Original Scene Research Articles

Construction of virtual test scenarios for intelligent car-pedestrian interaction

Aiming at the needs of intelligent car and pedestrian interaction testing under urban conditions, this paper proposes a scenario generation method that comprehensively considers the frequency of scenarios in the real world and the degree of challenge to human-vehicle interaction performance. First, according to the key features of intelligent car and pedestrian interaction, the original scene data of pedestrians crossing the road is extracted from the natural driving dataset; then, according to the needs of accelerated testing, a key scenario extraction method based on importance sampling theory is designed to extract and construct important scenarios for intelligent car-pedestrian interaction testing from the original scenes; finally, by comparing the data distribution of important scenarios with original scenarios, it is shown that this method can effectively screen out scenarios that may challenge safety performance during driving, thereby realizing accelerated testing, while taking into account the statistical characteristics of the test scenarios.

A Light-Weight Self-Supervised Infrared Image Perception Enhancement Method

Convolutional Neural Networks (CNNs) have achieved remarkable results in the field of infrared image enhancement. However, the research on the visual perception mechanism and the objective evaluation indicators for enhanced infrared images is still not in-depth enough. To make the subjective and objective evaluation more consistent, this paper uses a perceptual metric to evaluate the enhancement effect of infrared images. The perceptual metric mimics the early conversion process of the human visual system and uses the normalized Laplacian pyramid distance (NLPD) between the enhanced image and the original scene radiance to evaluate the image enhancement effect. Based on this, this paper designs an infrared image-enhancement algorithm that is more conducive to human visual perception. The algorithm uses a lightweight Fully Convolutional Network (FCN), with NLPD as the similarity measure, and trains the network in a self-supervised manner by minimizing the NLPD between the enhanced image and the original scene radiance to achieve infrared image enhancement. The experimental results show that the infrared image enhancement method in this paper outperforms existing methods in terms of visual perception quality, and due to the use of a lightweight network, it is also the fastest enhancement method currently.

High-risk powered two-wheelers scenarios generation for autonomous vehicle testing using WGAN

Objective Autonomous vehicles (AVs) have the potential to revolutionize the future of mobility by significantly improving traffic safety. This study presents a novel method for validating the safety performance of AVs in high-risk scenarios involving powered 2-wheelers (PTWs). By generating high-risk scenarios using in-depth crash data, this study is devoted to addressing the challenge of public road scenarios in testing, which often lack the necessary complexity and risk to effectively evaluate the capabilities of AVs in high-risk situations. Method Our approach employs a Wasserstein generative adversarial network (WGAN) to generate high-risk scenes, particularly focusing on PTW scenarios. By extracting 314 car-to-PTW crashes from the China In-depth Mobility Safety Study–Traffic Accident database, we simulate outcomes using PC-Crash software. The data are divided into scenes at 0.1-s intervals, with WGAN generating numerous high-risk scenes. By using a cumulative distribution function (CDF), we sampled and analyzed the vehicle’s dynamic information to generate complete scenarios applicable to the test. The validation process involves using the SVL Simulator and the Baidu Apollo joint simulation platform to evaluate the AV’s driving behavior and interactions with PTWs. Results This study evaluates model generation results by comparing distributions using Wasserstein distance as an indicator. The generator converges after approximately 200 epochs, with the iterator converging quickly. Subsequently, 10,000 new scenes are then generated. The distribution of several key parameters in the generated scenes can be found to approximate that of the original scenes. After sampling, the usability of generated scenarios is 64.76%. Virtual simulations confirm the effectiveness of the scenario generation method, with a generated scenario crash rate of 16.50% closely reflecting the original rate of 15.0%, showcasing the method’s capacity to produce realistic and hazardous scenarios. Conclusions The experimental results suggest that these scenarios exhibit a level of risk similar to the original crashes and are effective for testing AVs. Consequently, the generated scenarios enhance the diversity of the scenario library and accelerate the overall testing process of AVs.

RealFill: Reference-Driven Generation for Authentic Image Completion

Recent advances in generative imagery have brought forth outpainting and inpainting models that can produce high-quality, plausible image content in unknown regions. However, the content these models hallucinate is necessarily inauthentic, since they are unaware of the true scene. In this work, we propose RealFill, a novel generative approach for image completion that fills in missing regions of an image with the content that should have been there. RealFill is a generative inpainting model that is personalized using only a few reference images of a scene. These reference images do not have to be aligned with the target image, and can be taken with drastically varying viewpoints, lighting conditions, camera apertures, or image styles. Once personalized, RealFill is able to complete a target image with visually compelling contents that are faithful to the original scene. We evaluate RealFill on a new image completion benchmark that covers a set of diverse and challenging scenarios, and find that it outperforms existing approaches by a large margin. Project page: https://realfill.github.io.

