DIRSIG: Digital Imaging and Remote Sensing Image Generation Model: InfraredAirborne Validation and Input Parameter Analysis

摘要

Synthetic imagery is created by mathematically modeling the 'real world' and theimaging chain, encompassing everything from the target to the sensor characteristics. The Digital Imaging and Remote Sensing Image Generation Model (DIRSIG) attempts to model the real world. It creates synthetic images through the integration of scene geometry, ray-tracer, thermal, radiometry, and sensor submodels. The focus of this project lies in evaluating the ability of DIRSIG to recreate the imaging chain and produce high resolution synthetic imagery. DIRSIG synthetic imagery of the Kodak Hawkeye plant and the surrounding area was compared to aerial infrared imagery of the same region using root mean square error and rank order correlation. This comparison helped to validate the output from DIRSIG and detect inadequacies in the image chain model. In addition to validating DTRSIG, a procedure for optimizing the input parameters, incorporating a sensitivity analysis, was developed. This reduces the time involved in creating a realistic and accurate synthetic image.