Search

In this contribution, a flexible context-based security framework is proposed by exploring two types of context: distributed and local. While the former consists in processing information from a set of spatially distributed sources, the second accounts for the local environment surrounding the monitored system. The joint processing of these two types of information allows the identification of the anomaly cause, differentiating between natural and attack-related events, and the suggestion of the best mitigation strategy. In this work, the proposed framework is applied the Cyber Physical Systems scenario. More in detail, we focus on the distributed context analysis investigating the definition of a 2D representation of network traffic data. The suitability of four representation variables has been evaluated, and the variable selection has been performed.

Immersive visual technologies and Internet of things are considered the main paradigms in defining new models of immersive training for critical safety applications. Accordingly, matching appropriate devices and technologies with the training requirements and its individual, technological, and operational factors is mandatory to achieve the training goals. In this context, we investigate and classify different factors based on immersive technologies, emerging devices, and sensors. Furthermore, we propose a new model for cognitive immersive visual technologies for critical safety solutions. The proposed model is validated in a use case of a fire brigade intervention after an earthquake. The use of this model mitigates heterogeneity conflicts due to the variety of immersive technologies for a specific situation by respecting stakeholders' requirements. This paper helps future researchers, industrial and enterprise stockholders to select adequate immersive visual technologies, devices, and engines for creating further critical safety solutions.

Font matter of the Adjunct Proceedings of the ACM International Conference on Interactive Media Experiences: IMX 2022

Traditionally, the quality of a multimedia system was mainly assessed through the evaluation of its Quality of Service (QoS) that is by evaluating system parameters such as bandwidth, latency, jitter, throughput, transmission delay, availability, etc. However, these metrics often failed to capture the actual end-user perceived quality, which has prompted the development of the construct of Quality of Experience (QoE), widely understood as an interaction of the technical features of multimedia systems with perceptual, and cognitive/emotional factors involved in the interpretation of those features by users. This chapter addresses the open issues in the field of QoS and QoE assessments. First, the perceptual characteristics of the multiview content are analyzed, and then a survey on the existing approaches for QoS and QoE estimation is performed. The analysis is then focused on the subjective aspects of QoE assessment, by describing the standard methodologies currently used and new trends based on human factors research. Finally, the chapter offers a few guidelines for future research directions in the field.

Traditional acquisition systems record the scene from only one point of view while free-viewpoint television (FTV) or free-viewpoint video (FVV) allows the rendering of a complete representation of the scene. To this aim, it is necessary to use several input cameras, ideally as many as the possible viewing positions of the user. Due to the complexity and cost of such camera set up, a tractable approach consists in acquiring or transmitting only few viewpoints, while intermediate or missing points can be obtained by view interpolation. This is possible thanks to view synthesis techniques that exploit the geometric information of a scene, available for example in MVD (Multiview Video plus Depth) sequences. FTV is an important step towards interactivity since observers can freely change their point of view while exploring the scene. Such capability comes with new technological constraints that must be assessed not only in terms of perceived quality, but especially from the Quality of Experience (QoE) point of view [1]. In fact, FTV brings new issues in QoE assessment, due to the variety of use cases in which this technique can be employed. Furthermore, the free navigation capability is a completely new task in which user experience and QoE assessment methodology have both to be investigated, revisiting traditional approaches. In the following, an overview of the artefacts caused by virtual views generation, together with some options for content exploration are presented. Then, the state of the art on testing protocols and objective quality metrics to assess the quality of rendered FTV views is presented. Finally, we describe available datasets that could support the activities of further objective quality measures development.

The most emerging video technologies for cultural and social activities refer to 3D visual data processing and rendering. In fact, the use of 3D images and videos allows end-users to benefit from a depth perception of the recorded scene, being thus suitable for utilizations in entertainment, communication and telephony, training and learning, medicine, and so on. In the present paper, we take into account some characteristics of the human visual perception to define a joint motion-disparity processing approach, which is employed to enhance 3D video contents by reducing the feeling of artificial clarity experienced by the viewers, thus resulting in improved user acceptance and satisfaction. Experimental results have been conducted in terms of subjective tests, in order to evaluate the effectiveness of the proposed 3D video enhancement scheme.