Animal sauve enfant abandonné : quand la topique animalière vient éclairer autrement les topoï répertoriés (à partir du Shâhnâmeh de Ferdousi)

<p>The article continues with the post entitled "Poetic memory or the use of topos "ENFANT ABANDONNE" in Milan Kundera’s The unbearable lightness of being " published on the SATOR research notebook by Madeleine Jeay, in which she shows that Kundera works a topical situation corresponding to a number of topoi listed in SATORBASE as ADOPTER ENFANT TROUVE, DESOBEIR ORDRE DE TROUVER ENFANT, ENLEVER SECRETEMENT ENFANT etc. The article continues this reflection and focusses on the animal topic in the Shah Nameh of Ferdousi, reflecting on the role of the Simorgh (this mythical bird that plays an essential role in Persian culture and that is found in the Manteq-et-their (The dialogue of the birds) of Farid eddine Attar and in the metaphysical and philosophical texts of Avicenna and Sohravardi. In the Shah Nameh the Simorgh rescues Zal, abandoned as a child by his father, King Sam, and raises him until the latter, now an adult, finds his father who grants him the royal filiation. And the child is fed at night by a gazelle. Such topical configurations that could be called ANIMAL SAVES CHILD ABANDONS are very widespread. We question the original scene of a close relationship between human sovereignty and animal sovereignty, the part of sovereignty being equally distributed between man and animal which is here the inseparable adjuvant. The methodological question of topical naming is also addressed.</p>

A hybrid data fusion approach with twin CNN architecture for enhancing image source identification in IoT environment

AbstractWith the proliferation of digital devices in internet of things (IoT) environment featuring advanced visual capabilities, the task of Image Source Identification (ISI) has become increasingly vital for legal purposes, ensuring the verification of image authenticity and integrity, as well as identifying the device responsible for capturing the original scene. Over the past few decades, researchers have employed both traditional and machine‐learning methods to classify image sources. In the current landscape, data‐driven approaches leveraging deep learning models have emerged as powerful tools for achieving higher accuracy and precision in source prediction. The primary focus of this research is to address the complexities arising from diverse image sources and variable quality in IoT‐generated multimedia data. To achieve this, a Hybrid Data Fusion Approach is introduced, leveraging multiple sources of information to bolster the accuracy and robustness of ISI. This fusion methodology integrates diverse data streams from IoT devices, including metadata, sensor information, and contextual data, amalgamating them into a comprehensive data set for analysis. This study introduces an innovative approach to ISI through the implementation of a Twin Convolutional Neural Network Architecture (TCA) aimed at enhancing the efficacy of source classification. In TCA, the first CNN architecture, referred to as DnCNN, is employed to eliminate noise from the original data set, generating 256 × 256 patches for both training and testing. Subsequently, the second CNN architecture is employed to classify images based on features extracted from various convolutional layers using a 3 × 3 filter, thereby enhancing prediction efficiency. The proposed model demonstrates exceptional accuracy in effectively classifying image sources, showcasing its potential as a robust solution in the realm of ISI.

The Scene of the Council Hall: the Storyline and the Structure of the Grand Final Scene in the Second Edition of Verdi’s Opera Simon Boccanegra

The comparison of different authorial redactions of operas presents one of the through themes of world Verdi studies. The opera Simon Boccanegra holds a unique position, since the second version of the opera was created by Verdi almost a quarter of a century following the first. The depth of the changes, first of all, in the final scene of the first act, for which a new interpretation of the content was found is stipulated by such a temporal distance. Also important is the fact that the edition of Simon Boccanegra became the first large-scale collaborative work of Verdi and Arrigo Boito. In the two versions of the opera, the interconnection between the storyline and the structural laws is examined in the context of the general evolution of Verdi’s pivotal final scenes. In the original version, the composer still followed the conventional traditions, having created the most conservatively structured scene of that kind of all of his works from the 1850s. In the version of 1881, the final scene was transformed into one of the boldest and most original mass scenes in Verdi’s music. While initially the introduction of the episode missing in the play by Antonio García Gutiérrez (the festivities in honor of the anniversary of Simon Boccanegra having been elected as the doge of Genoa) was stipulated by the tradition of the mass grand final scene, the typical structure of which was in many ways “prompted” the scenario, in the second redaction, already the storyline and the verbal text of the scene of the Council session predetermined its musical-dramaturgical structure. However, even in this innovative final scene, Verdi does not abandon certain attributes of the typical structure. The result was a paradoxically organic combination of tradition and innovation of the scene in the Council Hall, which has become one of the pinnacles of the combined artistic work of the outstanding musician and the playwright.

