@inproceedings{GraesslDeinzerNiemann, author = {Gr{\"a}ßl, Christoph and Deinzer, Frank and Niemann, Heinrich}, title = {Continuous Parametrization of Normal Distributions for Improving the Discrete Statistical Eigenspace Approach for Object Recognition}, series = {Pattern Recognition and Information Processing}, volume = {1}, booktitle = {Pattern Recognition and Information Processing}, editor = {Krasnoproshin, V. and Ablameyko, S. and Soldek, J.}, pages = {73 -- 77}, language = {en} } @inproceedings{GraesslDeinzerMatternetal., author = {Gr{\"a}ßl, Christoph and Deinzer, Frank and Mattern, F. and Niemann, Heinrich}, title = {Improving Statistical Object Recognition Approaches by a Parameterization of Normal Distributions}, series = {6th German-Russian IAPR Workshop on Pattern Recognition and Image Understanding}, booktitle = {6th German-Russian IAPR Workshop on Pattern Recognition and Image Understanding}, pages = {38 -- 41}, language = {en} } @inproceedings{GraesslDeinzerMatternetal., author = {Gr{\"a}ßl, Christoph and Deinzer, Frank and Mattern, F. and Niemann, Heinrich}, title = {Improving Statistical Object Recognition Approaches by a Parameterization of Normal Distributions}, series = {6th German-Russian IAPR Workshop on Pattern Recognition and Image Understanding}, volume = {14}, booktitle = {6th German-Russian IAPR Workshop on Pattern Recognition and Image Understanding}, number = {2}, issn = {1054-6618}, pages = {222 -- 230}, abstract = {As statistical approaches play an important role in object recognition, we present a novel approach which is based on object mod- els consisting of normal distributions for each training image. We show how to parameterize the mean vector and covariance matrix independently from the interpolation technique and formulate the classification and localization as a continuous optimization problem. This enables the computation of object poses which have never been seen during training. For interpolation, we present four different techniques which are compared in an experiment with real images. The results show the benefits of our method both in classification rate and pose estimation accuracy.}, language = {en} }