Jim R. Parker


Laboratory for Computer Vision
Computer Science Department
University of Calgary
2500 University Drive NW
Calgary, Alberta, Canada
email: parker@cpsc.ucalgary.ca

Research projects include:
o Combining multiple classifiers
o Fitting multiple Gaussians to data
o Symbol Recognition
Others will be added here as time permits.

Papers

Rank and Response Combination From Confusion Matrix Data
J. R. Parker

The use of prior behavior of a classifier, as measured by the confusion matrix, can yield useful information for merging multiple classifiers. In particular, response vectors can be estimated and a ranking of possible classes can be produced which can allow Borda type reconciliation methods to be applied. A combination of real data and the simulation of multiple classifiers is used to evaluate this idea, and to compare with eleven other classifier combination techniques. Millions of classifications were used in the evaluation.

Simulated Annealing for Fitting Linear Combinations of Gaussians to Data
J. R. Parker

It is difficult to find a good fit of a combination of Gaussians to arbitrary empirical data. The surface defined by the objective function contains many local minima, which trap gradient descent algorithms and cause sto chastic methods to tarry unreasonably in the vicinity. A number of techniques for accelerating convergence when using simulated annealing are presented. These are tested on a sample of known Gaussian combinations and are com pared for accuracy and resource consumption. A single `best' set of techniques is found which gives good results on the test samples and on empirical data.

Vector Templates for Symbol Recognition
J.R. Parker, Juraj Pivovarov, Dominik Royko

Template matching is rarely used for symbol recognition in practice because of the variation in scale, orientation, and line thickness and position found in actual images. An answer to this is to use an adaptable template, one that can be scaled and rotated easily. Here the use of such a template, based on vector representation of symbols, is explored. Results from handprinted digits and electronic symbols will be presented.