Determination of the SO/PHI-HRT wavefront degradation using multiple defocused images

Context. The Polarimetric and Helioseismic Imager on board the Solar Orbiter mission (SO/PHI) offers refocusing capabilities to cope with the strongly varying thermal environment of the optical system along the spacecraft’s elliptical orbit. The series of images recorded during in-flight focus calibrations can be employed for phase diversity analyses. Aims. In this work we infer the wavefront degradation caused by the thermo-optical effects in the High Resolution Telescope (HRT) from images taken during the fine and coarse focus scans performed in the commissioning phase of the instrument. The difference between these two series of images are mainly related to the employed defocused step (smaller for the fine scans) and the signal-to-noise ratio (higher for the coarse scans). We use the retrieved wavefronts to reconstruct the original scene observed during the calibration of the instrument. Methods. We applied a generalized phase diversity algorithm that allowed us to use several images taken with different amounts of defocus to sense the wavefront degradation caused by the instrument. The algorithm also uses information from both the inferred wavefront and the series of images to restore the solar scene. Results. We find that most of the retrieved Zernike coefficients tend to converge to the same value when increasing the number of images employed for PD for both the fine and the coarse focusing scans. The restored scenes also show signs of convergence, and the merit function is minimized more as K increases. Apart from a defocus, the inferred wavefronts are consistent for the two datasets (λ/10 − λ/11). For the fine scan images, the quiet-sun contrast improves from 4.5% for the original focused image up to about 10%. For the coarse scan images, the contrast of the restored scene is as high as 11%.

Research Progress of Large Viewing‐Area 3D Holographic Near‐Eye Display

AbstractWith the development of the information industry, the wearable three‐dimensional (3D) near‐eye display that can provide complete visual information has attracted increasing attention. Compared with other existing 3D display technologies, the holographic display is a promising true 3D display technology that can fully record and reproduce the original 3D scenes. The ability to present a continuous parallax and depth perception gives holography technology an unprecedented promise in the next‐generation 3D near‐eye display. However, currently available holographic near‐eye display systems can only provide a limited viewing area (including the field of view and eyebox), greatly restricting its practical application. Focusing on the development requirements of large viewing‐area 3D holographic near‐eye display, the reasons are introduced first why the viewing area of current holographic displays is limited and then the existing solutions from different aspects, including multiplexing‐based methods, steering viewing window methods, aperiodic wavefront modulation methods, and some emerging holographic display technologies are summarized. These various solutions are analyzed respectively, and their representative works are investigated. In addition, the different applications of holography in the near‐eye display system are also discussed. Finally, the advantages and disadvantages of these techniques are compared and perspectives are given.

Spectrum Extension of a Real-Aperture Microwave Radiometer Using a Spectrum Extension Convolutional Neural Network for Spatial Resolution Enhancement

Enhancing the spatial resolution of real-aperture microwave radiometers is an essential research topic. The accuracy of the numerical values of brightness temperatures (BTs) observed using microwave radiometers directly affects the precision of the retrieval of marine environmental parameters. Hence, ensuring the accuracy of the enhanced brightness temperature values is of paramount importance when striving to enhance spatial resolution. A spectrum extension (SE) method is proposed in this paper, which restores the suppressed high-frequency components in the scene BT spectrum through frequency domain transformation and calculations, specifically, dividing the observed BT spectrum by the conjugate of the antenna pattern spectrum and applying a Taylor approximation to suppress error amplification, thereby extending the observed BT spectrum. By using a convolutional neural network to correct errors in the calculated spectrum and then reconstructing the BT through inverse fast Fourier transform (IFFT), the enhanced BTs are obtained. Since the extended BT spectrum contains more high-frequency components, namely, the spectrum is closer to that of the original scene BT, the reconstructed BT not only achieves an enhancement in spatial resolution, but also an improvement in the accuracy of BT values. Both the results from simulated data and satellite-measured data processing illustrate that the SE method is able to enhance the spatial resolution of real-aperture microwave radiometers and concurrently improve the accuracy of BT values.

