Multi-instrument Analysis from Point and Raster Data.

Sang-Chul Lee and Peter Bajcsy

Technical Report NCSA-ALG04-0001, February 2004

In this report, we address the problem of multi-instrument analysis from point and raster data. A camera-based sensor acquires raster data (images). Point-based sensors are attached to the object of interest and report accurate measurements at a sparse set of locations.

Our work tackles the problem of raster and point data integration, and includes sensor registration, point data interpolation, variable transformation, data overlay and value comparison. We describe a few necessary steps one has to perform in order to form two comparable data sets in terms of (1) their coordinate systems, (2) spatial resolution and (3) physical entities.

The objective of the processing steps is to resolve the above issues by (a) spatial registration of both data sets, (b) B-spline interpolation of point data, and (c) variable transformations of point data according to structural engineering formulas.

We present our preliminary results from the on-going research that is a part of the National Earthquake Engineering Simulation (NEES) project and conducted in collaboration with the Civil and Environmental Engineering Department, UIUC.

Our future work will focus on developing a new uncertainty model for data fusion so that we can predict and optimize setup parameters in the future multi-instrument experiments, for example, point sensor spacing, camera distance, point sensor locations or spatial overlap of sensor measurements.