View-Based Strategies for 3D Object Recognition

A persistent issue of debate in the area of 3D object recognition concerns the nature of the experientially acquired object models in the primate visual system. One prominent proposal in this regard has expounded the use of object centered models, such as representations of the objects' 3D structures in a coordinate frame independent of the viewing parameters [Marr and Nishihara, 1978]. In contrast to this is another proposal which suggests that the viewing parameters encountered during the learning phase might be inextricably linked to subsequent performance on a recognition task [Tarr and Pinker, 1989; Poggio and Edelman, 1990]. The 'object model', according to this idea, is simply a collection of the sample views encountered during training. Given that object centered recognition strategies have the attractive feature of leading to viewpoint independence, they have garnered much of the research effort in the field of computational vision. Furthermore, since human recognition performance seems remarkably robust in the face of imaging variations [Ellis et al., 1989], it has often been implicitly assumed that the visual system employs an object centered strategy. In the present study we examine this assumption more closely. Our experimental results with a class of novel 3D structures strongly suggest the use of a view-based strategy by the human visual system even when it has the opportunity of constructing and using object-centered models. In fact, for our chosen class of objects, the results seem to support a stronger claim: 3D object recognition is 2D view-based.

Visual Recognition and Categorization on the Basis of Similarities to Multiple Class Prototypes

To recognize a previously seen object, the visual system must overcome the variability in the object's appearance caused by factors such as illumination and pose. Developments in computer vision suggest that it may be possible to counter the influence of these factors, by learning to interpolate between stored views of the target object, taken under representative combinations of viewing conditions. Daily life situations, however, typically require categorization, rather than recognition, of objects. Due to the open-ended character both of natural kinds and of artificial categories, categorization cannot rely on interpolation between stored examples. Nonetheless, knowledge of several representative members, or prototypes, of each of the categories of interest can still provide the necessary computational substrate for the categorization of new instances. The resulting representational scheme based on similarities to prototypes appears to be computationally viable, and is readily mapped onto the mechanisms of biological vision revealed by recent psychophysical and physiological studies.

Slow and Smooth: A Bayesian Theory for the Combination of Local Motion Signals in Human Vision

In order to estimate the motion of an object, the visual system needs to combine multiple local measurements, each of which carries some degree of ambiguity. We present a model of motion perception whereby measurements from different image regions are combined according to a Bayesian estimator --- the estimated motion maximizes the posterior probability assuming a prior favoring slow and smooth velocities. In reviewing a large number of previously published phenomena we find that the Bayesian estimator predicts a wide range of psychophysical results. This suggests that the seemingly complex set of illusions arise from a single computational strategy that is optimal under reasonable assumptions.

View-based approaches to spatial representation in human vision

In an immersive virtual environment, observers fail to notice the expansion of a room around them and consequently make gross errors when comparing the size of objects. This result is difficult to explain if the visual system continuously generates a 3-D model of the scene based on known baseline information from interocular separation or proprioception as the observer walks. An alternative is that observers use view-based methods to guide their actions and to represent the spatial layout of the scene. In this case, they may have an expectation of the images they will receive but be insensitive to the rate at which images arrive as they walk. We describe the way in which the eye movement strategy of animals simplifies motion processing if their goal is to move towards a desired image and discuss dorsal and ventral stream processing of moving images in that context. Although many questions about view-based approaches to scene representation remain unanswered, the solutions are likely to be highly relevant to understanding biological 3-D vision.

Humans ignore motion and stereo cues in favor of a fictional stable world

As we move through the world, our eyes acquire a sequence of images. The information from this sequence is sufficient to determine the structure of a three-dimensional scene, up to a scale factor determined by the distance that the eyes have moved [1, 2]. Previous evidence shows that the human visual system accounts for the distance the observer has walked [3,4] and the separation of the eyes [5-8] when judging the scale, shape, and distance of objects. However, in an immersive virtual-reality environment, observers failed to notice when a scene expanded or contracted, despite having consistent information about scale from both distance walked and binocular vision. This failure led to large errors in judging the size of objects. The pattern of errors cannot be explained by assuming a visual reconstruction of the scene with an incorrect estimate of interocular separation or distance walked. Instead, it is consistent with a Bayesian model of cue integration in which the efficacy of motion and disparity cues is greater at near viewing distances. Our results imply that observers are more willing to adjust their estimate of interocular separation or distance walked than to accept that the scene has changed in size.