Training Methods of Multi-Label Prediction Classifiers for Hyperspectral Remote Sensing Images

Hyperspectral remote sensing images, with their amalgamation of spectral richness and geometric precision, encapsulate intricate, non-linear information that poses significant challenges to traditional machine learning methodologies. Deep learning techniques, recognised for their superior representation learning capabilities, exhibit enhanced proficiency in managing such intricate data. In this study, we introduce a novel approach in hyperspectral image analysis focusing on multi-label, patch-level classification, as opposed to applications in the literature concentrating predominantly on single-label, pixel-level classification for hyperspectral remote sensing images. The proposed model comprises a two-component deep learning network and employs patches of hyperspectral remote sensing scenes with reduced spatial dimensions yet with a complete spectral depth derived from the original scene. Additionally, this work explores three distinct training schemes for our network: Iterative, Joint, and Cascade. Empirical evidence suggests the Joint approach as the optimal strategy, but it requires an extensive search to ascertain the optimal weight combination of the loss constituents. The Iterative scheme facilitates feature sharing between the network components from the early phases of training and demonstrates superior performance with complex, multi-labelled data. Subsequent analysis reveals that models with varying architectures, when trained on patches derived and annotated per our proposed single-label sampling procedure, exhibit commendable performance.

Distribution Network Optimization Dispatch Method Based on Renewable Energy Scenario Reduction

The scenario analysis method is an important method to adapt to the optimal dispatch of power systems with a high percentage of new energy sources. As a hot research topic in scenario analysis methods, the significance of scenario reduction is to describe many complex scenarios that feature a small number of representative scenarios to achieve the purpose of reducing computational complexity. In this paper, a distribution network optimization dispatch method based on the hierarchical clustering algorithm for scenario reduction is proposed considering wind power output and photovoltaic output. Firstly, the original scenes are quickly categorized because of the hierarchical clustering algorithm to derive typical scenarios. Secondly, an optimal dispatch model of the distribution network containing renewable energy is established. Finally, the arithmetic example is verified by using the IEEE33 distribution network, and the experimental results validate the validity and superiority of the proposed scenario reduction method.

Foreign Movies and TV Dramas as the Source of Political Argot in an Authoritarian Context: Memes and Creative Resistance in Chinese Social Media

ABSTRACT In the Web 2.0 era, memes have emerged as playful and versatile forms of communication in the daily interactions of social media users. In countries that embrace digital authoritarianism, memes have been deployed as a vehicle of political communication under the online surveillance system. Based on long-term observation of Chinese cyberspace, the author adopts a deep China approach and conducts multimodal critical discourse analysis (MCDA) of a series of political memes which poach transnational TV series and movie scenarios or characters. The remix of symbols from foreign TV dramas and movies with indigenous Chinese symbols devotes to three fundamental dimensions of meaning-making strategies: direct captures, reproduction of the original scene, and the reworking of specific elements. By producing, deploying, and proliferating ironic political memes, Chinese netizens have the capacity to engage in innovative discursive guerrilla wars under the rapidly evolving landscape of surveillance.

Estimation of Air Light With Deep Learning for a Near Real-Time Image Dehazing System

Haze which can be created by natural or synthetic factors, degrades the visual quality and human sight distance. Visible objects become invisible or scarcely visible. The physics of the degrading function due to haze has been modelled by Atmospheric Light Scattering (ALS) Model. Therefore, from a single hazy image, by using proper methods, it is possible to recover the original scene. In dehazing methods, which solve the ALS function, there are basically two steps: First one is the estimation of the air light present at the time of the image capturing and the second one is the estimation of transmission of the corresponding scene. One of the most effective method which is used for air light estimation is QuadTree decomposition. For this method, tests show that the most amount of the dehazing time is consumed to estimate the air light. For the case of High Definition (HD) imagery, the estimation of air light consumes huge time. Therefore, it cannot be possible to achieve a real-time or near real-time dehazing on traditional hardware. In this study, a novel convolutional neural network model is developed to estimate the air light directly from the hazy image quickly. The estimated air light then is used with Atmospheric Light Scattering model to handle the recovered image. Results show that the time cost is reduced by 56.0% and 65% for image resolutions of (640x480) and (1920x1080) compared to the QuadTree Decomposition method used in ALS based dehazing methods, without losing the visual quality of the dehazed image.