In this paper, a new image database, TID2008, for evaluation of full-reference visual quality assessment metrics is described. It contains 1700 test images (25 reference images, 17 types of distortions for each reference image, 4 different levels of each type of distortion). A comparative analysis of TID20008 and its nearest analog LIVE Database (only 5 types of distortions, less statistical accuracy) is presented. Mean Opinion Scores (MOS) for TID2008 database have been obtained as a result of more than 800 experiments. During these tests, observers from three countries (Finland, Italy, and Ukraine) have carried out about 256000 individual human quality judgments. The obtained MOS can be used for effective testing of different visual quality metrics as well as for the design of new metrics. Using the designed image database, we have tested several known quality metrics. The values of rank correlations of Spearman and Kendall with the considered metrics and Mean Opinion Score obtained by exploiting TID2008 in experiments are presented. The metrics are verified for both full set of distorted test images in TID2008 and for particular subsets of TID2008 that include distortions most important for digital image processing applications. It is demonstrated that for the considered wide range of possible distortion types no existing metric performs well enough. In aggregate, the best results are provided by Multiscale Structural Similarity Index Metric (MSSIM). For this metric Spearman Correlation with MOS is 0.853, Kendall correlation with MOS is 0.654. Analysis has been also performed for particular subsets of distortion types. This analysis has shown that for most typical practical applications like image filtering and compression the metrics PSNR-HVS and PSNR-HVS-M produce reasonably good results (Spearman correlation is 0.929, Kendall correlation is 0.765). The designed test image database is available for downloading and utilization in scientific investigations.

Publication in the conference proceedings of EUSIPCO, Florence, Italy, 2006

Assessing the quality of images is a challenging task. To achieve this goal, images must be evaluated by a pool of subjects following a well-defined protocol or an objective quality metric must be defined. In this work, an objective quality metric based on deep neural network is proposed. The metric takes into account the human vision system by computing the saliency map and natural scene statistics features of the image under test. The neural network is composed by two modules: the convolutional layers and the regression units. The first one is trained by using preprocessed distorted images. The feature weights of the first module are smoothed by exploiting the estimated saliency map. The latter module is fit with the ground truth quality scores of the input image and the scaled feature weights obtained from first module by using visual sensitivity factor of image obtained using natural scene statistics features. The performances of the proposed metric have been evaluated by using four datasets: LIVEIQA, TID2013, CSIQ, and KADID10K. The achieved results show the effectiveness of the proposed system in closely matching the predicted quality scores with the ground truth ones.

The objective of a sound event detector is to recognize anomalies in an audio clip and return their onset and offset. However, detecting sound events in noisy environments is a challenging task. This is due to the fact that in a real audio signal several sound sources co-exist and that the characteristics of polyphonic audio are different from isolated recordings. It is also necessary to consider the presence of noise (e.g. thermal and environmental). In this contribution, we present a sound anomaly detection system based on a fully convolutional network which exploits image spatial filtering and an Atrous Spatial Pyramid Pooling module. To cope with the lack of datasets specifically designed for sound event detection, a dataset for the specific application of noisy bus environments has been designed. The dataset has been obtained by mixing background audio files, recorded in a real environment, with anomalous events extracted from monophonic collections of labelled audios. The performances of the proposed system have been evaluated through segment-based metrics such as error rate, recall, and F1-Score. Moreover, robustness and precision have been evaluated through four different tests. The analysis of the results shows that the proposed sound event detector outperforms both state-of-the-art methods and general purpose deep learning-solutions.

Immersive media, such as free view point video and 360° video, are expected to be dominant as broadcasting services. The light field (LF) imaging is being considered as a next generation imaging technology offering the possibility to provide new services, including six degree-of-freedom video. The drawback of this technology is in the size of the generated content thus requiring novel compression systems and the design of ad-hoc methodologies for evaluating the perceived quality. In this paper, the relation between depth map quality and overall quality of LF image is studied. Next, a reduced reference quality assessment metric for LF images is presented. To predict the quality of distorted LF images, the measure of distortion in the depth map is exploited. To test and validate the proposed framework, a subjective experiment has been performed, and a LF image quality dataset has been created. The dataset is also used for evaluating the performance of state-of-the-art quality metrics, when applied to LF images. The achieved results evidence that the estimated quality score by the proposed framework has a significant correlation with subjective quality rating. Consequently, reference data can be delivered to the clients thus allowing the local estimation of the perceived quality of service.

Recent years have witnessed a rapid growth of user generated videos thanks to the availability of affordable video recording devices and the extreme popularity of social media platforms. Accordingly, there is a challenge for designing highly efficient video quality assessment models to monitor, control, and optimize this content. In this contribution, a novel no-reference video quality metric for user generated video is presented. It exploits the spatial and temporal information contained in the center patch of video frames and a Support Vector Regressor system for computing the objective score. Experimental results show the effectiveness of the proposed approach. To promote reproducible research and public evaluation, an implementation of RM3VQA has been made available online: https://github.com/pramitmazumdar/RM3VQA.

Data-sets play a crucial role in advancing the research. However, getting access to real-world data becomes difficult when it comes to critical infrastructures and more so if that data is being acquired for security research. In this work, a comprehensive dataset from a real-world smart electric grid testbed is collected and shared with the research community. A few of the unique features of the dataset and testbed are highlighted.