Light field

About: Light field is a research topic. Over the lifetime, 5357 publications have been published within this topic receiving 87424 citations.

Memory-oriented Decoder for Light Field Salient Object Detection

Abstract: Light field data have been demonstrated in favor of many tasks in computer vision, but existing works about light field saliency detection still rely on hand-crafted features. In this paper, we present a deep-learning-based method where a novel memory-oriented decoder is tailored for light field saliency detection. Our goal is to deeply explore and comprehensively exploit internal correlation of focal slices for accurate prediction by designing feature fusion and integration mechanisms. The success of our method is demonstrated by achieving the state of the art on three datasets. We present this problem in a way that is accessible to members of the community and provide a large-scale light field dataset that facilitates comparisons across algorithms. The code and dataset will be made publicly available.

Short-term variability of the underwater light field in the oligotrophic ocean in response to surface waves and clouds

Abstract: Examples of short-term variations of the underwater light field associated with surface-wave focusing and cloud cover, and occurring at intermediate depths of the euphotic zone in the oligotrophic ocean are discussed. Time series of water pressure, spectral downwelling irradiance (Ed), and spectral upwelling radiance (Lu) at 412, 443, 490, 510, 555, 665, and 683 nm were acquired with moored instruments. The mooring was deployed in the Sargasso Sea near Bermuda in September 1994. The measurements were made at 15 and 35 m once every hour from 6 a.m. to 8 p.m. local time, with a 6 Hz sampling rate over time periods of 2 or 5 min duration. Time series of Ed, Lu, and water pressure were subject to spectral analysis. Under clear sky conditions, the power spectra of Ed at the blue/green wavelengths exhibited distinct maxima due to focusing of sunlight by surface waves. These maxima were located at a frequency higher than the dominant frequency in the power spectrum of water pressure variations. Similar maxima were present in the power spectra of Lu, but the position of the peak was dependent on light wavelength. The coefficient of variation for Ed and Lu depended on light wavelength and increased towards the red wavelengths. This coefficient was much smaller for Lu than for Ed, which reflects the fact that the upwelling light field is less affected by wave focusing than the downwelling light field. At depths of our measurements, the light fluctuations in the red can be attributed to variations in inelastically scattered radiation, that is, the fluorescence of phytoplankton cells and Raman scattering. This is because the fluoresced light and Raman scattered light in the red follow fluctuations that occur at shorter excitation wavelengths. When clouds covered the sky, we observed a decrease of the coefficient of variation for Ed and Lu. In this case, the fluctuations of light were well correlated with variations of water pressure associated with water surface displacement due to the wave motion.

On the number of samples needed in light field rendering with constant-depth assumption

Abstract: While several image-based rendering techniques have been proposed to successfully render scenes/objects from a large collection (e.g., thousands) of images without explicitly recovering 3D structures, the minimum number of images needed to achieve a satisfactory rendering result remains an open problem. This paper is the first attempt to investigate the lower bound for the number of samples needed in the Lumigraph/light field rendering. To simplify the analysis, we consider an ideal scene with only a point that is between a minimum and a maximum range. Furthermore, constant-depth assumption and bilinear interpolation are used for rendering. The constant-depth assumption serves to choose "nearby" rays for interpolation. Our criterion to determine the lower bound is to avoid horizontal and vertical double images, which are caused by interpolation using multiple nearby rays. This criterion is based on the causality requirement in scale-space theory, i.e., no "spurious details" should be generated while smoothing. Using this criterion, closed-form solutions of lower bounds are obtained for both 3D plenoptic function (Concentric Mosaics) and 4D plenoptic function (light field). The bounds are derived completely from the aspect of geometry and are closely related to the resolution of the camera and the depth range of the scene. These tower bounds are further verified by our experimental results.

A spectrally averaged model of light penetration and photosynthesis

Abstract: A model was developed which predicts the daily photosynthesis of a vertical pigment profile divided into a number of homogeneous layers. A spectral model (irradiance divided into a large number of wavebands) was used to derive simple empirical equations for calculating spectrally averaged values of two parameters- the vertical light attenuation coefficient and the chlorophyll-specific absorption of algaefor each layer as a function of its pigment content and position in the water column. The empirical equations are not dependent on the layer depths chosen, i.e. the same equations can be used for any given set of depths. The spectrally averaged parameters can be used with analytic integrals to give a computationally rapid and accurate result. The model is therefore ideally suited for general circulation models. When choosing models of light penetration and photosynthesis for a given application, one should consider the need for both accuracy and computational efficiency of predictions. Sometimes both are required, particularly when embedding ecosystem models in general circulation models (GCMs) of the world’s oceans. The penetration of photosynthetically active radiation (PAR) into seawater depends on the nature of the light field and the absorbing properties of water and substances present in it. Photosynthesis depends on the distribution of algal pigments and their photosynthetic response to PAR. The complicated nature of these processes means that achieving a balance between accuracy (achieved by introducing detail) and computational efficiency (achieved through simplicity) is not easy. Both light attenuation and photosynthesis are affected by the spectral properties of the light field. For example, red wavelengths are rapidly absorbed in the upper 10 m of the water column, whereas blue wavelengths penetrate much deeper (Jerlov 1968). Furthermore, light at different wavelengths is absorbed with different efficiencies by algae. The use of nonspectral models of light attenuation and photosynthesis can overestimate daily primary