<p>This master's thesis develops an algorithm for tracking of cars robust enough to handle turning cars. It is implemented in the image processing environment Image Processing Application Programming Interface (IPAPI) for use with the WITAS project. </p><p>Firstly, algorithms, comparable with one currently used in the WITAS-project, are studied. The focus is on how rotation, that originates from the turning of the cars, affects tracking performance. The algorithms studied all perform an exhaustive search over a region, close to the last known position of the object being tracked, to find a match. After this, an iterative algorithm, based on the idea that a car can only rotate, translate and change scale, is introduced. The algorithm estimates the parameters describing this rotation, translation, and change of scale, iteratively. The iterative process needs a initial parameter estimate that is accurate enough for the algorithm to converge. The developed algorithm is based on an earlier publication on the subject, however the mathematical description, and deduction, of it is taken one step further than in this publication. </p><p>The iterative algorithm used performs well under the assumption that the data used fulfills some basic criteria. These demands comprises: placement of camera, template size as well as how the parameters may vary between two observations. The iterative algorithm is also potentially faster than exhaustive search methods, because few iterations are needed when the parameters change slowly. Better initial parameters should improve stability and speed of convergation. Other suggestions that could give better performance is discussed, e.g., methods to better extract the target from the surroundings.</p>

This master's thesis develops an algorithm for tracking of cars robust enough to handle turning cars. It is implemented in the image processing environment Image Processing Application Programming Interface (IPAPI) for use with the WITAS project. Firstly, algorithms, comparable with one currently used in the WITAS-project, are studied. The focus is on how rotation, that originates from the turning of the cars, affects tracking performance. The algorithms studied all perform an exhaustive search over a region, close to the last known position of the object being tracked, to find a match. After this, an iterative algorithm, based on the idea that a car can only rotate, translate and change scale, is introduced. The algorithm estimates the parameters describing this rotation, translation, and change of scale, iteratively. The iterative process needs a initial parameter estimate that is accurate enough for the algorithm to converge. The developed algorithm is based on an earlier publication on the subject, however the mathematical description, and deduction, of it is taken one step further than in this publication. The iterative algorithm used performs well under the assumption that the data used fulfills some basic criteria. These demands comprises: placement of camera, template size as well as how the parameters may vary between two observations. The iterative algorithm is also potentially faster than exhaustive search methods, because few iterations are needed when the parameters change slowly. Better initial parameters should improve stability and speed of convergation. Other suggestions that could give better performance is discussed, e.g., methods to better extract the target from the surroundings.

<p>The WITAS project aims to develop technologies to enable an Unmanned Airial Vehicle (UAV) to operate autonomously and intelligently, in applications such as traffic surveillance and remote photogrammetry. Many of the necessary control and reasoning tasks, e.g. state estimation, reidentification, planning and diagnosis, involve prediction as an important component. Prediction relies on models, and such models can take a variety of forms. Model design involves many choices with many alternatives for each choice, and each alternative carries advantages and disadvantages that may be far from obvious. In spite of this, and of the important role of prediction in so many areas, the problem of predictive model design is rarely studied on its own.</p><p>In this thesis, we examine a range of applications involving prediction and try to extract a set of choices and alternatives for model design. As a case study, we then develop, evaluate and compare two different model designs for a specific prediction problem encountered in the WITAS UAV project. The problem is to predict the movements of a vehicle travelling in a traffic network. The main difficulty is that uncertainty in predictions is very high, du to two factors: predictions have to be made on a relatively large time scale, and we have very little information about the specific vehicle in question. To counter uncertainty, as much use as possible must be made of knowledge about traffic in general, which puts emphasis on the knowledge representation aspect of the predictive model design.</p><p>The two mode design we develop differ mainly in how they represent uncertainty: the first uses coarse, schema-based representation of likelihood, while the second, a Markov model, uses probability. Preliminary experiments indicate that the second design has better computational properties, but also some drawbacks: model construction is data intensive and the resulting models are somewhat opaque.</p> / Report code: LiU-Tek-Lic-2002:15.

The WITAS project aims to develop technologies to enable an Unmanned Airial Vehicle (UAV) to operate autonomously and intelligently, in applications such as traffic surveillance and remote photogrammetry. Many of the necessary control and reasoning tasks, e.g. state estimation, reidentification, planning and diagnosis, involve prediction as an important component. Prediction relies on models, and such models can take a variety of forms. Model design involves many choices with many alternatives for each choice, and each alternative carries advantages and disadvantages that may be far from obvious. In spite of this, and of the important role of prediction in so many areas, the problem of predictive model design is rarely studied on its own. In this thesis, we examine a range of applications involving prediction and try to extract a set of choices and alternatives for model design. As a case study, we then develop, evaluate and compare two different model designs for a specific prediction problem encountered in the WITAS UAV project. The problem is to predict the movements of a vehicle travelling in a traffic network. The main difficulty is that uncertainty in predictions is very high, du to two factors: predictions have to be made on a relatively large time scale, and we have very little information about the specific vehicle in question. To counter uncertainty, as much use as possible must be made of knowledge about traffic in general, which puts emphasis on the knowledge representation aspect of the predictive model design. The two mode design we develop differ mainly in how they represent uncertainty: the first uses coarse, schema-based representation of likelihood, while the second, a Markov model, uses probability. Preliminary experiments indicate that the second design has better computational properties, but also some drawbacks: model construction is data intensive and the resulting models are somewhat opaque. / <p>Report code: LiU-Tek-Lic-2002:15.</p>