@techreport{GallRosenhahnSeidel2007,
TITLE = {Clustered stochastic optimization for object recognition and pose estimation},
AUTHOR = {Gall, J{\"u}rgen and Rosenhahn, Bodo and Seidel, Hans-Peter},
LANGUAGE = {eng},
URL = {http://domino.mpi-inf.mpg.de/internet/reports.nsf/NumberView/2007-4-001},
NUMBER = {MPI-I-2007-4-001},
INSTITUTION = {Max-Planck-Institut f{\"u}r Informatik},
ADDRESS = {Saarbr{\"u}cken},
YEAR = {2007},
DATE = {2007},
ABSTRACT = {We present an approach for estimating the 3D position and in case of <br>articulated objects also the joint configuration from segmented 2D <br>images. The pose estimation without initial information is a challenging <br>optimization problem in a high dimensional space and is essential for <br>texture acquisition and initialization of model-based tracking <br>algorithms. Our method is able to recognize the correct object in the <br>case of multiple objects and estimates its pose with a high accuracy. <br>The key component is a particle-based global optimization method that <br>converges to the global minimum similar to simulated annealing. After <br>detecting potential bounded subsets of the search space, the particles <br>are divided into clusters and migrate to the most attractive cluster as <br>the time increases. The performance of our approach is verified by means <br>of real scenes and a quantative error analysis for image distortions. <br>Our experiments include rigid bodies and full human bodies.},
TYPE = {Research Report / Max-Planck-Institut für Informatik},
}