A Joint Illumination and Shape Model for Visual Tracking

Kale, Amit; Jaynes, Christopher

doi:10.1109/cvpr.2006.30

Cited by 19 publications

(41 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The set of Legendre polynomials provide a good platform for such representation, since the basis functions are smoothly changing themselves. Legendre polynomials have got a great application in certain aspects of image processing in designing illumination models for object tracking [17] and generating shape signatures for supervised segmentation [18]. Using Legendre polynomials model intensity based appearance allows the estimating functions to vary spatially, but at the same time variation is constrained by the inherent smoothness of the polynomials.…”

Section: B Choice Of Basis Functionmentioning

confidence: 99%

Region Based Segmentation in Presence of Significant Intensity Inhomogeneity using Legendre Polynomials

Rangaswamy¹

2017

IJRASET

View full text Add to dashboard Cite

we propose a region based segmentation method which is capable of segmenting objects in presence of significant intensity inhomogeneity. Present method use some form of local processing to tackle intra region inhomogeneity, which further makes such methods susceptible to local minima. Here we present a framework which generalizes the traditional Chan-Vese algorithm. In contrast with existing local techniques, we represent the illumination of the regions of interest in a lower dimensional subspace using set of pre-specified basis functions. This representation can accommodate heterogeneous objects, even in presence of noise. We compare our results with several techniques like biological/biomedical images with tubular or filamentous structures. As the, we achieve a 44% increase in performance, which demonstrates efficacy of the method. I. INTRODUCTION Active contours [1]-[4]are popular for image segmentation. Active contours have ability to elastically deform and delineate object boundaries with sub-pixel accuracy. Furthermore, the energy optimization framework, which serves as the basic platform for most active contour based techniques, can be handle easily to introduce ad-additional constraints based on shape, appearance etc. to assist segmentation. Geometric active contours are used when the application requires the propagating curves to be able to adapt to the varying topology of the underlying object by automated splitting or merging. The geometric active contours may be divided in two types. Edge based techniques and region based techniques. Edge based techniques [2], [5], [6] perform curve evolution geometrically, with the stopping criteria controlled by edge dependent features extracted from the image. However, in many applications where the edge information is unreliable, region based techniques are used. Chan and Vese [3] proposed a level set formulation to minimize the Mumford Shah functional [7] for segmentation. The Chan-Vese frame-work models the image as a set of constant illumination regions and performs a two-class segmentation by computing the optimal partition which hereby satisfies the constant illumination constraint. The authors also propose a multi-phase variant [8] of their approach to perform multi-class grouping.

show abstract

Section: B Choice Of Basis Functionmentioning

confidence: 99%

Region Based Segmentation in Presence of Significant Intensity Inhomogeneity using Legendre Polynomials

Rangaswamy¹

2017

IJRASET

View full text Add to dashboard Cite

show abstract

“…For our problem, consider the random walk model on illumination coef cients given in (2). As explained in the introduction, the rate of change of illumination over time (quanti ed by the illumination change covariance) is much larger when the car transitions from a shadowy region to a bright/sunlit region or vice versa than when it is in a shadowy or bright region.…”

Section: Detecting and Changing The System Modelmentioning

confidence: 99%

“…System Model: The state, X t , consists of a 3-dimensional motion vector u t which contains x-y translation and scale, and a 7 dimensional illumination coef cients vector (illumination is parameterized using a Legendre basis) as in [2], i.e. Xt = [u t Λ t ] The system model is a random walk model on object motion, u t and on illumination coef cients, Λ t i.e.…”

Section: State Space Model and The Problemmentioning

confidence: 99%

“…a face with light falling at different angles on different parts of the face, usually happens when light source is near the object, invalidating assumptions about point light sources at in nity), the maximum dimension of illumination change is equal to the number of image pixels. Of course, the variability is never that large, and as been demonstrated in previous work [2], usually a 3 to 7 dimensional basis suf ces for modeling illumination, but even that will increase the total state space dimension to somewhere between 9-16. It is well known that as state space dimension increases, number of particles required to track using a PF increases [3].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Generalized ELL for detecting and tracking through illumination model changes

