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

The article focuses on a novel data processing pipeline designed to automatically detect flares and moving celestial objects in digitized multiple-exposure photo-plates, which are historical astronomical photographic plates capturing the same sky region multiple times on a single plate. Combining classical image processing algorithms with machine learning techniques, including a convolutional neural network for target classification, the pipeline addresses challenges posed by artifacts, varying exposure strategies, and complex star image distributions. Tested on digitized plates from the China Astronomical Plates Data, the pipeline effectively separates star chains, cross-matches detected objects with existing catalogs to identify transients, and performs simplified photometric calibration to detect stellar flares. This approach enables efficient extraction of temporal variation information from archival astronomical data, with plans to apply it to a large dataset of over 4,600 digitized plates for future discoveries.