Camera Modeling/Kalman Filter Source Code

Jump to The Nature of Mathematical Modeling Homepage

Jump back to my NMM index page

Source Code

I chose to work in ANSI C, because of its portability and ease of integration into other projects I am working on. The kalman filter code is embedded in an application which loads the simulated data from files (stored in ASCII format), performs the monte-carlo simulation, and outputs the results back in to text files.

This is code that is specific to the model used in this particular project :

test.c - "Wrapper" application
test.h - Model initialization parameters
kalman_camera.c - Model Specific Kalman Filter Routines
eval_camera.c - Model Specific Lev.-Mar. Routines
Makefile - DEC Alpha/OSF UNIX Makefile for app.

This code implements the basic algorithms used, and is model independent in nature :

matmath.h - Matrix math Declarations
matmath.c - Matrix math Routines
kalman.h - Kalman Filter Declarations
kalman.c - Kalman Filtering Routines
mrqmin.c - Levenberg-Marquardt Minimization Routines
random.c - Random Number Generators

Data Preparation and Analysis

Instead of writing standalone utilities for generating the simulation data, or using a pre-packaged utility, I chose to use MatLab. It allowed me to play around with the data and get a feel for the measurement model before the simulations, and easily analyze and plot the results.

Here are the MatLab files I used. The top-level worksheet guides you through the process, automatically doing some of the work and providing the rest in "commented out" form.

Worksheet for preparing simulated data
synth.m - Synthesizes a trajectory
traj_plot.m - Plots a trajectory
project.txt - Projects a single pt in 3D
do_proj.txt - Peforms the imaging model
calc_e.txt - Calculates the est. error
motbas.txt - Motion params used by workpage
feature0.txt - Synthetic Feature set #0
feature1.txt - Synthetic Feature set #1

wad@media.mit.edu