Explore the vast range of titles available.

In this paper, we continue the research cycle on the properties of convolutional neural network-based image recognition systems and ways to improve noise immunity and robustness. Currently, a popular research area related to artificial neural networks is adversarial attacks. The adversarial attacks on the image are not highly perceptible to the human eye, and they also drastically reduce the neural network’s accuracy. Image perception by a machine is highly dependent on the propagation of high frequency distortions throughout the network. At the same time, a human efficiently ignores high-frequency distortions, perceiving the shape of objects as a whole. We propose a technique to reduce the influence of high-frequency noise on the CNNs. We show that low-pass image filtering can improve the image recognition accuracy in the presence of high-frequency distortions in particular, caused by adversarial attacks. This technique is resource efficient and easy to implement. The proposed technique makes it possible to measure up the logic of an artificial neural network to that of a human, for whom high-frequency distortions are not decisive in object recognition.

Distortions in scanning electron microscope (SEM) images compromise characterization accuracy and restrict reliable quantitative analysis. Quantifying and correcting these distortions remains challenging due to the complexity of their inherent sources, such as scanning coil hysteresis and electronic circuit response delays. To address this, we independently developed a scanning controller and software system that enables customizable scanning strategies and is crucial for capturing unprocessed raw data. We utilized the characteristic row misalignment of snake scanning to split images into sub-images, measure offsets using the ORB algorithm, and apply pixel compensation. Experimental validation shows that corrected images exhibit reduced distortion artifacts, with structural similarity comparable to raster scanning results and improved reference-free quality metrics. The distortion magnitude is independent of magnification, primarily governed by dwell time, and stabilizes at a minimum level when the dwell time reaches a critical threshold. This work clarifies the relationship between scanning parameters and distortion behavior, guiding the optimization of SEM scanning strategies. Furthermore, it offers a potential scalable framework for distortion correction in related microscopy techniques. Many of these techniques also face distortion issues from hardware hysteresis or circuit delays, similar to SEM.

Blind image quality assessment is a challenging task particularly due to the unavailability of reference information. Training a deep neural network requires a large amount of training data which is not readily available for image quality. Transfer learning is usually opted to overcome this limitation and different deep architectures are used for this purpose as they learn features differently. After extensive experiments, we have designed a deep architecture containing two CNN architectures as its sub-units. Moreover, a self-collected image database BIQ2021 is proposed with 12,000 images having natural distortions. The self-collected database is subjectively scored and is used for model training and validation. It is demonstrated that synthetic distortion databases cannot provide generalization beyond the distortion types used in the database and they are not ideal candidates for general-purpose image quality assessment. Moreover, a large-scale database of 18.75 million images with synthetic distortions is used to pretrain the model and then retrain it on benchmark databases for evaluation. Experiments are conducted on six benchmark databases three of which are synthetic distortion databases (LIVE, CSIQ and TID2013) and three are natural distortion databases (LIVE Challenge Database, CID2013 and KonIQ-10 k). The proposed approach has provided a Pearson correlation coefficient of 0.8992, 0.8472 and 0.9452 subsequently and Spearman correlation coefficient of 0.8863, 0.8408 and 0.9421. Moreover, the performance is demonstrated using perceptually weighted rank correlation to indicatethe perceptual superiority of the proposed approach. Multiple experiments are conducted to validate the generalization performance of the proposed model by training on different subsets of the databases and validating on the test subset of BIQ2021 database.

Highly divergent estimates of benthic cover of sponges have been reported for Caribbean mesophotic reefs (90–100 m) based on quadrat point-intercept data collection using 2 methods: visual surveys conducted in situ by technical divers, and analyses of photographs taken by unmanned underwater vehicles (UUVs). The second method has been criticized for potential errors from image distortion caused by variable camera angle relative to the substratum, but without a broader comparison of both methods. We find that studies that have used the UUV-based method are advantageous for a number of reasons, most importantly: (1) access to the full mesophotic zone, (2) higher sample replication, and (3) reduced likelihood of sampling bias. For tech diving surveys conducted at 91 m, i.e. the deepest depth reported using this method but only midway through the mesophotic zone, studies have reported particularly high sponge cover (∼80 vs. <10% for UUV-based surveys), which may be a consequence of low replication and targeted sampling influenced by very short working times under hazardous conditions. When evaluating benthic abundance metrics from photographs, issues associated with variable substratum angle are common to any topographically complex surface, particularly within a quadrat. Nevertheless, point-intercept estimates are not dependent on quadrat area and are not subject to error due to image distortion or surface complexity. Unlike visual census data from tech dives, UUV photographs can be validated by taxonomic experts and archived for re-analysis. Past tech diving surveys should be repeated using the UUV-based method with greater replication over the full range of the mesophotic zone in order to reconcile divergent estimates of benthic cover.