A demonstration of 'broken' visual space

It has long been assumed that there is a distorted mapping between real and ‘perceived’ space, based on demonstrations of systematic errors in judgements of slant, curvature, direction and separation. Here, we have applied a direct test to the notion of a coherent visual space. In an immersive virtual environment, participants judged the relative distance of two squares displayed in separate intervals. On some trials, the virtual scene expanded by a factor of four between intervals although, in line with recent results, participants did not report any noticeable change in the scene. We found that there was no consistent depth ordering of objects that can explain the distance matches participants made in this environment (e.g. A > B > D yet also A < C < D) and hence no single one-to-one mapping between participants’ perceived space and any real 3D environment. Instead, factors that affect pairwise comparisons of distances dictate participants’ performance. These data contradict, more directly than previous experiments, the idea that the visual system builds and uses a coherent 3D internal representation of a scene.

Adjoint methods in data assimilation for estimating model error

Data assimilation aims to incorporate measured observations into a dynamical system model in order to produce accurate estimates of all the current (and future) state variables of the system. The optimal estimates minimize a variational principle and can be found using adjoint methods. The model equations are treated as strong constraints on the problem. In reality, the model does not represent the system behaviour exactly and errors arise due to lack of resolution and inaccuracies in physical parameters, boundary conditions and forcing terms. A technique for estimating systematic and time-correlated errors as part of the variational assimilation procedure is described here. The modified method determines a correction term that compensates for model error and leads to improved predictions of the system states. The technique is illustrated in two test cases. Applications to the 1-D nonlinear shallow water equations demonstrate the effectiveness of the new procedure.

Microsaccades distinguish between global and local visual processing

Much is known about the functional mechanisms involved in visual search. Yet, the fundamental question of whether the visual system can perform different types of visual analysis at different spatial resolutions still remains unsettled. In the visual-attention literature, the distinction between different spatial scales of visual processing corresponds to the distinction between distributed and focused attention. Some authors have argued that singleton detection can be performed in distributed attention, whereas others suggest that even such a simple visual operation involves focused attention. Here we showed that microsaccades were spatially biased during singleton discrimination but not during singleton detection. The results provide support to the hypothesis that some coarse visual analysis can be performed in a distributed attention mode.

Simulated variations in atmospheric CO2 over a Wisconsin forest using a coupled ecosystem-atmosphere model

Ecosystem fluxes of energy, water, and CO2 result in spatial and temporal variations in atmospheric properties. In principle, these variations can be used to quantify the fluxes through inverse modelling of atmospheric transport, and can improve the understanding of processes and falsifiability of models. We investigated the influence of ecosystem fluxes on atmospheric CO2 in the vicinity of the WLEF-TV tower in Wisconsin using an ecophysiological model (Simple Biosphere, SiB2) coupled to an atmospheric model (Regional Atmospheric Modelling System). Model parameters were specified from satellite imagery and soil texture data. In a companion paper, simulated fluxes in the immediate tower vicinity have been compared to eddy covariance fluxes measured at the tower, with meteorology specified from tower sensors. Results were encouraging with respect to the ability of the model to capture observed diurnal cycles of fluxes. Here, the effects of fluxes in the tower footprint were also investigated by coupling SiB2 to a high-resolution atmospheric simulation, so that the model physiology could affect the meteorological environment. These experiments were successful in reproducing observed fluxes and concentration gradients during the day and at night, but revealed problems during transitions at sunrise and sunset that appear to be related to the canopy radiation parameterization in SiB2.

An urban parameterization for a global climate model. Part I: Formulation and evaluation for two cities

Urbanization, the expansion of built-up areas, is an important yet less-studied aspect of land use/land cover change in climate science. To date, most global climate models used to evaluate effects of land use/land cover change on climate do not include an urban parameterization. Here, the authors describe the formulation and evaluation of a parameterization of urban areas that is incorporated into the Community Land Model, the land surface component of the Community Climate System Model. The model is designed to be simple enough to be compatible with structural and computational constraints of a land surface model coupled to a global climate model yet complex enough to explore physically based processes known to be important in determining urban climatology. The city representation is based upon the “urban canyon” concept, which consists of roofs, sunlit and shaded walls, and canyon floor. The canyon floor is divided into pervious (e.g., residential lawns, parks) and impervious (e.g., roads, parking lots, sidewalks) fractions. Trapping of longwave radiation by canyon surfaces and solar radiation absorption and reflection is determined by accounting for multiple reflections. Separate energy balances and surface temperatures are determined for each canyon facet. A one-dimensional heat conduction equation is solved numerically for a 10-layer column to determine conduction fluxes into and out of canyon surfaces. Model performance is evaluated against measured fluxes and temperatures from two urban sites. Results indicate the model does a reasonable job of simulating the energy balance of cities.

