Citation Context

... developed in this paper are based on minimizing the sum-of-squared differences (SSD) between two regions. Although this idea has been successfully employed in many contexts including stereo matching =-=[21]-=-, optical flow computation [22], and visual motion analysis [23], previous SSD-based tracking algorithms have suffered from a variety of limitations. Many algorithms have modeled the motion of the tar...

Citation Context

...sensitivity to outliers by solving a robust optimization problem of the form 1 6 4 3 2 7 8 2 79  ORI f t I t � � = r x; , - x, 0 xŒ5 , (49) where r is one of a variety of “robust” regression metrics =-=[39]-=-. It is well-known that optimization of (49) is closely related to another approach to robust estimation—iteratively reweighted least squares (IRLS). We have chosen to implement the optimization using...

Citation Context

...red differences (SSD) between two regions. Although this idea has been successfully employed in many contexts including stereo matching [21], optical flow computation [22], and visual motion analysis =-=[23]-=-, previous SSD-based tracking algorithms have suffered from a variety of limitations. Many algorithms have modeled the motion of the target region as pure 0162-8828/98/$10.00 © 1998 IEEE1026 IEEE TRA...

Citation Context

... follows that for any time t > t0, there is a parameter vector m * (t) such that I(x, t0) = I(f(x; m * (t)), t) for all x Œ 5. (1) This is a generalization of the so-called image constancy assumption =-=[34]-=-. Thus, the motion parameter vector of the target region can be estimated at time t by minimizing the following least squares objective function 1 6 4 3 2 7 8 2 79  O � = I fx; � , t - I x, t0 xŒ5 2 ...

Citation Context

...mptions are often violated [23]. More elaborate tracking algorithms have included parametrized models for articulation [24], [25] or nonrigid deformations [26], [27] as well as linear image subspaces =-=[28]-=-, [29]. However, the resulting algorithms rely on nonlinear optimization techniques which require from several seconds to several minutes per frame to compute. Furthermore, none explicitly address the...

Citation Context

...based control [1], [2], [3], [4], human-computer interfaces [5], [6], [7], surveillance [8], [9], agricultural automation [10], [11], medical imaging [12], [13], and visual reconstruction [14], [15], =-=[16]-=-. The central challenge in visual tracking is to determine the image configuration of a target region (or features) of an object as it moves through a camera’s field of view. This is done by solving w...

Citation Context

...s are often violated [23]. More elaborate tracking algorithms have included parametrized models for articulation [24], [25] or nonrigid deformations [26], [27] as well as linear image subspaces [28], =-=[29]-=-. However, the resulting algorithms rely on nonlinear optimization techniques which require from several seconds to several minutes per frame to compute. Furthermore, none explicitly address the probl...

Citation Context

...hose with value one) are the expected target. Let Q(x) to be the pixel values in the eight-neighborhood of the image coordinate x plus the value at x itself. We use two common morphological operators =-=[40]-=- erode x max v (55) dilate x 05 = v∈Q05 x 05 = min v∈Q05 x v . (56) When applied to a weighting matrix image, erode has the effect of removing small areas of outlier pixels, while dilate increases the...

Citation Context

...sed on minimizing the sum-of-squared differences (SSD) between two regions. Although this idea has been successfully employed in many contexts including stereo matching [21], optical flow computation =-=[22]-=-, and visual motion analysis [23], previous SSD-based tracking algorithms have suffered from a variety of limitations. Many algorithms have modeled the motion of the target region as pure 0162-8828/98...

Citation Context

...ll addressed by warping. We are also looking at the problem of extending the method to utilize shape information on the target when such information is available [43]. In particular, it is well known =-=[44]-=- that under orthographic projection, the image deformations of a surface due to motion can be described with a linear motion model. This suggests that our methods can be extended to handle such models...

Citation Context

...n a target region—models that can be used to predict the movement of the points and track the target through an image sequence. In the case of illumination, we exploit the observations of [17], [18], =-=[19]-=- to model image variation due to changing illumination by low-dimensional linear subspaces. We then show that these models can be incorporated into an efficient estimation algorithm which establishes ...

Citation Context

...aphically. While computationally efficient, over a long sequence these assumptions are often violated [23]. More elaborate tracking algorithms have included parametrized models for articulation [24], =-=[25]-=- or nonrigid deformations [26], [27] as well as linear image subspaces [28], [29]. However, the resulting algorithms rely on nonlinear optimization techniques which require from several seconds to sev...

Citation Context

...icient, over a long sequence these assumptions are often violated [23]. More elaborate tracking algorithms have included parametrized models for articulation [24], [25] or nonrigid deformations [26], =-=[27]-=- as well as linear image subspaces [28], [29]. However, the resulting algorithms rely on nonlinear optimization techniques which require from several seconds to several minutes per frame to compute. F...

Citation Context

...ntly propagated through the entire sequence. Another well-established route toward efficient tracking is to detect and track only a sparse collection of features (or contours) [30], [11], [31], [32], =-=[33]-=-. As such methods use local detection of areas of high contrast change, they tend to be insensitive to global changes in the intensity and/or composition of the incident illumination. However, in many...