Remote sensing imagery enables object detection systems to localize and classify targets for critical applications like surveillance and autonomous driving. However, distortions introduced during image acquisition, transmission, or compression degrade the detection performance, posing challenges for real-world applications. This study conducts a comprehensive robustness evaluation of seven state-of-the-art YOLO models, including YOLOv5, YOLOv7, YOLOv8, YOLOv9, YOLOv10, YOLOv11, and the modified YOLOv5 against four common distortions: Additive White Gaussian Noise (AWGN), JPEG and JPEG2000 compressions, and Gaussian blurring. Using the DOTA-v1.0 dataset, we generated 40 distortion test sets (10 levels per distortion type). The obtained results demonstrate that all distortions degrade performance across all evaluated models. YOLOv9 outperforms others YOLO models in terms of mean average precision under different distortions. YOLOv7 and YOLOv10 exhibit the weakest robustness, whereas YOLOv11 shows low resistance to AWGN distortion.

Space-surface bistatic synthetic aperture radar (SS-BiSAR) offers an additional observation angle for monostatic spaceborne SAR, making it a promising technology for high-accuracy deformation retrieval technology in local regions. Repeat-pass SS-BiSAR tomography can accurately estimate the surfaces of buildings and steep areas, effectively removing terrain phases during deformation retrieving. However, inaccuracies in the orbital ephemeris can lead to image geometry distortion, reducing image pair coherence, introducing interferometric phase errors, and consequently deteriorating tomographic precision. This paper precisely models the image geometry distortion and interferometric phase error caused by repeat-pass ephemeris error. We propose an ephemeris correction method based on the chirp-Z transform to address these issues. Furthermore, we introduce an accurate tomography model to improve 3D reconstruction accuracy. Our first SS-BiSAR tomography experiment, conducted using the Chinese Lutan-1 satellite, demonstrates that the correlation coefficient is improved by 0.16 after ephemeris error correction. Moreover, the density and precision of the tomographic point cloud are improved by 13.7% and 12.1%, respectively.

In Part I of this diptych, we outlined the theory and an analysis methodology for quantitative phase recovery from real-space distortions of Fresnel images acquired in the parallel mode of transmission electron microscopy (TEM). In that work, the properties of the method, termed TEM-differential phase contrast (TEM-DPC), were highlighted through the use of simulated data. In this work, we explore the use of the TEM-DPC technique with experimental cryo-TEM images of a thin lamella of a low-temperature two-dimensional (2D) ferromagnetic material, K 2 CuF 4 , to perform two tasks. First, using images recorded below the ordering temperature, we compare the TEM-DPC method with the transport of intensity one for phase recovery and discuss the relative advantages the former has for experimental data. Second, by tracking the induction of the sample as it is driven through a phase transition by heating, we extract estimates for the critical temperature and critical exponent of the order parameter. The value of the latter is consistent with the 2D XY class, raising the prospect that a Kosterlitz–Thoules transition may have occurred.

The aberration-corrected scanning transmission electron microscope has great sensitivity to environmental or instrumental disturbances such as acoustic, mechanical, or electromagnetic interference. This interference can introduce distortions to the images recorded and degrade both signal noise and resolution performance. In addition, sample or stage drift can cause the images to appear warped and leads to unreliable lattice parameters being exhibited. Here a detailed study of the sources, natures, and effects of imaging distortions is presented, and from this analysis a piece of image reconstruction code has been developed that can restore the majority of the effects of these detrimental image distortions for atomic-resolution data. Example data are presented, and the performance of the restored images is compared quantitatively against the as-recorded data. An improvement in apparent resolution of 16% and an improvement in signal-to-noise ratio of 30% were achieved, as well as correction of the drift up to the precision to which it can be measured.

Recently, with more portable diagnostic devices being moved to people anywhere, point-of-care (PoC) imaging has become more convenient and more popular than the traditional “bed imaging”. Instant image segmentation, as an important technology of computer vision, is receiving more and more attention in PoC diagnosis. However, the image distortion caused by image preprocessing and the low resolution of medical images extracted by PoC devices are urgent problems that need to be solved. Moreover, more efficient feature representation is necessary in the design of instant image segmentation. In this paper, a new feature representation considering the relationships among local features with minimal parameters and a lower computational complexity is proposed. Since a feature window sliding along a diagonal can capture more pluralistic features, a Diagonal-Axial Multi-Layer Perceptron is designed to obtain the global correlation among local features for a more comprehensive feature representation. Additionally, a new multi-scale feature fusion is proposed to integrate nonlinear features with linear ones to obtain a more precise feature representation. Richer features are figured out. In order to improve the generalization of the models, a dynamic residual spatial pyramid pooling based on various receptive fields is constructed according to different sizes of images, which alleviates the influence of image distortion. The experimental results show that the proposed strategy has better performance on instant image segmentation. Notably, it yields an average improvement of 1.31% in Dice than existing strategies on the BUSI, ISIC2018 and MoNuSeg datasets.

The acquisition of images with a fish-eye lens can cause serious image distortion because of the short focal length of the lens. As a result, it is difficult to use the obtained image information. To make use of the effective information in the image, these distorted images must first be corrected into the perspective of projection images in accordance with the human eye's observation abilities. To solve this problem, this study presents an adaptive classiffication fitting method for fish-eye image correction. The degree of distortion in the image is represented by the difference value of the distances from the distorted point and undistorted point to the center of the image. The target points selected in the image are classified by the difference value. In the areas classified by different distortion differences, different parameter curves were used for fitting and correction. The algorithm was verified through experiments. The results showed that this method has a substantial correction effect on fish-eye images taken by different fish-eye lenses.