Graph laplacian regularization with sparse coding in secure image restoration and representation for Internet of Things

Image restoration (IR) attempts to recreate the original (ideal) scene from a degraded observation. The goal of image restoration is to avert the process of image deterioration that happens during image acquisition and processing. Blurring and noise are two common types of picture capture degradation. When the blurring function is unknown, the image restoration issue is referred to as blind image restoration. Existing methods of image restoration do not provide efficient results due to time complexity, computational complexity, large-scale scaling factor and limited input. Therefore, in order to overcome those problems, an Improved Graph Laplacian Regularization with Sparse Coding by integrating Internet of Things (IoT) is developed in this research. The process of removing de-noised portions of image to texture layer and cartoon layer is carried out by Morphological Component Analysis (MCA). An improved Graph Laplacian regularized method and Simultaneous Sparse Coding with Gaussian Scale Mixture (SSC-GSM) methods is performed on texture layer, cartoon layer and restored image. A Levin’s dataset and a real-time dataset such as industrial manufacturing products are analyzed in this research. The proposed Graph Laplacian algorithm and sparse coding model are better efficient in identifying sharper texture information and getting a restored image. The proposed method proves that it has better performance in terms of Peak Signal-to-Noise Ratio (PSNR) of 35.82 and Structural Similarity Index Measure (SSIM) of 0.94 than the existing methods.

Aging attenuates the memory advantage for unexpected objects in real-world scenes

Across the adult lifespan memory processes are subject to pronounced changes. Prior knowledge and expectations might critically shape functional differences; however, corresponding findings have remained ambiguous so far. Here, we chose a tailored approach to scrutinize how schema (in-)congruencies affect older and younger adults’ memory for objects embedded in real-world scenes, a scenario close to everyday life memory demands. A sample of 23 older (52–81 years) and 23 younger adults (18–38 years) freely viewed 60 photographs of scenes in which target objects were included that were either congruent or incongruent with the given context information. After a delay, recognition performance for those objects was determined. In addition, recognized objects had to be matched to the scene context in which they were previously presented. While we found schema violations beneficial for object recognition across age groups, the advantage was significantly less pronounced in older adults. We moreover observed an age-related congruency bias for matching objects to their original scene context. Our findings support a critical role of predictive processes for age-related memory differences and indicate enhanced weighting of predictions with age. We suggest that recent predictive processing theories provide a particularly useful framework to elaborate on age-related functional vulnerabilities as well as stability.

Assessing Precision and Dependability of Reconstructed Three-Dimensional Modeling for Vehicles at Crash Scenes using Unmanned Aircraft System

This study focuses on the accuracy assessment of 3D reconstructions of crime scenes using Unmanned Aircraft Systems (UAS) and Terrestrial Laser Scanners (TLS) data for forensic crash investigation. Forensic crash investigation involves meticulously analyzing physical evidence, vehicles, and human factors in road collisions to determine the sequence of events. Preserving the original state of the crash scene before cleaning is essential for accurate forensic analysis. However, this preservation process can disrupt normal activities and demand considerable time. Geomatic technology, specifically UAS or drones, offers a potential solution for efficient and precise forensic mapping. The application of UAS technology enables swift data collection, leading to cost savings, enhanced safety, and data utilization. This study aims to assess the suitability of UAS techniques for forensic mapping, encompassing both relative and absolute accuracy. This research uses a UAS to rapidly and comprehensively capture evidence from a simulated crash site using predefined flight paths. The acquired image data is then processed utilizing Agisoft Metashape software, generating a detailed 3D model of the crash scene. This model can be enriched with annotations, measurements, and pertinent information. A comparative analysis is performed by preparing a table that contrasts the absolute and relative accuracy of UAS-collected data with that obtained from TLS, which serves as a benchmark. The results reveal that the UAS demonstrates a relative accuracy Root Mean Square Error (RMSE) of approximately ±4.1 cm compared to TLS. Concerning absolute precision, the UAS-produced RMSE values are determined as ±0.20719 for the X coordinate, ±0.164 for the Y coordinate, and ±0.001584 for the Z coordinate compared to GNSS data, which functions as the benchmark. The utilization of UAS technology offers a non-invasive measurement approach that eliminates direct physical contact between the operator and the documented object. This non-intrusive method ensures the preservation of the original scene characteristics and has shown its superiority over conventional approaches in managing crash scenes. Overall, this study underscores the potential of UAS technology in accurately reconstructing crime scenes for forensic investigation purposes.