Climate and carbon cycle changes from 1850 to 2100 in MPI-ESM simulations for the Coupled Model Intercomparison Project phase 5

The new Max-Planck-Institute Earth System Model (MPI-ESM) is used in the Coupled Model Intercomparison Project phase 5 (CMIP5) in a series of climate change experiments for either idealized CO2-only forcing or forcings based on observations and the Representative Concentration Pathway (RCP) scenarios. The paper gives an overview of the model configurations, experiments related forcings, and initialization procedures and presents results for the simulated changes in climate and carbon cycle. It is found that the climate feedback depends on the global warming and possibly the forcing history. The global warming from climatological 1850 conditions to 2080–2100 ranges from 1.5°C under the RCP2.6 scenario to 4.4°C under the RCP8.5 scenario. Over this range, the patterns of temperature and precipitation change are nearly independent of the global warming. The model shows a tendency to reduce the ocean heat uptake efficiency toward a warmer climate, and hence acceleration in warming in the later years. The precipitation sensitivity can be as high as 2.5% K−1 if the CO2 concentration is constant, or as small as 1.6% K−1, if the CO2 concentration is increasing. The oceanic uptake of anthropogenic carbon increases over time in all scenarios, being smallest in the experiment forced by RCP2.6 and largest in that for RCP8.5. The land also serves as a net carbon sink in all scenarios, predominantly in boreal regions. The strong tropical carbon sources found in the RCP2.6 and RCP8.5 experiments are almost absent in the RCP4.5 experiment, which can be explained by reforestation in the RCP4.5 scenario.

Climatic impacts of land-use change due to crop yield increases and a universal carbon tax from a scenario model*

Future land cover will have a significant impact on climate and is strongly influenced by the extent of agricultural land use. Differing assumptions of crop yield increase and carbon pricing mitigation strategies affect projected expansion of agricultural land in future scenarios. In the representative concentration pathway 4.5 (RCP4.5) from phase 5 of the Coupled Model Intercomparison Project (CMIP5), the carbon effects of these land cover changes are included, although the biogeophysical effects are not. The afforestation in RCP4.5 has important biogeophysical impacts on climate, in addition to the land carbon changes, which are directly related to the assumption of crop yield increase and the universal carbon tax. To investigate the biogeophysical climatic impact of combinations of agricultural crop yield increases and carbon pricing mitigation, five scenarios of land-use change based on RCP4.5 are used as inputs to an earth system model [Hadley Centre Global Environment Model, version 2-Earth System (HadGEM2-ES)]. In the scenario with the greatest increase in agricultural land (as a result of no increase in crop yield and no climate mitigation) there is a significant -0.49 K worldwide cooling by 2100 compared to a control scenario with no land-use change. Regional cooling is up to -2.2 K annually in northeastern Asia. Including carbon feedbacks from the land-use change gives a small global cooling of -0.067 K. This work shows that there are significant impacts from biogeophysical land-use changes caused by assumptions of crop yield and carbon mitigation, which mean that land carbon is not the whole story. It also elucidates the potential conflict between cooling from biogeophysical climate effects of land-use change and wider environmental aims.

Including an ocean carbon cycle model into iLOVECLIM (v1.0)

The atmospheric carbon dioxide concentration plays a crucial role in the radiative balance and as such has a strong influence on the evolution of climate. Because of the numerous interactions between climate and the carbon cycle, it is necessary to include a model of the carbon cycle within a climate model to understand and simulate past and future changes of the carbon cycle. In particular, natural variations of atmospheric CO2 have happened in the past, while anthropogenic carbon emissions are likely to continue in the future. To study changes of the carbon cycle and climate on timescales of a few hundred to a few thousand years, we have included a simple carbon cycle model into the iLOVECLIM Earth System Model. In this study, we describe the ocean and terrestrial biosphere carbon cycle models and their performance relative to observational data. We focus on the main carbon cycle variables including the carbon isotope ratios δ13C and the Δ14C. We show that the model results are in good agreement with modern observations both at the surface and in the deep ocean for the main variables, in particular phosphates, dissolved inorganic carbon and the carbon isotopes.