Kale¹,

Vaswani

2008

2008 15th IEEE International Conference on Image Processing

View full text Add to dashboard Cite

In previous work, we developed the Illum-PF-MT, which is the PF-MT idea applied to the problem of tracking temporally and spatially varying illumination change. In many practical problems, the rate at which illumination changes varies over time. For e.g. when a car transitions from shadow to sunlight or vice-versa the rate of illumination change is much higher than when it is in shadow or in sunlight. One way to model illumination change in such problems is using a Gaussian random walk model with two values of the change covariance -a large covariance when a transition is detected and a much smaller one when no transition is detected. But to use such a model, one needs to rst detect the transition. The transition is a natural one and so it happens gradually (unlike a sudden manual dimming of the light in the room) and thus existing change detection statistics which are designed only for sudden changes are unable to detect the transition. In this paper, we propose to use the recently proposed generalized ELL (gELL) idea which uses the tracked part of the change to detect it and hence detects such partially trackable changes very quickly. Since gELL detects much before loss of track occurs, one is able to transition to the transition model and back without ever losing track. Also, for the rst time, we demonstrate the use of gELL in combination with the PF-MT algorithm which is more stable to model change than the original PF.

show abstract

“…For example, the growth model of [3]. Another many-to-one example is when the observation is a product of functions of two subsets of states plus noise, for example, bearings-only tracking [3] or illumination and motion tracking [12], [13].…”

Section: Introductionmentioning

confidence: 99%

Particle Filtering for Large-Dimensional State Spaces With Multimodal Observation Likelihoods

Vaswani

2008

IEEE Trans. Signal Process.

View full text Add to dashboard Cite

Abstract-We study efficient importance sampling techniques for particle filtering (PF) when either (a) the observation likelihood (OL) is frequently multimodal or heavy-tailed, or (b) the state space dimension is large or both. When the OL is multimodal, but the state transition pdf (STP) is narrow enough, the optimal importance density is usually unimodal. Under this assumption, many techniques have been proposed. But when the STP is broad, this assumption does not hold. We study how existing techniques can be generalized to situations where the optimal importance density is multimodal, but is unimodal conditioned on a part of the state vector.Sufficient conditions to test for the unimodality of this conditional posterior are derived. Our result is directly extendable to testing for unimodality of any posterior.The number of particles N to accurately track using a PF increases with state space dimension, thus making any regular PF impractical for large dimensional tracking problems. But in most such problems, most of the state change occurs in only a few dimensions, while the change in the rest of the dimensions is small. Using this property, we propose to replace importance sampling from a large part of the state space (whose conditional posterior is narrow enough) by just tracking the mode of the conditional posterior. This introduces some extra error, but it also greatly reduces the importance sampling dimension. The net effect is much smaller error for a given N, especially when the available N is small. An important class of large dimensional problems with multimodal OL is tracking spatially varying physical quantities such as temperature or pressure in a large area using a network of sensors which may be nonlinear and/or may have nonnegligible failure probabilities. Improved performance of our proposed algorithms over existing PFs is demonstrated for this problem.Index Terms-Importance sampling for multimodal posteriors, large dimensional sequential state estimation, particle filtering, posterior mode tracking, tracking spatially varying physical quantities.

show abstract

A Joint Illumination and Shape Model for Visual Tracking

Cited by 19 publications

References 12 publications

Region Based Segmentation in Presence of Significant Intensity Inhomogeneity using Legendre Polynomials

Region Based Segmentation in Presence of Significant Intensity Inhomogeneity using Legendre Polynomials

Generalized ELL for detecting and tracking through illumination model changes

Particle Filtering for Large-Dimensional State Spaces With Multimodal Observation Likelihoods

Contact Info

Product

Resources

About