Homography Constraint Research Articles

Monocular vision based 6D object localization for service robot’s intelligent grasping

Intelligent grasping is still a hard problem for home service robots. There are two major issues in the intelligent grasping, i.e. the object recognition and the pose estimation. To grasp casually placed objects, the robot needs the object’s full 6 degrees of freedom pose data. To deal with the challenges such as illumination changes, cluttered background, occlusion, etc., we propose a monocular vision based object recognition and 6D pose estimation method. The SIFT feature point matching and brute-force search algorithm is used to do a tentative object recognition. The object recognition result is then verified with the homography constraint. After passing the verification, the 6D pose estimation is obtained through the decomposition of the homography matrix and the result is refined using the Levenberg–Marquardt algorithm. We embed our pose estimation method in a tracking by detection framework to keep computing and refining the pose during the whole approaching procedure. To test our method, a robot arm of seven degrees of freedom was utilized for a group of grasping experiments. The experimental results showed that our approach successfully recognized and grasped a variety of household objects with decent accuracy.

Applying distribution of feature points to detect multiple plane

For detecting multiple planar regions rapidly in the image sequence, this paper proposed a novel algorithm. First of all, the method uses RANSAC to detect a dominant homography, then, it calculates the poles corresponding to the two images based on homography constraint and classifies all the matching feature points into two sets. When we detect the homography corresponding to more planes based on RANSAC once more, only three pairs of feature points should be selected from the classified feature points, that is define the to-be-determined plane homography model together with a pair of poles, and this can improve the detection efficiency. Simulation and real experiment results show that the proposed algorithm is robust to the existence of mismatched features and is applicable to either stereo or motion sequence images.

Extraction of the Region Feasible for Safe Ambulation from Head-Mounted Wearable Camera Images Employing Genetic Algorithm on Homography Constraints

The objects have potentially dangerousness for pedestrians, such as illegally-parked bicy-cles, power poles, stopping vehicles, etc., could be found in a street. Those objects are threats to thesafety of elderly and disabled people with impaired vision, because it is difficult for them to find thoseobjects. In this paper, to enhance the elderly and disabled people's mobility, we propose the methodto estimate the region which is feasible for people's mobility in life space. In this method, the imagesare obtained from a head-mounted wearable camera, and the suited region for walking is estimatedusing 3-dimensional analysis and genetic algorithm. The performance and the validity are shown byapplying the proposed method to a number of scenes in life space.

Detecting and Extracting the Photo Composites Using Planar Homography and Graph Cut

With the advancement of photo and video editing tools, it has become fairly easy to tamper with photos and videos. One common way is to insert visually plausible composites into target images and videos. In this paper, we propose an automatic fake region detection method based on the planar homography constraint, and an automatic extraction method using graph cut with online feature/parameter selection. Two steps are taken in our method: 1) the targeting step, and 2) the segmentation step. First, the fake region is located roughly by enforcing the planar homography constraint. Second, the fake object is segmented via graph cut with the initialization given by the targeting step. To achieve an automatic segmentation, the optimal features and parameters for graph cut are dynamically selected via the proposed online feature/parameter selection. Performance of this method is evaluated on both semisimulated and real images. Our method works efficiently on images as long as there are regions satisfying the planar homography constraint, including image pairs captured by the approximately cocentered cameras, image pairs photographing planar or distant scenes, and a single image with duplications.

Multiple camera people detection and tracking using support integration

This paper proposes a method to locate and track people by combining evidence from multiple cameras using the homography constraint. The proposed method use foreground pixels from simple background subtraction to compute evidence of the location of people on a reference ground plane. The algorithm computes the amount of support that basically corresponds to the “foreground mass” above each pixel. Therefore, pixels that correspond to ground points have more support. The support is normalized to compensate for perspective effects and accumulated on the reference plane for all camera views. The detection of people on the reference plane becomes a search for regions of local maxima in the accumulator. Many false positives are filtered by checking the visibility consistency of the detected candidates against all camera views. The remaining candidates are tracked using Kalman filters and appearance models. Experimental results using challenging data from PETS’06 show good performance of the method in the presence of severe occlusion. Ground truth data also confirms the robustness of the method.

