Images of outdoor scenes captured in bad weather suffer from poor contrast. Under conditions, the light reaching a camera is severely scattered by atmosphere. The resulting decay contrast varies across scene and exponential depths points. Therefore, traditional space invariant image processing techniques are not sufficient to remove effects images. We present physics-based model that describes appearances uniform conditions. Changes intensities points under different conditions provide...

We present an approach to easily remove the effects of haze from images. It is based on fact that usually airlight scattered by atmospheric particles partially polarized. Polarization filtering alone cannot effects, except in restricted situations. Our method, however, works under a wide range and viewing conditions. analyze image formation process, taking into account polarization scattering. then invert process enable removal The method can be used with as few two images taken through...

Current vision systems are designed to perform in clear weather. Needless say, any outdoor application, there is no escape from "bad" Ultimately, computer must include mechanisms that enable them function (even if somewhat less reliably) the presence of haze, fog, rain, hail and snow. We begin by studying visual manifestations different weather conditions. For this, we draw on what already known about atmospheric optics. Next, identify effects caused bad can be turned our advantage. Since...

Conventional vision systems are designed to perform in clear weather. However, any outdoor system is incomplete without mechanisms that guarantee satisfactory performance under poor weather conditions. It known the atmosphere can significantly alter light energy reaching an observer. Therefore, atmospheric scattering models must be used make robust bad In this paper, we develop a geometric framework for analyzing chromatic effects of scattering. First, study simple color model and verify it...

We present an approach for easily removing the effects of haze from passively acquired images. Our is based on fact that usually natural illuminating light scattered by atmospheric particles (airlight) partially polarized. Optical filtering alone cannot remove effects, except in restricted situations. method, however, stems physics-based analysis works under a wide range and viewing conditions, even if polarization low. The does not rely specific scattering models such as Rayleigh knowledge...

We present a vision system that automatically predicts yield in vineyards accurately and with high resolution. Yield estimation traditionally requires tedious hand measurement, which is destructive, sparse sampling, inaccurate. Our method efficient, high‐resolution, it the first such evaluated realistic experimentation over several years hundreds of vines spread acres different vineyards. Other existing research limited to small test sets 10 or less, just isolated grape clusters, tightly...

Images of outdoor scenes captured in bad weather suffer from poor contrast. Under conditions, the light reaching a camera is severely scattered by atmosphere. The resulting decay contrast varies across scene and exponential depths points. Therefore, traditional space invariant image processing techniques are not sufficient to remove effects images. In this paper, we present fast physics-based method compute structure hence restore two or more images taken previous techniques, our does...

Virtually all structured light methods assume that the scene and sources are immersed in pure air is neither scattered nor absorbed. Recently, however, lighting has found growing application underwater aerial imaging, where scattering effects cannot be ignored. In this paper, we present a comprehensive analysis of two representative - stripe range scanning photometric stereo presence scattering. For both methods, derive physical models for appearances surface medium. Based on these models,...

The harvest yield in vineyards can vary significantly from year to and also spatially within plots due variations climate, soil conditions pests. Fine grained knowledge of crop yields allow viticulturists better manage their vineyards. current industry practice for prediction is destructive, expensive sparse - during the growing season samples are taken extrapolated determine overall yield. We present an automated method that uses computer vision detect count grape berries. could potentially...

Given a single outdoor image, we present method for estimating the likely illumination conditions of scene. In particular, compute probability distribution over sun position and visibility. The relies on combination weak cues that can be extracted from different portions image: sky, vertical surfaces, ground. While no cue reliably estimate by itself, each one reinforce others to yield more robust estimate. This is combined with data-driven prior computed dataset 6 million Internet photos. We...

Global illumination effects such as inter-reflections, diffusion and sub-surface scattering severely degrade the performance of structured light-based 3D scanning. In this paper, we analyze errors caused by global in shape recovery. Based on analysis, design light patterns that are resilient to individual using simple logical operations tools from combinatorial mathematics. Scenes exhibiting multiple phenomena handled combining results a small ensemble patterns. This combination also allows...

Depth sensing is crucial for 3D reconstruction and scene understanding. Active depth sensors provide dense metric measurements, but often suffer from limitations such as restricted operating ranges, low spatial resolution, sensor interference, high power consumption. In this paper, we propose a deep learning (DL) method to estimate per-pixel its uncertainty continuously monocular video stream, with the goal of effectively turning an RGB camera into RGB-D camera. Unlike prior DL-based...

We present a novel theory of Fermat paths light between known visible scene and an unknown object not in the line sight transient camera. These either obey specular reflection or are reflected by object's boundary, hence encode shape hidden object. prove that correspond to discontinuities measurements. then derive constraint relates spatial derivatives path lengths at these surface normal. Based on this theory, we algorithm, called Flow, estimate non-line-of-sight Our method allows, for...

Programmable coding of light between a source and sensor has led to several important results in computational illumination, imaging display. Little is known, however, about how utilize energy most effectively, especially for applications live imaging. In this paper, we derive novel framework maximize efficiency by "homogeneous matrix factorization" that respects the physical constraints many mechanisms (DMDs/LCDs, lasers, etc. ). We demonstrate energy-efficient using two prototypes based on...

We consider the task of 3D pose estimation and tracking multiple people seen in an arbitrary number camera feeds. propose TesseTrack <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> , a novel top-down approach that simultaneously reasons about individuals' body joint reconstructions associations space time single end-to-end learnable framework. At core our is spatio-temporal formulation operates common voxelized feature aggregated from...

Virtually all methods in image processing and computer vision, for removing weather effects from images, assume single scattering of light by particles the atmosphere. In reality, multiple are significant. A common manifestation is appearance glows around sources bad weather. Modeling critical to understanding complex on hence essential improving performance outdoor vision systems. We develop a new physics-based model rays as they travel source an observer. This valid various conditions...

We consider real-time rendering of scenes in participating media, capturing the effects light scattering fog, mist and haze. While a number sophisticated approaches based on Monte Carlo finite element simulation have been developed, those methods do not work at interactive rates. The most common are essentially simple variants OpenGL fog model. easy to use specify, that model excludes many important qualitative like glows around sources, impact volumetric appearance surfaces such as...

The visual world around us displays a rich set of volumetric effects due to participating media. appearance these media is governed by several physical properties such as particle densities, shapes and sizes, which must be input (directly or indirectly) rendering algorithm generate realistic images. While there has been significant progress in developing techniques (for instance, Monte Carlo methods analytic approximations), are very few that measure estimate for relevance computer graphics....