Data processing pipeline for multiple-exposure photo-plate digital archives

Abstract

Photo-plates are an invaluable historical legacy that have been used for over a hundred years to capture images of celestial objects. By digitizing these photo-plates and processing the images with digital image processing methods, scientists can study celestial objects that exhibit temporal variations. Multiple-exposure photo-plates are a unique type of observation data that can capture images of the same sky at different observation times in a single photo-plate. Such photo-plates can be used to discover flares or moving targets with rapid variations, but they are difficult to process automatically due to their complex observation strategies. This paper proposes a pipeline based on classical data-processing algorithms and machine-learning algorithms to detect flares or moving targets in multiple-exposure photo-plate images automatically. The pipeline was used to process several digitized multiple-exposure photo-plate images from the China Astronomical Plates Data, and preliminary results indicate that the pipeline is effective. In the future, we plan to use our method to discover more celestial objects with temporal variations from photo-plate digital archives.