Estudio e implementación del algoritmo PACO para la detección de exoplanetas en imágenes de alto contraste.
Loading...
Date
2025
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de Concepción
Abstract
La detección directa de exoplanetas mediante Imágenes de Alto Contraste (HCI) es una técnica fundamental para el estudio de sistemas planetarios jóvenes y masivos, pero su eficacia está limitada por el intenso patrón de ruido granular generado por la estrella anfitriona. Los algoritmos de post-procesamiento tradicionales, basados en la sustracción de una PSF de referencia, a menudo introducen sesgos fotométricos que comprometen la caracterización física de los objetos detectados. El algoritmo PACO (PAtch COvariances) ofrece una solución a este problema, utilizando un enfoque de aprendizaje estadístico para modelar localmente el ruido de fondo sin alterar la imagen, lo que permite una detección robusta y una estimación no sesgada del flujo del planeta. A pesar de sus ventajas teóricas, actualmente solo existe una implementación de código abierto, la cual carece de validación, presenta diferencias al algoritmo original y tiene algunos problemas de rendimiento, lo que limita su aplicación en la comunidad astronómica. Este trabajo presenta el desarrollo y la validación de una nueva implementación del algoritmo PACO, diseñada para ser computacionalmente eficiente y fiel a su fundamentación matemática. El objetivo es proporcionar una herramienta confiable y bien documentada que facilite la detección y caracterización de exoplanetas, y que sirva como base para el desarrollo de futuros métodos híbridos de análisis.
The direct detection of exoplanets or High-Contrast Imaging (HCI) is a fundamental technique for studying young, massive planetary systems, but its effectiveness is limited by the intense speckle noise generated by the host star. Traditional post-processing algorithms, based on the subtraction of a reference PSF, often introduce photometric biases that compromise the physical characterization of detected objects. The PACO (PAtch COvariances) algorithm offers a solution to this problem, using a statistical learning approach to locally model the background noise without altering the image, which allows for robust detection and an unbiased estimation of the planet’s flux. Despite its theoretical advantages, a validated, open-source implementation faithful to the original algorithm and with good performance is not currently available, limiting its application within the astronomical community. This work presents the development and validation of a new implementation of the PACO algorithm, designed to be computationally efficient and faithful to its mathematical foundation. The objective is to provide the community with a reliable and welldocumented tool that facilitates the precise detection and characterization of exoplanets, and that can serve as a foundation for the development of future hybrid analysis methods.
The direct detection of exoplanets or High-Contrast Imaging (HCI) is a fundamental technique for studying young, massive planetary systems, but its effectiveness is limited by the intense speckle noise generated by the host star. Traditional post-processing algorithms, based on the subtraction of a reference PSF, often introduce photometric biases that compromise the physical characterization of detected objects. The PACO (PAtch COvariances) algorithm offers a solution to this problem, using a statistical learning approach to locally model the background noise without altering the image, which allows for robust detection and an unbiased estimation of the planet’s flux. Despite its theoretical advantages, a validated, open-source implementation faithful to the original algorithm and with good performance is not currently available, limiting its application within the astronomical community. This work presents the development and validation of a new implementation of the PACO algorithm, designed to be computationally efficient and faithful to its mathematical foundation. The objective is to provide the community with a reliable and welldocumented tool that facilitates the precise detection and characterization of exoplanets, and that can serve as a foundation for the development of future hybrid analysis methods.
Description
Tesis presentada para optar al título de Ingeniero/a Civil Informático/a.
Keywords
Algoritmos, Imágenes digitales, Fotografía de planetas, Astronomía Observaciones