VSImG: A high frame rate bitmap based display system for neuroscience research

This paper describes a visual stimulus generator (VSImG) capable of displaying a gray-scale, 256 x 256 x 8 bitmap image with a frame rate of 500 Hz using a boustrophedonic scanning technique. It is designed for experiments with motion-sensitive neurons of the fly`s visual system, where the flicker fusion frequency of the photoreceptors can reach up to 500 Hz. Devices with such a high frame rate are not commercially available, but are required, if sensory systems with high flicker fusion frequency are to be studied. The implemented hardware approach gives us complete real-time control of the displacement sequence and provides all the signals needed to drive an electrostatic deflection display. With the use of analog signals, very small high-resolution displacements, not limited by the image`s pixel size can be obtained. Very slow image displacements with visually imperceptible steps can also be generated. This can be of interest for other vision research experiments. Two different stimulus files can be used simultaneously, allowing the system to generate X-Y displacements on one display or independent movements on two displays as long as they share the same bitmap image. (C) 2011 Elsevier B.V. All rights reserved.

TCA Evaluation : Lab Measurements, Modelling and System Simulations

The study reported here is part of a large project for evaluation of the Thermo-Chemical Accumulator (TCA), a technology under development by the Swedish company ClimateWell AB. The studies concentrate on the use of the technology for comfort cooling. This report concentrates on measurements in the laboratory, modelling and system simulation. The TCA is a three-phase absorption heat pump that stores energy in the form of crystallised salt, in this case Lithium Chloride (LiCl) with water being the other substance. The process requires vacuum conditions as with standard absorption chillers using LiBr/water. Measurements were carried out in the laboratories at the Solar Energy Research Center SERC, at Högskolan Dalarna as well as at ClimateWell AB. The measurements at SERC were performed on a prototype version 7:1 and showed that this prototype had several problems resulting in poor and unreliable performance. The main results were that: there was significant corrosion leading to non-condensable gases that in turn caused very poor performance; unwanted crystallisation caused blockages as well as inconsistent behaviour; poor wetting of the heat exchangers resulted in relatively high temperature drops there. A measured thermal COP for cooling of 0.46 was found, which is significantly lower than the theoretical value. These findings resulted in a thorough redesign for the new prototype, called ClimateWell 10 (CW10), which was tested briefly by the authors at ClimateWell. The data collected here was not large, but enough to show that the machine worked consistently with no noticeable vacuum problems. It was also sufficient for identifying the main parameters in a simulation model developed for the TRNSYS simulation environment, but not enough to verify the model properly. This model was shown to be able to simulate the dynamic as well as static performance of the CW10, and was then used in a series of system simulations. A single system model was developed as the basis of the system simulations, consisting of a CW10 machine, 30 m2 flat plate solar collectors with backup boiler and an office with a design cooling load in Stockholm of 50 W/m2, resulting in a 7.5 kW design load for the 150 m2 floor area. Two base cases were defined based on this: one for Stockholm using a dry cooler with design cooling rate of 30 kW; one for Madrid with a cooling tower with design cooling rate of 34 kW. A number of parametric studies were performed based on these two base cases. These showed that the temperature lift is a limiting factor for cooling for higher ambient temperatures and for charging with fixed temperature source such as district heating. The simulated evacuated tube collector performs only marginally better than a good flat plate collector if considering the gross area, the margin being greater for larger solar fractions. For 30 m2 collector a solar faction of 49% and 67% were achieved for the Stockholm and Madrid base cases respectively. The average annual efficiency of the collector in Stockholm (12%) was much lower than that in Madrid (19%). The thermal COP was simulated to be approximately 0.70, but has not been possible to verify with measured data. The annual electrical COP was shown to be very dependent on the cooling load as a large proportion of electrical use is for components that are permanently on. For the cooling loads studied, the annual electrical COP ranged from 2.2 for a 2000 kWh cooling load to 18.0 for a 21000 kWh cooling load. There is however a potential to reduce the electricity consumption in the machine, which would improve these figures significantly. It was shown that a cooling tower is necessary for the Madrid climate, whereas a dry cooler is sufficient for Stockholm although a cooling tower does improve performance. The simulation study was very shallow and has shown a number of areas that are important to study in more depth. One such area is advanced control strategy, which is necessary to mitigate the weakness of the technology (low temperature lift for cooling) and to optimally use its strength (storage).