Object Matching in Distributed Video Surveillance Systems by LDA-Based Appearance Descriptors

Establishing correspondences among object instances is still challenging in multi-camera surveillance systems, especially when the cameras’ fields of view are non-overlapping. Spatiotemporal constraints can help in solving the correspondence problem but still leave a wide margin of uncertainty. One way to reduce this uncertainty is to use appearance information about the moving objects in the site. In this paper we present the preliminary results of a new method that can capture salient appearance characteristics at each camera node in the network. A Latent Dirichlet Allocation (LDA) model is created and maintained at each node in the camera network. Each object is encoded in terms of the LDA bag-of-words model for appearance. The encoded appearance is then used to establish probable matching across cameras. Preliminary experiments are conducted on a dataset of 20 individuals and comparison against Madden’s I-MCHR is reported.

An Invariant Representation for Matching Trajectories across Uncalibrated Video Streams

We introduce a view-point invariant representation of moving object trajectories that can be used in video database applications. It is assumed that trajectories lie on a surface that can be locally approximated with a plane. Raw trajectory data is first locally approximated with a cubic spline via least squares fitting. For each sampled point of the obtained curve, a projective invariant feature is computed using a small number of points in its neighborhood. The resulting sequence of invariant features computed along the entire trajectory forms the view invariant descriptor of the trajectory itself. Time parametrization has been exploited to compute cross ratios without ambiguity due to point ordering. Similarity between descriptors of different trajectories is measured with a distance that takes into account the statistical properties of the cross ratio, and its symmetry with respect to the point at infinity. In experiments, an overall correct classification rate of about 95% has been obtained on a dataset of 58 trajectories of players in soccer video, and an overall correct classification rate of about 80% has been obtained on matching partial segments of trajectories collected from two overlapping views of outdoor scenes with moving people and cars.

Isoperimetric Graph Partitioning for Data Clustering and Image Segmentation

Spectral methods of graph partitioning have been shown to provide a powerful approach to the image segmentation problem. In this paper, we adopt a different approach, based on estimating the isoperimetric constant of an image graph. Our algorithm produces the high quality segmentations and data clustering of spectral methods, but with improved speed and stability.

Isoperimetric Partitioning: A New Algorithm for Graph Partitioning

Temporal structure is skilled, fluent action exists at several nested levels. At the largest scale considered here, short sequences of actions that are planned collectively in prefronatal cortex appear to be queued for performance by a cyclic competitive process that operates in concert with a parallel analog representation that implicitly specifies the relative priority of elements of the sequence. At an intermediate scale, single acts, like reaching to grasp, depend on coordinated scaling of the rates at which many muscles shorten or lengthen in parallel. To ensure success of acts such as catching an approaching ball, such parallel rate scaling, which appears to be one function of the basal ganglia, must be coupled to perceptual variables such as time-to-contact. At a finer scale, within each act, desired rate scaling can be realized only if precisely timed muscle activations first accelerate and then decelerate the limbs, to ensure that muscle length changes do not under- or over- shoot the amounts needed for precise acts. Each context of action may require a different timed muscle activation pattern than similar contexts. Because context differences that require different treatment cannot be known in advance, a formidable adaptive engine-the cerebellum-is needed to amplify differences within, and continuosly search, a vast parallel signal flow, in order to discover contextual "leading indicators" of when to generate distinctive patterns of analog signals. From some parts of the cerebellum, such signals control muscles. But a recent model shows how the lateral cerebellum may serve the competitive queuing system (frontal cortex) as a repository of quickly accessed long-term sequence memories. Thus different parts of the cerebellum may use the same adaptive engine design to serve the lowest and highest of the three levels of temporal structure treated. If so, no one-to-one mapping exists between leveels of temporal structure and major parts of the brain. Finally, recent data cast doubt on network-delay models of cerebellar adaptive timing.

The Graph Analysis Toolbox: Image Processing on Arbitrary Graphs

Office of Naval Research (N00014-01-1-0624)

Integration of micro- and macroscopic models for pedestrian evacuation simulation

Simulation of pedestrian evacuations of smart buildings in emergency is a powerful tool for building analysis, dynamic evacuation planning and real-time response to the evolving state of evacuations. Macroscopic pedestrian models are low-complexity models that are and well suited to algorithmic analysis and planning, but are quite abstract. Microscopic simulation models allow for a high level of simulation detail but can be computationally intensive. By combining micro- and macro- models we can use each to overcome the shortcomings of the other and enable new capability and applications for pedestrian evacuation simulation that would not be possible with either alone. We develop the EvacSim multi-agent pedestrian simulator and procedurally generate macroscopic flow graph models of building space, integrating micro- and macroscopic approaches to simulation of the same emergency space. By “coupling” flow graph parameters to microscopic simulation results, the graph model captures some of the higher detail and fidelity of the complex microscopic simulation model. The coupled flow graph is used for analysis and prediction of the movement of pedestrians in the microscopic simulation, and investigate the performance of dynamic evacuation planning in simulated emergencies using a variety of strategies for allocation of macroscopic evacuation routes to microscopic pedestrian agents. The predictive capability of the coupled flow graph is exploited for the decomposition of microscopic simulation space into multiple future states in a scalable manner. By simulating multiple future states of the emergency in short time frames, this enables sensing strategy based on simulation scenario pattern matching which we show to achieve fast scenario matching, enabling rich, real-time feedback in emergencies in buildings with meagre sensing capabilities.

Matching and sorting in online dating

Using data on user attributes and interactions from an online dating site, we estimate mate preferences, and use the Gale-Shapley algorithm to predict stable matches. The predicted matches are similar to the actual matches achieved by the dating site, and the actual matches are approximately efficient. Out-of-sample predictions of offline matches, i.e., marriages, exhibit assortative mating patterns similar to those observed in actual marriages. Thus, mate preferences, without resort to search frictions, can generate sorting in marriages. However, we underpredict some of the correlation patterns; search frictions may play a role in explaining the discrepancy.

