enhance MVCT image quality using a deep learning-based denoising method. Approach. We propose an unpaired MVCT denoising network using a coupled generative adversarial network framework (DeCoGAN). Our approach assumes that a universal latent code within a shared latent space can reconstruct any given pair of images. By employing an encoder, we enforce this shared-latent space constraint, facilitating the conversion of low-quality (noisy) MVCT images into high-quality (denoised) counterparts. The network learns the joint distribution of images from both domains by leveraging samples from their respective marginal distributions, enhanced by adversarial training for effective denoising. Main Results. Compared to an analytical algorithm (BM3D) and three deep learning-based methods (RED-CNN, WGAN-VGG and CycleGAN), the proposed method excels in preserving image details and enhancing human visual perception by removing most noise and retaining structural features. Quantitative analysis demonstrates that our method achieves the highest peak signal-to-noise ratio and Structural Similarity Index Measurement values, indicating superior denoising performance. Significance. The proposed DeCoGAN method shows remarkable MVCT denoising performance, making it a promising tool in the field of radiation therapy.

HyperPredict: Estimating Hyperparameter Effects for Instance-Specific Regularization in Deformable Image Registration

Methods for medical image registration infer geometric transformations that align pairs, or groups, of images by maximising an image similarity metric. This problem is ill-posed as several solutions may have equivalent likelihoods, also optimising purely for image similarity can yield implausible deformable transformations. For these reasons regularization terms are essential to obtain meaningful registration results. However, this requires the introduction of at least one hyperparameter, often termed λ, which serves as a trade-off between loss terms. In some approaches and situations, the quality of the estimated transformation greatly depends on hyperparameter choice, and different choices may be required depending on the characteristics of the data. Analyzing the effect of these hyperparameters requires labelled data, which is not commonly available at test-time. In this paper, we propose a novel method for evaluating the influence of hyperparameters and subsequently selecting an optimal value for given pair of images. Our approach, which we call HyperPredict, implements a Multi-Layer Perceptron that learns the effect of selecting particular hyperparameters for registering an image pair by predicting the resulting segmentation overlap and measures of deformation smoothness. This approach enables us to select optimal hyperparameters at test time without requiring labelled data, removing the need for a one-size-fits-all cross-validation approach. Furthermore, the criteria used to define optimal hyperparameter is flexible post-training, allowing us to efficiently choose specific properties (e.g. overlap of specific anatomical regions of interest, smoothness/plausibility of the final displacement field). We evaluate our proposed method on the OASIS brain MR standard benchmark dataset using a recent deep learning approach (cLapIRN) and an algorithmic method (Niftyreg). Our results demonstrate good performance in predicting the effects of regularization hyperparameters and highlight the benefits of our image-pair specific approach to hyperparameter selection.

Spatially Resolved Velocity Mapping of the Melt Plume During High-Pressure Gas Atomization of Liquid Metals

We present details of an image analysis algorithm designed specifically to determine the velocity of material in the melt plume during high-pressure, close-coupled gas atomization. Following high-speed filming (16,000 fps) pairs of images are used to identify and track dominant features within the plume. Due to the complexity of the atomization plume, relatively few features are tracked between any given pair of images, but by averaging over the many thousands of frames obtained during high-speed filming a spatially resolved map of the average velocity of material in the plume can be built up. Velocities in the plume are typically very low compared to that of the supersonic gas, being around 30 m s−1 on the margins of the plume where the melt interacts strongly with the gas and dropping to < 10 m s−1 in the center of the melt plume. Consequently, the efficiency of the atomizer in transferring kinetic energy from the gas to the melt is correspondingly very low, with this being estimated as being no more than 0.1 pct.

Image-Based Camera Localization Algorithm for Smartphone Cameras Based on Reference Objects

Navigating an individual in both outdoor and indoor environments would provide great facilities for the community. Image-based approaches are more affordable and feasible in comparison to the high cost of sensor devices and unavailability of guiding by GPS in indoor systems. Thus, we develop our system using cameras owing to their easy availability and accessibility. Primarily, detecting the location of personnel is needed to orientate a human. However, determining camera localization is a challenging task on the computer vision field. Therefore, in this study, a novel approach is proposed to address the problem using panoramic image sequences. We apply the scale invariant future transform for detecting corresponding points, and based on these points, a reference object is considered. The Levenberg–Marquardt optimization method is utilized to minimize reprojection errors for a given pair of images. As a better alternative to traditional methods, which apply the known surface of a reference object, our formulation rests on the scale ambiguity in different coordinate systems. Experiments are performed on Herz-Jesus-P8, Fountain-P11, and Fountain-P25 datasets to evaluate our dataset and compare our proposal with respect to prior works. We believe our proposal comparatively outperforms previous works indicating lower error rate in wide baselines between camera pairs.

309 環境内に複数存在する類似形状物の画像を用いた自動位置認識と把持ロボットシステムへの適用(OS3-2 生産システムの国際展開)

This article concerns the problem of measuring the position and orientation of a target object pair in the factory where a number of similar objects lie in the environment. Several techniques in the field of image processing have been adapted to recognize the target pair in the front correctly and to exclude the same objects located behind the target and/or the image of background. All the processes necessary to retrieve the required information from the pair of photos has been fully automated to form an algorithm. An example of information retrieval from the given pair of images is presented to show the effectiveness of the proposed algorithm.