Weakly supervised semantic segmentation for point cloud based on view-based adversarial training and self-attention fusion

Traditional methods of weakly supervised semantic segmentation (WSsegmentation) for point cloud scenes have several limitations including limited precision and difficulty in handling complex scenes due to imprecise labels or partial annotations. To address these issues, we perform view-based adversarial training on the original point cloud scene samples through view resampling and Gaussian noise perturbation to reduce overfitting. Combining a self-attention mechanism with multi-layer perceptrons and point cloud segmentation strategy, we can perform dimensionality enhancement and dimensionality reduction operations to better capture the local features of point cloud data. Finally, we can obtain the semantic segmentation results of point cloud scene by fusing local and global semantic features. In the design of the network loss function, we combine the Siamese loss and the smoothness loss and the cross-entropy loss to improve the ability and fidelity of semantic networks. Specifically, the Siamese loss is used to compute the distance between different augmented point cloud data in their feature embedding space and the smoothness loss is used to penalize the discontinuity of semantic information between adjacent regions. The proposed weakly supervised segmentation network achieves an overall segmentation accuracy close to fully supervised segmentation methods and outperforms most of the existing weakly supervised segmentation methods by 5% to 10% in scene segmentation in terms of mIoU on S3DIS, ShapeNet, and PartNet datasets. Extensive experiments demonstrate the robustness, effectiveness, and generalization of the proposed point cloud segmentation network.

Priors Guided Extreme Underwater Image Compression for Machine Vision and Human Vision

A low bit-rate compression is required for underwater images due to the limited bandwidth in underwater acoustic communication, which limits performances of both machine analysis and human perception in most underwater applications. Few existing compression methods consider the unique characteristics of underwater images such as color shift and haze effect to better fulfill requirements of various applications under low bit-rates. To address this problem, we propose a novel extreme underwater image compression framework, which can provide scalability to support machine vision and human vision with the assistance of underwater priors. Specifically, the base layer is composed of a feature extractor and a generator, where global structural edges and high-level features of regions with a significant impact on machine analysis are extracted and used for reconstructing a feature-matching image for analysis purpose. Considering the negative influence of underwater imaging processes on machine vision, in this article, a feature degradation removal module guided by underwater priors is proposed to alleviate feature-level degradation via taking analysis-friendly enhanced images as auxiliary information. As for the enhancement layer aiming for human vision, the residual between the original image and the reconstruction from base layer is compressed. A feature attention block and a background light recovery block are designed utilizing features extracted from enhanced images and the underwater before further recovering the original scene with a good perception quality under low bit-rates. Experimental results demonstrate the superiority of our framework in both machine vision tasks and perception quality compared with traditional compression methods and learned-based methods.

Exploiting Light Polarization for Deep HDR Imaging from a Single Exposure

In computational photography, high dynamic range (HDR) imaging refers to the family of techniques used to recover a wider range of intensity values compared to the limited range provided by standard sensors. Classical techniques consist of acquiring a scene-varying exposure to compensate for saturated and underexposed regions, followed by a non-linear compression of intensity values called tone mapping. Recently, there has been a growing interest in estimating HDR images from a single exposure. Some methods exploit data-driven models trained to estimate values outside the camera’s visible intensity levels. Others make use of polarimetric cameras to reconstruct HDR information without exposure bracketing. In this paper, we present a novel HDR reconstruction method that employs a single PFA (polarimetric filter array) camera with an additional external polarizer to increase the scene’s dynamic range across the acquired channels and to mimic different exposures. Our contribution consists of a pipeline that effectively combines standard HDR algorithms based on bracketing and data-driven solutions designed to work with polarimetric images. In this regard, we present a novel CNN (convolutional neural network) model that exploits the underlying mosaiced pattern of the PFA in combination with the external polarizer to estimate the original scene properties, and a second model designed to further improve the final tone mapping step. The combination of such techniques enables us to take advantage of the light attenuation given by the filters while producing an accurate reconstruction. We present an extensive experimental section in which we validate the proposed method on both synthetic and real-world datasets specifically acquired for the task. Quantitative and qualitative results show the effectiveness of the approach when compared to state-of-the-art methods. In particular, our technique exhibits a PSNR (peak signal-to-noise ratio) on the whole test set equal to 23 dB, which is 18% better with respect to the second-best alternative.