Practical improvements to simultaneous computation of multi-view geometry and radial lens distortion

This paper discusses practical issues related to the use of the division model for lens distortion in multi-view geometry computation. A data normalisation strategy is presented, which has been absent from previous discussions on the topic. The convergence properties of the Rectangular Quadric Eigenvalue Problem solution for computing division model distortion are examined. It is shown that the existing method can require more than 1000 iterations when dealing with severe distortion. A method is presented for accelerating convergence to less than 10 iterations for any amount of distortion. The new method is shown to produce equivalent or better results than the existing method with up to two orders of magnitude reduction in iterations. Through detailed simulation it is found that the number of data points used to compute geometry and lens distortion has a strong influence on convergence speed and solution accuracy. It is recommended that more than the minimal number of data points be used when computing geometry using a robust estimator such as RANSAC. Adding two to four extra samples improves the convergence rate and accuracy sufficiently to compensate for the increased number of samples required by the RANSAC process.

Evaluation of two-view geometry methods with automatic ground-truth generation

A large number of methods have been published that aim to evaluate various components of multi-view geometry systems. Most of these have focused on the feature extraction, description and matching stages (the visual front end), since geometry computation can be evaluated through simulation. Many data sets are constrained to small scale scenes or planar scenes that are not challenging to new algorithms, or require special equipment. This paper presents a method for automatically generating geometry ground truth and challenging test cases from high spatio-temporal resolution video. The objective of the system is to enable data collection at any physical scale, in any location and in various parts of the electromagnetic spectrum. The data generation process consists of collecting high resolution video, computing accurate sparse 3D reconstruction, video frame culling and down sampling, and test case selection. The evaluation process consists of applying a test 2-view geometry method to every test case and comparing the results to the ground truth. This system facilitates the evaluation of the whole geometry computation process or any part thereof against data compatible with a realistic application. A collection of example data sets and evaluations is included to demonstrate the range of applications of the proposed system.

Applications of image-based, multi-view stereo reconstruction methods = (Aplicaciones de los métodos de reconstrucción estéreo multivista)

These slides present several 3-D reconstruction methods to obtain the geometric structure of a scene that is viewed by multiple cameras. We focus on the combination of the geometric modeling in the image formation process with the use of standard optimization tools to estimate the characteristic parameters that describe the geometry of the 3-D scene. In particular, linear, non-linear and robust methods to estimate the monocular and epipolar geometry are introduced as cornerstones to generate 3-D reconstructions with multiple cameras. Some examples of systems that use this constructive strategy are Bundler, PhotoSynth, VideoSurfing, etc., which are able to obtain 3-D reconstructions with several hundreds or thousands of cameras. En esta presentación se tratan varios métodos de reconstrucción 3-D para la obtención de la estructura geométrica de una escena que es visualizada por varias cámaras. Se enfatiza la combinación de modelado geométrico del proceso de formación de la imagen con el uso de herramientas estándar de optimización para estimar los parámetros característicos que describen la geometría de la escena 3-D. En concreto, se presentan métodos de estimación lineales, no lineales y robustos de las geometrías monocular y epipolar como punto de partida para generar reconstrucciones con tres o más cámaras. Algunos ejemplos de sistemas que utilizan este enfoque constructivo son Bundler, PhotoSynth, VideoSurfing, etc., los cuales, en la práctica pueden llegar a reconstruir una escena con varios cientos o miles de cámaras.

Large scale multi-view stereopsis evaluation

The seminal multiple view stereo benchmark evaluations from Middlebury and by Strecha et al. have played a major role in propelling the development of multi-view stereopsis methodology. Although seminal, these benchmark datasets are limited in scope with few reference scenes. Here, we try to take these works a step further by proposing a new multi-view stereo dataset, which is an order of magnitude larger in number of scenes and with a significant increase in diversity. Specifically, we propose a dataset containing 80 scenes of large variability. Each scene consists of 49 or 64 accurate camera positions and reference structured light scans, all acquired by a 6-axis industrial robot. To apply this dataset we propose an extension of the evaluation protocol from the Middlebury evaluation, reflecting the more complex geometry of some of our scenes. The proposed dataset is used to evaluate the state of the art multiview stereo algorithms of Tola et al., Campbell et al. and Furukawa et al. Hereby we demonstrate the usability of the dataset as well as gain insight into the workings and challenges of multi-view stereopsis. Through these experiments we empirically validate some of the central hypotheses of multi-view stereopsis, as well as determining and reaffirming some of the central challenges.

Shape from photographs:a multi-view stereo pipeline

Acquiring 3D shape from images is a classic problem in Computer Vision occupying researchers for at least 20 years. Only recently however have these ideas matured enough to provide highly accurate results. We present a complete algorithm to reconstruct 3D objects from images using the stereo correspondence cue. The technique can be described as a pipeline of four basic building blocks: camera calibration, image segmentation, photo-consistency estimation from images, and surface extraction from photo-consistency. In this Chapter we will put more emphasis on the latter two: namely how to extract geometric information from a set of photographs without explicit camera visibility, and how to combine different geometry estimates in an optimal way. © 2010 Springer-Verlag Berlin Heidelberg.