Complex Wavelet Structural Similarity: A New Image Similarity Index

We introduce a new measure of image similarity called the complex wavelet structural similarity (CW-SSIM) index and show its applicability as a general purpose image similarity index. The key idea behind CW-SSIM is that certain image distortions lead to consistent phase changes in the local wavelet coefficients, and that a consistent phase shift of the coefficients does not change the structural content of the image. By conducting four case studies, we have demonstrated the superiority of the CW-SSIM index against other indices (e.g., Dice, Hausdorff distance) commonly used for assessing the similarity of a given pair of images. In addition, we show that the CW-SSIM index has a number of advantages. It is robust to small rotations and translations. It provides useful comparisons even without a preprocessing image registration step, which is essential for other indices. Moreover, it is computationally less expensive.

Semi-automatic identification of print layers from a sequence of sample images: A case study from banknote print inspection

This article presents an approach for finding displacements of print layers from sequences of sample images semi-automatically with the aim to simplify and shorten the setup of inspection systems for printing processes in which the perfect alignment of print layers cannot be guaranteed. The basic idea behind the proposed approach is to identify pixels which are likely to have the same displacements for a given pair of images. This relatively coarse information is computed for several pairs of sample images and aggregated in order to identify regions that tend to have the same displacement over a large proportion of image pair comparisons. This idea is motivated and justified in detail. The test cases considered in this study are data from banknote print inspection. We use these data to illustrate the steps of the algorithm. The examples demonstrate the method’s capability to sensibly identify print layers, even if they overlap partially. Although the paper concentrates on a particular case study, the method can be used in any print inspection process with similar characteristics.

Matching of medical images by self-organizing neural networks

A general approach to the problem of image matching which exploits a multi-scale representation of local image structure and the principles of self-organizing neural networks is introduced. The problem considered is relevant in many imaging applications and has been largely investigated in medical imagery, especially as regards the integration of different imaging procedures. A given pair of images to be matched, named target and stimulus respectively, are represented by Gabor Wavelets. Correspondence is computed by exploiting the learning procedure of a neural network derived from Kohonen’s SOM. The SOM units coincide with the pixels of the target image and their weight are pointers to those of the stimulus images. The standard SOM rule is modified so as to account for image features. The properties of our method are tested by experiments performed on synthetic images. The considered implementation has shown that is able to recover a wide range of transformations including global affine transformations and local distortions. Tests in the presence of additive noise indicate considerable robustness against statistical variability. Applications to clinical images are presented.

Color gamut mapping based on a perceptual image difference measure

Gamut mapping is a color transformation technique to solve a problem caused by mismatch of gamuts among imaging devices. One plausible goal of gamut mapping is to find a reproduction that is perceptually closest to the corresponding original image when an exact color matching is not possible. Several measures to quantify the perceptual difference between images have been proposed and applied to the gamut mapping problem. However most of the measures, such as average color difference, are applied on a pixel-wise basis and show poor correlation with human visual perception. This article describes a model of the perceptual image difference for a given pair of images, which takes the human's contrast sensitivity into account and applies the model to a gamut mapping for generating a reproduction with minimum perceptual image difference. The model has a multispatial-frequency channel structure with tunable peak gains for each channel, which are determined by psychophysical experiments, so that the model output fits the observer's sensitivity to the image difference. A gamut-mapped image with minimum perceptual image difference is obtained by an iterative minimization process. To evaluate the proposed method, subjective evaluation experiments are performed to construct ratio scales that measure perceptual image difference of gamut-mapped reproductions generated by the proposed and pixel-wise methods. Results show that the reproductions by the proposed method are perceived as perceptually closest to the original, and the model's estimate of perceptual difference correlates better with the experimentally measured perceived image difference than other pixel-wise measures. © 1999 John Wiley & Sons, Inc. Col Res Appl, 24, 280–291, 1999

Content-based image retrieval using a composite color-shape approach

We have proposed a composite feature measure which combines the shape and color features of an image based on a clustering technique. We have also developed a similarity measure to compute the degree of match between a given pair of images. This technique can be used for content-based image retrieval of images using shape and/or color. We have tested our technique on two image databases: one consisting of 100 synthetic images, and another database consisting of 500 actual trademarks images. Test results of the proposed scheme for retrieval of images using only shape, only color, and a weighted combination of the two are presented. The efficiency of retrieval is found to be very high and the experimental results are promising for practical applications.

Chapter 27 Computer-Assisted Analysis of Flagellar Structure

Publisher Summary The identification of polypeptides as subunits of specific axonemal structures has arisen because of analysis of flagellar mutants of Chlamydomonas reinhardtii by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). This approach has been particularly useful in sorting out the complexity of the inner dynein arms. Methods such as cross-correlation analysis and rotational averaging of structures possessing ninefold axial-symmetry have been used to average transverse sections of insect sperm flagella. Averages of doublet cross sections have been obtained to make qualitative comparisons between inner and outer dynein arm mutants in Chlamydomonas . Described in this chapter are the steps involved in obtaining averages of axonemal components and the methods for making statistical comparisons between them. There is illustration of specimen preparation and electron microscopy. There is image analysis; the programs involve digitization, display, and quantitative analysis of images. The hardware used for the analysis includes a Dage MTI video camera. Images of axonemes containing outer doublets in cross section are found relatively quickly and easily in sections of pellets. Axonemes are chosen for analysis if they have a complete set of nine outer-doublet microtubules, an intact central pair, and protofilaments that are visible in at least one outer-doublet microtubule. The individual doublets at each position are aligned, averaged, and normalized, and statistical comparisons made to determine if there are doublet-specific differences in dynein structure in a mutant strain. There is analysis of longitudinal sections. Statistical comparisons of different samples are performed by a nested analysis of variance, using the sums of squared deviations computed from the means, standard deviations, and sample number for each individual sample, then test is done for the significance of differences in specified areas of the dynein arms or at each pixel in a given pair of images by taking into account the intra- and inter-sample variability.