An Efficient Astronomical Image Representation for Solar System Object Searches

Abstract

Astronomical images are typically thought of as two-dimensional rectangular projections onto the celestial sphere. Such images may be defined by the right ascension (α) and declination (δ) of the four corners of the image, or by the α and δ of the image center, width and height of the image, and the rotation around the image center. The computations involved in determining the location of a moving object in relation to an image exposure involves several conversions of the object's position into observer centered α and δ as both the object and observer move over time. These conversions, although not complex, are computationally expensive as they involve trigonometric transformations. Since the position of an object is relative to the observer, there is little opportunity to optimize this heavy computation against multiple images, or against multiple image collections. The viewing frustum, the threedimensional volume of space covered by an observer's view of a scene, is a long-used computer scene-rendering tool that allows world object positions to be transformed and clipped to a viewing port. This tool has been adapted to represent astronomical images as image frustums, supporting a one-time presearch optimization of individual image representations, while dramatically reducing object-image intersection detection complexity. © 2014. The Astronomical Society of the Pacific. All rights reserved.