¤   3 / 4 / 03 - PROJECT RELEASE 3.1 (VERY BETA, MAYBE EVEN SUB-ALPHA)

Project Description

     - Evan Herbst

     - Easy Camera Calibration

     - Background:

Calibrating digital cameras is a process in progress. This project investigates the role of corner detection in camera calibration. The corner detection method of calibration involves finding a designated type of spot (corners) in some pictures taken by the camera in question and finding the geometric relationships between all the pictures given, which then provides information about the camera's characteristics: focal length, skew, etc.

     - Description:

I am using several processes already written. Mark Livingston, the NRL researcher with whom I am working, has produced a file for detecting corners in images, another for increasing the accuracy of the detected corners, and another for calculating various types of error between a given calibration model and the imaged data. The calibration procedure itself was written by Zhengyou Zhang of Microsoft Research; it takes output similar to that of Mark's second file and produces input for Mark's error statistician. The role of this project is to manipulate corner positions that have been detected and analyzed so that the result of Zhang's algorithm, easycalib, is as accurate as possible as measured by the statistix program. As of yet there are no details available on how this manipulation will be done.

¤   3 / 11 / 03 - Beginning to think calibrate will be successful. Multiple errors removed.

¤   3 / 13 / 03 - I have gotten (apparently useful) results from calibrate! Intrinsic parameters solved for, at least almost correctly. Output will be posted soon if possible.