Strategy-proofness versus Efficiency in Matching with Indifferences: Redesigning the NYC High School Match

The design of the New York City (NYC) high school match involved trade-offs among efficiency, stability, and strategy-proofness that raise new theoretical questions. We analyze a model with indifferences-ties-in school preferences. Simulations with field data and the theory favor breaking indifferences the same way at every school-single tiebreaking-in a student-proposing deferred acceptance mechanism. Any inefficiency associated with a realized tiebreaking cannot be removed without harming student incentives. Finally, we empirically document the extent of potential efficiency loss associated with strategy-proofness and stability, and direct attention to some open questions. (JEL C78, D82, I21).

Optical-CT 3D Dosimetry Using Fresnel Lenses with Minimal Refractive-Index Matching Fluid.

Telecentric optical computed tomography (optical-CT) is a state-of-the-art method for visualizing and quantifying 3-dimensional dose distributions in radiochromic dosimeters. In this work a prototype telecentric system (DFOS-Duke Fresnel Optical-CT Scanner) is evaluated which incorporates two substantial design changes: the use of Fresnel lenses (reducing lens costs from $10-30K t0 $1-3K) and the use of a 'solid tank' (which reduces noise, and the volume of refractively matched fluid from 1 ltr to 10 cc). The efficacy of DFOS was evaluated by direct comparison against commissioned scanners in our lab. Measured dose distributions from all systems were compared against the predicted dose distributions from a commissioned treatment planning system (TPS). Three treatment plans were investigated including a simple four-field box treatment, a multiple small field delivery, and a complex IMRT treatment. Dosimeters were imaged within 2 h post irradiation, using consistent scanning techniques (360 projections acquired at 1 degree intervals, reconstruction at 2mm). DFOS efficacy was evaluated through inspection of dose line-profiles, and 2D and 3D dose and gamma maps. DFOS/TPS gamma pass rates with 3%/3mm dose difference/distance-to-agreement criteria ranged from 89.3% to 92.2%, compared to from 95.6% to 99.0% obtained with the commissioned system. The 3D gamma pass rate between the commissioned system and DFOS was 98.2%. The typical noise rates in DFOS reconstructions were up to 3%, compared to under 2% for the commissioned system. In conclusion, while the introduction of a solid tank proved advantageous with regards to cost and convenience, further work is required to improve the image quality and dose reconstruction accuracy of the new DFOS optical-CT system.

Parallel dynamic graph-partitioning for unstructured meshes

A parallel method for the dynamic partitioning of unstructured meshes is described. The method introduces a new iterative optimization technique known as relative gain optimization which both balances the workload and attempts to minimize the interprocessor communications overhead. Experiments on a series of adaptively refined meshes indicate that the algorithm provides partitions of an equivalent or higher quality to static partitioners (which do not reuse the existing partition) and much more rapidly. Perhaps more importantly, the algorithm results in only a small fraction of the amount of data migration compared to the static partitioners.

A multilevel algorithm for force-directed graph drawing

We describe a heuristic method for drawing graphs which uses a multilevel technique combined with a force-directed placement algorithm. The multilevel process groups vertices to form clusters, uses the clusters to define a new graph and is repeated until the graph size falls below some threshold. The coarsest graph is then given an initial layout and the layout is successively refined on all the graphs starting with the coarsest and ending with the original. In this way the multilevel algorithm both accelerates and gives a more global quality to the force- directed placement. The algorithm can compute both 2 & 3 dimensional layouts and we demonstrate it on a number of examples ranging from 500 to 225,000 vertices. It is also very fast and can compute a 2D layout of a sparse graph in around 30 seconds for a 10,000 vertex graph to around 10 minutes for the largest graph. This is an order of magnitude faster than recent implementations of force-directed placement algorithms.

A framework for historical case-based reasoning

This paper presents a framework for Historical Case-Based Reasoning (HCBR) which allows the expression of both relative and absolute temporal knowledge, representing case histories in the real world. The formalism is founded on a general temporal theory that accommodates both points and intervals as primitive time elements. A case history is formally defined as a collection of (time-independent) elemental cases, together with its corresponding temporal reference. Case history matching is two-fold, i.e., there are two similarity values need to be computed: the non-temporal similarity degree and the temporal similarity degree. On the one hand, based on elemental case matching, the non-temporal similarity degree between case histories is defined by means of computing the unions and intersections of the involved elemental cases. On the other hand, by means of the graphical presentation of temporal references, the temporal similarity degree in case history matching is transformed into conventional graph similarity measurement.

A multilevel algorithm for force-directed graph-drawing

We describe a heuristic method for drawing graphs which uses a multilevel framework combined with a force-directed placement algorithm. The multilevel technique matches and coalesces pairs of adjacent vertices to define a new graph and is repeated recursively to create a hierarchy of increasingly coarse graphs, G0, G1, …, GL. The coarsest graph, GL, is then given an initial layout and the layout is refined and extended to all the graphs starting with the coarsest and ending with the original. At each successive change of level, l, the initial layout for Gl is taken from its coarser and smaller child graph, Gl+1, and refined using force-directed placement. In this way the multilevel framework both accelerates and appears to give a more global quality to the drawing. The algorithm can compute both 2 & 3 dimensional layouts and we demonstrate it on examples ranging in size from 10 to 225,000 vertices. It is also very fast and can compute a 2D layout of a sparse graph in around 12 seconds for a 10,000 vertex graph to around 5-7 minutes for the largest graphs. This is an order of magnitude faster than recent implementations of force-directed placement algorithms.