Learning and Exploiting Camera Geometry for Computer Vision

This work explores the use of statistical methods in describing and estimating camera poses, as well as the information feedback loop between camera pose and object detection. Surging development in robotics and computer vision has pushed the need for algorithms that infer, understand, and utilize information about the position and orientation of the sensor platforms when observing and/or interacting with their environment.

The first contribution of this thesis is the development of a set of statistical tools for representing and estimating the uncertainty in object poses. A distribution for representing the joint uncertainty over multiple object positions and orientations is described, called the mirrored normal-Bingham distribution. This distribution generalizes both the normal distribution in Euclidean space, and the Bingham distribution on the unit hypersphere. It is shown to inherit many of the convenient properties of these special cases: it is the maximum-entropy distribution with fixed second moment, and there is a generalized Laplace approximation whose result is the mirrored normal-Bingham distribution. This distribution and approximation method are demonstrated by deriving the analytical approximation to the wrapped-normal distribution. Further, it is shown how these tools can be used to represent the uncertainty in the result of a bundle adjustment problem.

Another application of these methods is illustrated as part of a novel camera pose estimation algorithm based on object detections. The autocalibration task is formulated as a bundle adjustment problem using prior distributions over the 3D points to enforce the objects' structure and their relationship with the scene geometry. This framework is very flexible and enables the use of off-the-shelf computational tools to solve specialized autocalibration problems. Its performance is evaluated using a pedestrian detector to provide head and foot location observations, and it proves much faster and potentially more accurate than existing methods.

Finally, the information feedback loop between object detection and camera pose estimation is closed by utilizing camera pose information to improve object detection in scenarios with significant perspective warping. Methods are presented that allow the inverse perspective mapping traditionally applied to images to be applied instead to features computed from those images. For the special case of HOG-like features, which are used by many modern object detection systems, these methods are shown to provide substantial performance benefits over unadapted detectors while achieving real-time frame rates, orders of magnitude faster than comparable image warping methods.

The statistical tools and algorithms presented here are especially promising for mobile cameras, providing the ability to autocalibrate and adapt to the camera pose in real time. In addition, these methods have wide-ranging potential applications in diverse areas of computer vision, robotics, and imaging.

Hyperspectral Image Classification with Nonlinear Methods

This thesis introduces two related lines of study on classification of hyperspectral images with nonlinear methods. First, it describes a quantitative and systematic evaluation, by the author, of each major component in a pipeline for classifying hyperspectral images (HSI) developed earlier in a joint collaboration [23]. The pipeline, with novel use of nonlinear classification methods, has reached beyond the state of the art in classification accuracy on commonly used benchmarking HSI data [6], [13]. More importantly, it provides a clutter map, with respect to a predetermined set of classes, toward the real application situations where the image pixels not necessarily fall into a predetermined set of classes to be identified, detected or classified with.

The particular components evaluated are a) band selection with band-wise entropy spread, b) feature transformation with spatial filters and spectral expansion with derivatives c) graph spectral transformation via locally linear embedding for dimension reduction, and d) statistical ensemble for clutter detection. The quantitative evaluation of the pipeline verifies that these components are indispensable to high-accuracy classification.

Secondly, the work extends the HSI classification pipeline with a single HSI data cube to multiple HSI data cubes. Each cube, with feature variation, is to be classified of multiple classes. The main challenge is deriving the cube-wise classification from pixel-wise classification. The thesis presents the initial attempt to circumvent it, and discuss the potential for further improvement.