Pairwise Markov fields for segmentation in astronomical hyperspectral images

Abstract

We consider the problem of segmentation in noisy, blurred astronomical hyperspectral images (HSI). Recent methods based on an hypothesis-testing framework handle the problem, but do not allow to use a prior on the result. This paper introduces a pairwise Markov field model, allowing the unsupervized Bayesian segmentation of faint sources in astronomical HSI. Results on synthetic images show that the segmentation methods outperform their state-of-the-art counterparts, and allow the detection at very low SNR. Besides, results on real images provide encouraging detections with respect to the application.

Pairwise and Hidden Markov Random Fields in Image Segmentation

“…The posterior distributions in a site for the two approaches are given in Eqs. (10) and (15). Identifying the three exponential terms in (10) as proportional to three densities f , g and h let us rewrite (10) as…”

“…(10) and (15). Identifying the three exponential terms in (10) as proportional to three densities f , g and h let us rewrite (10) as…”

“…We and colleagues have often employed the PMRF model in the past, see e.g. [8]- [10]. However the PMRF model has been rarely used in the community, contrary to the HMRF model.…”

Recent Advances in Counterfeit Art, Document, Photo, Hologram, and Currency Detection Using Hyperspectral Imaging

“…In remote sensing, it is used for geological exploration and soil characterization from a distance [ 72 , 73 , 74 , 75 , 76 , 77 , 78 , 79 ]. HSI paired with its remote sensing capability is often used in the field of astronomy for astronomic observation and space surveillance purposes [ 80 , 81 , 82 , 83 , 84 , 85 , 86 , 87 , 88 , 89 , 90 ]. In addition to these, environmental applications such as drought stress measurement, pollution detection, water resource analysis, space science, and vegetation monitoring also prove HSI’s reliability [ 91 , 92 , 93 , 94 , 95 , 96 , 97 , 98 , 99 , 100 , 101 , 102 , 103 , 104 ].…”

A General Parametrization Framework for Pairwise Markov Models: An Application to Unsupervised Image Segmentation

“…They rely on a transition distribution p θ θ θ (h k , x k |h k−1 , x k−1 ) and have also received a particular attention for image segmentation, see e.g. [3,4,5].…”