Three-dimensional object recognition from single two-dimensional images
Artificial Intelligence
Determination of the Attitude of 3D Objects from a Single Perspective View
IEEE Transactions on Pattern Analysis and Machine Intelligence
Determination of Camera Location from 2-D to 3-D Line and Point Correspondences
IEEE Transactions on Pattern Analysis and Machine Intelligence
Image and Vision Computing - Special issue on the first ECCV 1990
Fitting Parameterized Three-Dimensional Models to Images
IEEE Transactions on Pattern Analysis and Machine Intelligence
Three-dimensional computer vision: a geometric viewpoint
Three-dimensional computer vision: a geometric viewpoint
Model-based object pose in 25 lines of code
International Journal of Computer Vision - Special issue: image understanding research at the University of Maryland
Object pose from 2-D to 3-D point and line correspondences
International Journal of Computer Vision
Linear N-Point Camera Pose Determination
IEEE Transactions on Pattern Analysis and Machine Intelligence
Fast and Globally Convergent Pose Estimation from Video Images
IEEE Transactions on Pattern Analysis and Machine Intelligence
Perceptual Organization and Visual Recognition
Perceptual Organization and Visual Recognition
Hi-index | 0.00 |
This paper presents an adaptation of Lowe's numerical model-based camera localisation algorithm to the domain of indoor mobile robotics. While the original method is straightforward and even elegant, it nonetheless exhibits certain weaknesses. First, due to an affine approximation, the method is not consistent with perspective projection especially when the dimensions of objects seen are large in comparison with their distances to the camera. Next, the non-linearity of equations makes convergence properties sensitive both to the initial solution estimate and to noise. By taking the specificity and exigency of the mobile robotics domain into account, a new formulation of this method is proposed in order to improve efficiency, accuracy and robustness in the presence of noisy data and variable initial conditions. According to this formulation, line correspondences are used rather than points, the number of degrees of freedom is reduced, the affine approximation is removed and rotation is uncoupled from translation. Test results with both synthetic and real images illustrate the improvements expected from theoretical modifications.