Performance evaluation of panoramic electro-optic imagers using the TOD method

摘要

The triangle orientation discrimination (TOD) method is an emerging technique for the evaluation of electro-optical (EO) systems. In this method, the test pattern is a non-periodic equilateral triangle in one of four different orientations (apex up, down, left, or right), and the measurement procedure is a robust four-alternative forced-choice psychophysical process. This leads to a time-consuming task. Consequently, software models have been developed to replace the required human observers. These models base their decision on the orientation of the target using correlation between observed data and the set of four differently oriented targets. This study investigates for the first time how this method can be applied to highly distorted OE systems like hemispheric imagers. These types of systems have inherent large distortion, but the distortion should not be considered as an aberration but rather the result of the projection of a hemispheric field (3D) on a 2D sensor. The distortion deforms the image of the targets and image processing is usually performed to remove distortion and straighten the field of view. We present a comparison in accuracy and computational burden for the evaluation of EO system performance between cases where tested images are pre-processed and correlated to unchanged triangle targets and where untouched (distorted) images are correlated with position-wise distorted targets. This is a first evaluation of the application of the TOD with the goal of obtaining an image quality criterion for panoramic imagers.