Citation Context

... is subsequently propagated through the entire sequence. Another well-established route toward efficient tracking is to detect and track only a sparse collection of features (or contours) [30], [11], =-=[31]-=-, [32], [33]. As such methods use local detection of areas of high contrast change, they tend to be insensitive to global changes in the intensity and/or composition of the incident illumination. Howe...

Citation Context

...t + t) - I(0, t0)� 2 . (7) If the magnitude of the components of dm are small, then it is possible to apply continuous optimization procedures to a linearized version of the problem [29], [34], [21], =-=[36]-=-, [23]. The linearization is carried out by expanding I(m + dm, t + t) in a Taylor series about m and t, I(m + dm, t + t) = I(m, t) + M(m, t) dm + t It(m, t) + h.o.t, (8) where h.o.t denotes higher-or...

Citation Context

...rthographically. While computationally efficient, over a long sequence these assumptions are often violated [23]. More elaborate tracking algorithms have included parametrized models for articulation =-=[24]-=-, [25] or nonrigid deformations [26], [27] as well as linear image subspaces [28], [29]. However, the resulting algorithms rely on nonlinear optimization techniques which require from several seconds ...

Citation Context

...l to: tpami@computer.org, and reference IEEECS Log Number 107165. ISUAL tracking has emerged as an important component of systems in several application areas including vision-based control [1], [2], =-=[3]-=-, [4], human-computer interfaces [5], [6], [7], surveillance [8], [9], agricultural automation [10], [11], medical imaging [12], [13], and visual reconstruction [14], [15], [16]. The central challenge...

Citation Context

...bsequently propagated through the entire sequence. Another well-established route toward efficient tracking is to detect and track only a sparse collection of features (or contours) [30], [11], [31], =-=[32]-=-, [33]. As such methods use local detection of areas of high contrast change, they tend to be insensitive to global changes in the intensity and/or composition of the incident illumination. However, i...

Citation Context

...points within a target region—models that can be used to predict the movement of the points and track the target through an image sequence. In the case of illumination, we exploit the observations of =-=[17]-=-, [18], [19] to model image variation due to changing illumination by low-dimensional linear subspaces. We then show that these models can be incorporated into an efficient estimation algorithm which ...

Citation Context

... worst case, complete loss of the target. Recently, it has been shown that a relatively small number of “basis” images can often be used to account for large changes in illumination [19], [18], [17], =-=[37]-=-. Briefly, the reason for this is as follows. Consider a point p on a Lambertian surface and a collimated light source characterized by a vector s Œ � 3 , such that the direction of s gives the direct...

Citation Context

...nce IEEECS Log Number 107165. ISUAL tracking has emerged as an important component of systems in several application areas including vision-based control [1], [2], [3], [4], human-computer interfaces =-=[5]-=-, [6], [7], surveillance [8], [9], agricultural automation [10], [11], medical imaging [12], [13], and visual reconstruction [14], [15], [16]. The central challenge in visual tracking is to determine ...

Citation Context

...ision-based control [1], [2], [3], [4], human-computer interfaces [5], [6], [7], surveillance [8], [9], agricultural automation [10], [11], medical imaging [12], [13], and visual reconstruction [14], =-=[15]-=-, [16]. The central challenge in visual tracking is to determine the image configuration of a target region (or features) of an object as it moves through a camera’s field of view. This is done by sol...

Citation Context

... within a target region—models that can be used to predict the movement of the points and track the target through an image sequence. In the case of illumination, we exploit the observations of [17], =-=[18]-=-, [19] to model image variation due to changing illumination by low-dimensional linear subspaces. We then show that these models can be incorporated into an efficient estimation algorithm which establ...

Citation Context

...EEECS Log Number 107165. ISUAL tracking has emerged as an important component of systems in several application areas including vision-based control [1], [2], [3], [4], human-computer interfaces [5], =-=[6]-=-, [7], surveillance [8], [9], agricultural automation [10], [11], medical imaging [12], [13], and visual reconstruction [14], [15], [16]. The central challenge in visual tracking is to determine the i...

Citation Context

...aken under three linearly independent light source directions, one can reconstruct the image of the surface under a novel lighting direction by a linear combination of the three original images [37], =-=[38]-=-. In other words, if the surface is purely Lambertian and there is no shadowing, then all images under varying illumination lie within a 3D linear subspace of � N , the space of all possible images (w...

Citation Context

... Log Number 107165. ISUAL tracking has emerged as an important component of systems in several application areas including vision-based control [1], [2], [3], [4], human-computer interfaces [5], [6], =-=[7]-=-, surveillance [8], [9], agricultural automation [10], [11], medical imaging [12], [13], and visual reconstruction [14], [15], [16]. The central challenge in visual tracking is to determine the image ...

