Pose estimation from corresponding point data

Haralick, Robert M.; Joo, Hyonam; Lee, C.; Zhuang, Xinhua; Vaidya, Vinay G.; Kim, M.B.

doi:10.1109/21.44063

Cited by 521 publications

(191 citation statements)

References 17 publications

Supporting

Mentioning

188

Contrasting

Unclassified

Order By: Relevance

“…The first module determines feature matches between the previous and the incoming image and uses their 3D-2D correspondences to compute the camera pose (RANSAC, Fischler and Bolles 1981, and iteratively re-weighted least squares optimization, Haralick et al 1989). As a result, the epipolar geometry between current and previous image can be computed and be used to limit and speed up the search for extra feature matches.…”

Section: Structure-from-motionmentioning

confidence: 99%

3D Urban Scene Modeling Integrating Recognition and Reconstruction

et al. 2007

View full text Add to dashboard Cite

Supplying realistically textured 3D city models at ground level promises to be useful for pre-visualizing upcoming traffic situations in car navigation systems. Because this pre-visualization can be rendered from the expected future viewpoints of the driver, the required maneuver will be more easily understandable. 3D city models can be reconstructed from the imagery recorded by surveying vehicles. The vastness of image material gathered by these vehicles, however, puts extreme demands on vision algorithms to ensure their practical usability. Algorithms need to be as fast as possible and should result in compact, memory efficient 3D city models for future ease of distribution and visualization. For the considered application, these are not contradictory demands. Simplified geometry assumptions can speed up vision algorithms while automatically guaranteeing compact geometry models. In this paper, we present a novel city modeling framework which builds upon this philosophy to create 3D content at high speed.Objects in the environment, such as cars and pedestrians, may however disturb the reconstruction, as they violate the simplified geometry assumptions, leading to visually unpleasant artifacts and degrading the visual realism of N. Cornelis ( ) · K. Cornelis · L. Van Gool KU Leuven, the resulting 3D city model. Unfortunately, such objects are prevalent in urban scenes. We therefore extend the reconstruction framework by integrating it with an object recognition module that automatically detects cars in the input video streams and localizes them in 3D. The two components of our system are tightly integrated and benefit from each other's continuous input. 3D reconstruction delivers geometric scene context, which greatly helps improve detection precision. The detected car locations, on the other hand, are used to instantiate virtual placeholder models which augment the visual realism of the reconstructed city model.

show abstract

Section: Structure-from-motionmentioning

confidence: 99%

3D Urban Scene Modeling Integrating Recognition and Reconstruction

et al. 2007

View full text Add to dashboard Cite

show abstract

“…Given a set of point-to-point correspondences on a pair of 3D objects that we wish to align, several research groups have shown that we can compute the relative rotation between the two sets of data using least-squares techniques (Faugeras and Hebert 1986;Arun et al 1987;Haralick et al 1989). Once we have the 3D rotation, the relative 3D translation can be computed using the means of the two data sets.…”

Section: Pose-invariant and Pose-aligned Descriptorsmentioning

confidence: 99%

“…We compute these rotations using least squares (Arun et al 1987;Haralick et al 1989). First compute the cross covariance matrix, K given by:…”

Section: Pose Computationmentioning

confidence: 99%

From 3D Point Clouds to Pose-Normalised Depth Maps

2009

View full text Add to dashboard Cite

We consider the problem of generating either pairwise-aligned or pose-normalised depth maps from noisy 3D point clouds in a relatively unrestricted poses. Our system is deployed in a 3D face alignment application and consists of the following four stages: (i) data filtering, (ii) nose tip identification and sub-vertex localisation, (iii) computation of the (relative) face orientation, (iv) generation of either a pose aligned or a pose normalised depth map. We generate an implicit radial basis function (RBF) model of the facial surface and this is employed within all four stages of the process. For example, in stage (ii), construction of novel invariant features is based on sampling this RBF over a set of concentric spheres to give a spherically-sampled RBF (SSR) shape histogram. In stage (iii), a second novel descriptor, called an isoradius contour curvature signal, is defined, which allows rotational alignment to be determined using a simple process of 1D correlation. We test our system on both the University of York (UoY) 3D face dataset and the Face Recognition Grand Challenge (FRGC) 3D data. For the more challenging UoY data, our SSR descriptors significantly outperform three variants of spin images, successfully identifying nose vertices at a rate of 99.6%. Nose localisation performance on the higher quality FRGC data, which has only small pose variations, is 99.9%. Our best N. Pears ( ) · system successfully normalises the pose of 3D faces at rates of 99.1% (UoY data) and 99.6% (FRGC data).

show abstract

“…Dementhon et al [13] proposed a POSIT method, in which an approximate solution computed using the scaled orthographic projection camera model is iteratively refined to approach a full perspective solution. Haralick et al [14] proposed a relative pose estimation algorithm which simultaneously computes the object relative pose and the depth of the observed point. The algorithm eliminates the nonlinearity due to the perspective eliminated by the depth variable.…”

Section: Introductionmentioning

confidence: 99%

“…Inspired by [14] and [15], an inverse projection ray approach based iterative algorithm is proposed which is divided into depth recovery and absolute orientation stage. The proposed algorithm is proved to be of globally convergence.…”

Section: Introductionmentioning

confidence: 99%

Monocular vision-based iterative pose estimation algorithm from corresponding feature points

Zhang

Cao

Zhang

et al. 2010

Sci. China Inf. Sci.

View full text Add to dashboard Cite

To determine the relative pose between an object and a single camera using 2D-to-3D point correspondences, a kind of iterative methods based on inverse projection ray approach is proposed. An iterative algorithm which is divided into depth recovery stage and absolute orientation stage is also proposed. In the first stage, the optimal translation vector is first computed in terms of rotation matrix via least square method, then the depths of the observed points are estimated by projecting the estimated point orthogonally to the inverse projection ray defined by the image point, and finally 3D points are reconstructed using the estimated depths from previous step. In the second stage, the optimal rotation matrix is estimated by applying Umeyama algorithm to fitting of the 3D model points and 3D estimated points. The above two stages are repeated until the result converges. The global convergence of the two-stage iterative algorithm is proven based on the global convergence theorem. Finally, a spacecraft docking application is implemented to test the effectiveness and convergence of the proposed algorithm by mathematical simulation and physical simulation. CitationZhang S J, Cao X B, Zhang F, et al. Monocular vision-based iterative pose estimation algorithm from corresponding feature points.

show abstract

Pose estimation from corresponding point data

Cited by 521 publications

References 17 publications

3D Urban Scene Modeling Integrating Recognition and Reconstruction

3D Urban Scene Modeling Integrating Recognition and Reconstruction

From 3D Point Clouds to Pose-Normalised Depth Maps

Monocular vision-based iterative pose estimation algorithm from corresponding feature points

Contact Info

Product

Resources

About