===============================================================================
Projection Based M-Estimator 
Author: Raghav Subbarao
Robust Image Understanding Laboratory, Rutgers University
===============================================================================

Implements the Projection Based estimator based on:
R. Subbarao, P. Meer, "Projection Based M-Estimator",
Submitted to, IEEE Transactions on Pattern Analysis and MAchine Intelligence

Examples for using the base class for linear, heteroscedastic (ellipse and fundamental matrix) and subspace estimation are included in the program.

Using the binary:

To run the program click on the batch file "run.bat". The batch file should be edited for different options. The options are described below.

pbm [-METHOD] [DATAFILE] [NO. TRIALS]

METHOD:
-l       linear pbM
-f       fundamental pbM
-e       ellipse pbM
-m       multiple pbM

DATAFILE format (see also the examples):
nRows nColumns
input data points, each row is a measurement

the estimate and the corrected measurements are written in out.txt

Using the sources:

the project files for MS Visual C++ are included. 

To use the heteroscedastic pbM algorithm for a different problem, you should derive a class 
from cLinearHeivPbM and write the methods for Zi and Ci. Zi computes the nonlinear mapping of
the current measurement Xi. Ci computes the covariance matrix of the measurement Xi. See the 
examples provided in the code for fundamental matrices and ellipses.

Raghav Subbarao
rsubbara@caip.rutgers.edu

Robust Image Understanding Laboratory
www.caip.rutgers.edu/riul