Robust Algebraic Segmentation of Mixed Rigid-Body and Planar Motions from Two Views

This paper studies segmentation of multiple rigid-body motions in a 3-D dynamic scene under perspective camera projection. We consider dynamic scenes that contain both 3-D rigid-body structures and 2-D planar structures. Based on the well-known epipolar and homography constraints between two views, we propose a hybrid perspective constraint (HPC) to unify the representation of rigid-body and planar motions. Given a mixture of K hybrid perspective constraints, we propose an algebraic process to partition image correspondences to the individual 3-D motions, called Robust Algebraic Segmentation (RAS). Particularly, we prove that the joint distribution of image correspondences is uniquely determined by a set of (2K)-th degree polynomials, a global signature for the union of K motions of possibly mixed type. The first and second derivatives of these polynomials provide a means to recover the association of the individual image samples to their respective motions. Finally, using robust statistics, we show that the polynomials can be robustly estimated in the presence of moderate image noise and outliers. We conduct extensive simulations and real experiments to validate the performance of the new algorithm. The results demonstrate that RAS achieves notably higher accuracy than most existing robust motion-segmentation methods, including random sample consensus (RANSAC) and its variations. The implementation of the algorithm is also two to three times faster than the existing methods. The implementation of the algorithm and the benchmark scripts are available at http://perception.csl.illinois.edu/ras/ .

Human Tracking based on Multiple View Homography

We propose a method for detection and tracking for objects under multiple cameras system. To track objects, one need to establish correspondence objects among multiple views. We apply the principal axis of objects and the homography constraint to match objects across multiple cameras. The principal axis belongs to the silhouette of objects that is extracted by the background subtraction. We use the multiple back- ground model to the background subtraction. In an image sequence, many changes happen with respect to pixel intensity. This cannot be characterized by the single background model so that is necessary to use the multiple background model. Also, we use the median background model reducing some noises. The silhouette is detected by difference with background models and current image which includes moving objects. For calculating homography, we use landmarks on the ground plane in 3D space. The homography means the relation between two correspondence between two coinciding points from different views. The intersection of principal axes and ground plane in 3D space are the same point shown in each view. The intersection occurs when a principal axis in an image crosses to the transformed ground plane from another image. We con- struct the correspondence which means the relationship between intersection in current image and transformed intersection from the other image by homography constraint. Those correspondences confirm within a short distance measuring in the top viewed plane. Thus, we track a person by these corresponding points on the ground plane.

Robust Real-Time 3D Object Tracking with Interfering Background Visual Projections

This paper presents a robust real-time object tracking system for human computer interaction in mediated environments with interfering visual projection in the background. Two major contributions are made in our research to achieve robust object tracking. A reliable outlier rejection algorithm is developed using the epipolar and homography constraints to remove false candidates caused by interfering background projections and mismatches between cameras. To reliably integrate multiple estimates of the 3D object positions, an efficient fusion algorithm based on mean shift is used. This fusion algorithm can also reduce tracking errors caused by partial occlusion of the object in some of the camera views. Experimental results obtained in real life scenarios demonstrate that the proposed system is able to achieve decent 3D object tracking performance in the presence of interfering background visual projection.

Feature point matching of curved surface and robust uncertainty

Stereo matching is an important research area in stereovision and stereo matching of curved surface is especially crucial. A novel correspondence algorithm is presented and its matching uncertainty is computed robustly for feature points of curved surface. The corners are matched by using homography constraint besides epipolar constraint to solve the occlusion problem. The uncertainty sources are analyzed. A cost function is established and acts as an optimal rule to compute the matching uncertainty. An adaptive scheme Gauss weights are put forward to make the matching results robust to noises. It makes the practical application of corner matching possible. From the experimental results of an image pair of curved surface it is shown that computing uncertainty robustly can restrain the affection caused by noises to the matching precision.

