This paper presents the results of the depth estimation challenge for dense light fields, which took place at the second workshop on Light Fields for Computer Vision (LF4CV) in conjunction with CVPR 2017. The challenge consisted of submission to a recent benchmark [7], which allows a thorough performance analysis. While individual results are readily available on the benchmark web page http://www.lightfield-analysis.net, we take this opportunity to give a detailed overview of the current participants. Based on the algorithms submitted to our challenge, we develop a taxonomy of light field disparity estimation algorithms and give a report on the current state-of-the-art. In addition, we include more comparative metrics, and discuss the relative strengths and weaknesses of the algorithms. Thus, we obtain a snapshot of where light field algorithm development stands at the moment and identify aspects with potential for further improvement.

The transport sector is experiencing a fast and revolutionary development, moving towards the deployment of autonomous vehicles. Thus, the accuracy of the position information and the integrity of the navigation system have become key factors. In this paper, we address the problem of the enhancement of the integrity of the position provided by a Global Navigation Satellite System receiver by exploiting sensor fusion. To this aim, we estimate the lateral offset and heading of the vehicle with respect to a georeferenced roadway centerline from the images supplied by an on-board camera. Moreover, we perform integrity monitoring based on the implementation of the Solution Separation in the parity space. The numerical results indicate that the use of sensor fusion and digital map allows to attain a longitudinal Protection Level reduction with respect to the case in which sensor fusion is not exploited. More specifically, a decrease of about 70% is achieved when a single constellation is used, while reduction is less relevant, about 15%, when two constellations are employed.

Over the last few years applications based on the use of immersive environments, where physical and digital objects coexist and interact, have gained widespread attention. Thanks to the development of new visualization devices, even at low cost, and increasingly effective rendering and processing techniques, these applications are reaching a growing number of users. While the adoption of digital information makes it possible to provide immersive experiences in a number of different applications, there are still many unexplored aspects. In this work, a preliminary step to understand the impact of the scene content on human perception of the virtual 3D elements in a mixed reality has been performed. To this aim, a subjective test was designed and implemented to collect the reaction time of a set of users in a mixed reality application. In this test each user was asked to wear an augmented reality headset and to catch a virtual objects randomly appearing in the subject's field of view. We first estimated the detection accuracy through omitted, anticipated, and completed responses; then we related stimulus location, scene content and estimated accuracy. For this purpose, the area of stimulus presentation was divided into upper, lower, right, left, inner, and outer, to understand in which area responses were omitted and anticipated with respect to the central point of view. Experimental results show that, in addition to the saliency of the real scene, natural body gesture technology and limited field of view influenced human reaction time.

Global navigation satellite system (GNSS) positioning is essential for achieving absolute vehicular positioning in urban scenarios; however, it suffers from limited measurement redundancy and substantial faults caused by complex urban environments. In this work, we propose the subspace-based adaptive error modeling and fault detection and exclusion (FDE) method for pseudorange-based GNSS positioning in urban canyons, which integrates the adaptive error modeling into the FDE process and the positioning-solving process. Notably, we divide the pseudorange measurement space into subspaces regarding elevation angle and carrier-to-noise ratio (C/N0), each of which maintains a Gaussian mixture model (GMM) to adaptively characterize measurement error profiles. Results show that the proposed method has the ability to detect environmental changes. In addition, the proposed method outperforms the conventional FDE method with Gaussian assumptions, reducing the mean positioning error by 16 % and 9% in slightly and medium urbanized datasets, respectively. The impacts of step size (elevation angle and C/N0) and time window of the proposed method are discussed through controlled experiments.

The objective of this work is to evaluate the feasibility of the integration of visual sensors in the framework of GNSS-based vehicle localization systems. Undoubtedly GNSS plays a key role in vehicle localization. However, the impact of GNSS faults (e.g. satellite and constellation failure), signal degradation (e.g. ionospheric scintillations, multipath) and external threats (e.g. jamming and spoofing) need to be taken into account. To address this issue, we propose a three-level localization framework based on the use of additional visual sensors. Thanks to the visual input, the inaccurate GNSS measurements are refined by tracking georeferenced landmarks to obtain the vehicle absolute position. Moreover, the same sensors can be used to obtain the relative position and motion of the vehicle. Every time a new group of landmarks is detected, it is possible to repeat the absolute positioning process to cancel the drift accumulated by the relative localization procedure. This work represents the first step towards the realization of the proposed localization technique. Initial testing to evaluate its suitability has been carried out and is reported hereafter.