3D ellipsoid fitting for multi-view gait recognition

Gait recognition approaches continue to struggle with challenges including view-invariance, low-resolution data, robustness to unconstrained environments, and fluctuating gait patterns due to subjects carrying goods or wearing different clothes. Although computationally expensive, model based techniques offer promise over appearance based techniques for these challenges as they gather gait features and interpret gait dynamics in skeleton form. In this paper, we propose a fast 3D ellipsoidal-based gait recognition algorithm using a 3D voxel model derived from multi-view silhouette images. This approach directly solves the limitations of view dependency and self-occlusion in existing ellipse fitting model-based approaches. Voxel models are segmented into four components (left and right legs, above and below the knee), and ellipsoids are fitted to each region using eigenvalue decomposition. Features derived from the ellipsoid parameters are modeled using a Fourier representation to retain the temporal dynamic pattern for classification. We demonstrate the proposed approach using the CMU MoBo database and show that an improvement of 15-20% can be achieved over a 2D ellipse fitting baseline.

An adaptive spherical view representation for navigation in changing environments

Real-world environments such as houses and offices change over time, meaning that a mobile robot’s map will become out of date. In previous work we introduced a method to update the reference views in a topological map so that a mobile robot could continue to localize itself in a changing environment using omni-directional vision. In this work we extend this longterm updating mechanism to incorporate a spherical metric representation of the observed visual features for each node in the topological map. Using multi-view geometry we are then able to estimate the heading of the robot, in order to enable navigation between the nodes of the map, and to simultaneously adapt the spherical view representation in response to environmental changes. The results demonstrate the persistent performance of the proposed system in a long-term experiment.

Long-term experiments with an adaptive spherical view representation for navigation in changing environments

Real-world environments such as houses and offices change over time, meaning that a mobile robot’s map will become out of date. In this work, we introduce a method to update the reference views in a hybrid metrictopological map so that a mobile robot can continue to localize itself in a changing environment. The updating mechanism, based on the multi-store model of human memory, incorporates a spherical metric representation of the observed visual features for each node in the map, which enables the robot to estimate its heading and navigate using multi-view geometry, as well as representing the local 3D geometry of the environment. A series of experiments demonstrate the persistence performance of the proposed system in real changing environments, including analysis of the long-term stability.

Boosting-based transfer learning for multi-view head-pose classification from surveillance videos

This work proposes a boosting-based transfer learning approach for head-pose classification from multiple, low-resolution views. Head-pose classification performance is adversely affected when the source (training) and target (test) data arise from different distributions (due to change in face appearance, lighting, etc). Under such conditions, we employ Xferboost, a Logitboost-based transfer learning framework that integrates knowledge from a few labeled target samples with the source model to effectively minimize misclassifications on the target data. Experiments confirm that the Xferboost framework can improve classification performance by up to 6%, when knowledge is transferred between the CLEAR and FBK four-view headpose datasets.

An adaptation framework for head-pose classification in dynamic multi-view scenarios

Multi-view head-pose estimation in low-resolution, dynamic scenes is difficult due to blurred facial appearance and perspective changes as targets move around freely in the environment. Under these conditions, acquiring sufficient training examples to learn the dynamic relationship between position, face appearance and head-pose can be very expensive. Instead, a transfer learning approach is proposed in this work. Upon learning a weighted-distance function from many examples where the target position is fixed, we adapt these weights to the scenario where target positions are varying. The adaptation framework incorporates reliability of the different face regions for pose estimation under positional variation, by transforming the target appearance to a canonical appearance corresponding to a reference scene location. Experimental results confirm effectiveness of the proposed approach, which outperforms state-of-the-art by 9.5% under relevant conditions. To aid further research on this topic, we also make DPOSE- a dynamic, multi-view head-pose dataset with ground-truth publicly available with this paper.

Exploring Transfer Learning Approaches for Head Pose Classification from Multi-view Surveillance Images

Head pose classification from surveillance images acquired with distant, large field-of-view cameras is difficult as faces are captured at low-resolution and have a blurred appearance. Domain adaptation approaches are useful for transferring knowledge from the training (source) to the test (target) data when they have different attributes, minimizing target data labeling efforts in the process. This paper examines the use of transfer learning for efficient multi-view head pose classification with minimal target training data under three challenging situations: (i) where the range of head poses in the source and target images is different, (ii) where source images capture a stationary person while target images capture a moving person whose facial appearance varies under motion due to changing perspective, scale and (iii) a combination of (i) and (ii). On the whole, the presented methods represent novel transfer learning solutions employed in the context of multi-view head pose classification. We demonstrate that the proposed solutions considerably outperform the state-of-the-art through extensive experimental validation. Finally, the DPOSE dataset compiled for benchmarking head pose classification performance with moving persons, and to aid behavioral understanding applications is presented in this work.

ON THE UTILITY OF CANONICAL CORRELATION ANALYSIS FOR DOMAIN ADAPTATION IN MULTI-VIEW HEADPOSE ESTIMATION

The utility of canonical correlation analysis (CCA) for domain adaptation (DA) in the context of multi-view head pose estimation is examined in this work. We consider the three problems studied in 1], where different DA approaches are explored to transfer head pose-related knowledge from an extensively labeled source dataset to a sparsely labeled target set, whose attributes are vastly different from the source. CCA is found to benefit DA for all the three problems, and the use of a covariance profile-based diagonality score (DS) also improves classification performance with respect to a nearest neighbor (NN) classifier.