Multi camera image tracking

This paper presents a method for multi-camera image tracking in the context of image surveillance. The approach differs from most methods in that we exploit multiple camera views to resolve object occlusion. Moving objects are detected by using background subtraction. Viewpoint correspondence between the detected objects is then established by using the ground plane homography constraint. The Kalman Filter is then used to facilitate the tracking of objects in 3D. Tracking in 3D offers benefits in terms of allowing multiple views to be combined to generate a network field of view (FOV), i.e. the FOV of all the cameras combined. In addition, tracking the objects location in 3D allows us to use the Linear Kalman Filter, which is less cumbersome to implement than the Extended Kalman Filter (EKF). The current system is capable of tracking moving objects within three camera views

The impact of radial distortion on the self-calibration of rotating cameras

Recent methods of automatically calibrating the intrinsic parameters of cameras undergoing pure rotation are based on the infinite homography constraint, and have been found to be sensitive to radial distortion in the imagery. This paper develops a straightforward argument based on geometrical optics to show that increasing pin-cushion radial distortion will produce a gently worsening underestimate of the lens’ focal length, whereas increasing barrel radial distortion will produce a more sharply increasing overestimate followed by failure of the calibration. A second geometrical argument uses the approximation of a barrel-distorted image to a spherical projection to estimate the degree of distortion at which breakdown is likely to occur. The predictions are verified experimentally using data from real scenes with varying degrees of distortion and noise added. The paper also considers four methods of correcting the radial distortion within self-calibration. The first method pre-calibrates the distortion as a function of focal length, but the remainder assume no such prior knowledge. Although these prior-less methods are successful to an extent, everyday scenes are unlikely to provide image feature data of sufficient density and quality to make them fully viable.

3D EUCLIDEAN RECONSTRUCTION OF BUILDINGS FROM UNCALIBRATED IMAGE SEQUENCES

This paper presents a new approach for reconstructing realistic 3D models of buildings from uncalibrated image sequences taken by a hand-held camera. Firstly, correspondences between image pairs are established by using various computer vision tools, and then the fundamental matrix is estimated to high accuracy. Meanwhile, homography constraints are exploited to find more correspondences, to avoid degenerate cases and to obtain more accurate results. Secondly, rectified image pairs are resampled by using epipolar geometry constraints, where epipolar lines coincide with image scan-lines and disparities between the images are in the x-direction only. This allows subsequent stereoscopic analysis algorithms to easily take advantage of the epipolar constraint and reduce the search space to one dimension, namely along the horizontal row of the rectified images. Furthermore, dense stereo matching of the original image pairs is simple and low computational cost. Finally, the 3D model can be built through self-calibration, matching and Delaunay triangulation. The self-calibration method uses prior knowledge of orthogonal planes (lines) and parallel planes (lines) to act as constraints on the absolute quadric. A large number of experimental results show that this method improves the speed and accuracy of reconstructed 3D models and the 3D models obtained are more realistic.

Self-Calibration of Rotating and Zooming Cameras

In this paper we describe the theory and practice of self-calibration of cameras which are fixed in location and may freely rotate while changing their internal parameters by zooming. The basis of our approach is to make use of the so-called infinite homography constraint which relates the unknown calibration matrices to the computed inter-image homographies. In order for the calibration to be possible some constraints must be placed on the internal parameters of the camera. We present various self-calibration methods. First an iterative non-linear method is described which is very versatile in terms of the constraints that may be imposed on the camera calibration: each of the camera parameters may be assumed to be known, constant throughout the sequence but unknown, or free to vary. Secondly, we describe a fast linear method which works under the minimal assumption of zero camera skew or the more restrictive conditions of square pixels (zero skew and known aspect ratio) or known principal point. We show experimental results on both synthetic and real image sequences (where ground truth data was available) to assess the accuracy and the stability of the algorithms and to compare the result of applying different constraints on the camera parameters. We also derive an optimal Maximum Likelihood estimator for the calibration and the motion parameters. Prior knowledge about the distribution of the estimated parameters (such as the location of the principal point) may also be incorporated via Maximum a Posteriori estimation. We then identify some near-ambiguities that arise under rotational motions showing that coupled changes of certain parameters are barely observable making them indistinguishable. Finally we study the negative effect of radial distortion in the self-calibration process and point out some possible solutions to it.