Image Pre-Processing Framework for Time-Domain Astronomy in the Artificial Intelligence Era

TL;DR

The paper tackles the bottleneck of image pre-processing in AI-driven time-domain astronomy by introducing a GPU-accelerated, end-to-end framework that integrates key steps such as image quality assessment, background estimation/removal, alignment, subtraction, source detection, and grayscale transformation. It provides two operating modes—Eager for interactive development and Pipeline for large-scale training/deployment—utilizing CUDA, CuPy, and NVIDIA DALI to maximize throughput with minimal data transfer. The framework demonstrates substantial speedups over CPU-based tools (e.g., alignment >10x faster than Swarp, subtraction faster than HOTPANTS, overall pipeline >12x) while preserving accuracy comparable to established methods like SExtractor, HOTPANTS, and SWarp. The work, tested on simulated data and GWAC real observations, highlights practical impact for real-time AI model training and deployment in time-domain astronomy and is packaged as a Docker image to ease adoption.

Abstract

The rapid advancement of image analysis methods in time-domain astronomy, particularly those leveraging AI algorithms, has highlighted efficient image pre-processing as a critical bottleneck affecting algorithm performance. Image pre-processing, which involves standardizing images for training or deployment of various AI algorithms, encompasses essential steps such as image quality evaluation, alignment, stacking, background extraction, gray-scale transformation, cropping, source detection, astrometry, and photometry. Historically, these algorithms were developed independently by different research groups, primarily based on CPU architecture for small-scale data processing. This paper introduces a novel framework for image pre-processing that integrates key algorithms specifically modified for GPU architecture, enabling large-scale image pre-processing for different algorithms. To prepare for the new algorithm design paradigm in the AI era, we have implemented two operational modes in the framework for different application scenarios: Eager mode and Pipeline mode. The Eager mode facilitates real-time feedback and flexible adjustments, which could be used for parameter tuning and algorithm development. The pipeline mode is primarily designed for large scale data processing, which could be used for training or deploying of artificial intelligence models. We have tested the performance of our framework using simulated and real observation images. Results demonstrate that our framework significantly enhances image pre-processing speed while maintaining accuracy levels comparable to CPU based algorithms. To promote accessibility and ease of use, a Docker version of our framework is available for download in the PaperData Repository powered by China-VO, compatible with various AI algorithms developed for time-domain astronomy research.

Figures (10)

• Figure 1: The process of training, validating, updating, and deploying the image quality assessment function. As shown in the figure, we first train an autoencoder using high-quality images, validate and update the model periodically, and then deploy the trained model to filter out high-quality images.
• Figure 2: The process of obtaining the coordinate mapping. Since each pixel coordinate undergoes the same calculation, leveraging the advantages of GPU parallel computing can significantly reduce the time required for this process.
• Figure 3: The process of performing parallel convolution on the template image and generating the difference image. By leveraging the parallel computing power of the GPU, the time required for convolution can be significantly reduced.
• Figure 4: The flowchart of the source extraction algorithm based on GPU parallel computing. By leveraging the parallel computing power of the GPU, the time spent on the four stages—detection, decomposition, cleaning, and photometry—has been significantly reduced.
• Figure 5: The diagram of the pipeline designed to carry out data pre-processing for the GWAC.