Citation Context

...quences by an SGI Indy equipped with a 175 Mhz R4400 SC processor and VINO image acquisition system. 5.1 Implementation We have implemented the methods described above within the X Vision environment =-=[41]-=-. The implemented system incorporates all of the linear motion models described in Section 2, nonorthonormal illumination bases as described in Section 3, and outlier rejection using the algorithm des...

Citation Context

...ly efficient, over a long sequence these assumptions are often violated [23]. More elaborate tracking algorithms have included parametrized models for articulation [24], [25] or nonrigid deformations =-=[26]-=-, [27] as well as linear image subspaces [28], [29]. However, the resulting algorithms rely on nonlinear optimization techniques which require from several seconds to several minutes per frame to comp...

Citation Context

...e-mail to: tpami@computer.org, and reference IEEECS Log Number 107165. ISUAL tracking has emerged as an important component of systems in several application areas including vision-based control [1], =-=[2]-=-, [3], [4], human-computer interfaces [5], [6], [7], surveillance [8], [9], agricultural automation [10], [11], medical imaging [12], [13], and visual reconstruction [14], [15], [16]. The central chal...

Citation Context

...send e-mail to: tpami@computer.org, and reference IEEECS Log Number 107165. ISUAL tracking has emerged as an important component of systems in several application areas including vision-based control =-=[1]-=-, [2], [3], [4], human-computer interfaces [5], [6], [7], surveillance [8], [9], agricultural automation [10], [11], medical imaging [12], [13], and visual reconstruction [14], [15], [16]. The central...

Citation Context

...ding vision-based control [1], [2], [3], [4], human-computer interfaces [5], [6], [7], surveillance [8], [9], agricultural automation [10], [11], medical imaging [12], [13], and visual reconstruction =-=[14]-=-, [15], [16]. The central challenge in visual tracking is to determine the image configuration of a target region (or features) of an object as it moves through a camera’s field of view. This is done ...

Citation Context

... several application areas including vision-based control [1], [2], [3], [4], human-computer interfaces [5], [6], [7], surveillance [8], [9], agricultural automation [10], [11], medical imaging [12], =-=[13]-=-, and visual reconstruction [14], [15], [16]. The central challenge in visual tracking is to determine the image configuration of a target region (or features) of an object as it moves through a camer...

Citation Context

...andle changes of view or aspect not well addressed by warping. We are also looking at the problem of extending the method to utilize shape information on the target when such information is available =-=[43]-=-. In particular, it is well known [44] that under orthographic projection, the image deformations of a surface due to motion can be described with a linear motion model. This suggests that our methods...

Citation Context

.... ISUAL tracking has emerged as an important component of systems in several application areas including vision-based control [1], [2], [3], [4], human-computer interfaces [5], [6], [7], surveillance =-=[8]-=-, [9], agricultural automation [10], [11], medical imaging [12], [13], and visual reconstruction [14], [15], [16]. The central challenge in visual tracking is to determine the image configuration of a...

Citation Context

... important component of systems in several application areas including vision-based control [1], [2], [3], [4], human-computer interfaces [5], [6], [7], surveillance [8], [9], agricultural automation =-=[10]-=-, [11], medical imaging [12], [13], and visual reconstruction [14], [15], [16]. The central challenge in visual tracking is to determine the image configuration of a target region (or features) of an ...

Citation Context

... temporal correspondence of the target region by simultaneously determining both motion and illumination parameters. Finally, in the case of partial occlusion, we apply results from robust statistics =-=[20]-=- to develop automatic methods of rejecting occluded pixels in a computationally efficient manner. The result is a family of region-tracking algorithms which can easily track large image regions (for e...

Citation Context

...ems in several application areas including vision-based control [1], [2], [3], [4], human-computer interfaces [5], [6], [7], surveillance [8], [9], agricultural automation [10], [11], medical imaging =-=[12]-=-, [13], and visual reconstruction [14], [15], [16]. The central challenge in visual tracking is to determine the image configuration of a target region (or features) of an object as it moves through a...

Citation Context

...he difference between the two geometric models is clearly evident. 5.3 Human Face Tracking There has been a great deal of recent interest in face tracking in the computer vision literature [27], [6], =-=[42]-=-. Although faces can produce images with significant variation due to1034 IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 20, NO. 10, OCTOBER 1998 Fig. 3. Top row, excerpts from ...

Citation Context

...AL tracking has emerged as an important component of systems in several application areas including vision-based control [1], [2], [3], [4], human-computer interfaces [5], [6], [7], surveillance [8], =-=[9]-=-, agricultural automation [10], [11], medical imaging [12], [13], and visual reconstruction [14], [15], [16]. The central challenge in visual tracking is to determine the image configuration of a targ...

Citation Context

...tant component of systems in several application areas including vision-based control [1], [2], [3], [4], human-computer interfaces [5], [6], [7], surveillance [8], [9], agricultural automation [10], =-=[11]-=-, medical imaging [12], [13], and visual reconstruction [14], [15], [16]. The central challenge in visual tracking is to determine the image configuration of a target region (or features) of an object...