Segmentation and Detection of Extended Structures in Low Frequency Astronomical Surveys using Hybrid Wavelet Decomposition
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
The morphological complexity of extended real structures (such as SNRs, HII regions, bow shocks, etc.), and their wide variety in scale and surface brightness make their automatic detection and segmentation in large surveys a difficult task. We propose in this paper a segmentation method based on applying wavelet decomposition in the residual thresholded images. This strategy avoids the artifacts produced by strong sources in a straight wavelet decomposition. Our method successfully segments extended structures at different scales and therefore is suitable for further morphological analysis and object recognition processes. Results using images from radio and infrared wavelengths surveys show the validity of our approach. 1. Motivation and Objectives Large surveys reveal thousands of low spatial frequency objects, shown at different intensity scales. When imaging rich areas in the interstellar medium, many of the compact sources overlap with objects associated to extended, morphologically complex real structures, such as SNRs, HII regions, bow shocks, etc. The wide variety in spacial scale and surface brightness of these objects make their automatic detection and segmentation a difficult task. To illustrate these facts, we use in this paper the image corresponding to the high galactic longitude end of the Phase I Canadian Galactic Plane Survey (CGPS hereafter, see Taylor et al. 2003). Figure 1 shows, on the top, the image composition corresponding to mosaics V1, V2, W1, W2, X1, X2, Y1 and Y2 of the CGPS 1420 MHz, continuum. We have eliminated 0.1% of the intensity outliers in order to visualize some dozens of sources. Nevertheless, the data corresponding to these mosaics contain thousands of objects, as it is illustrated when we display the sub-image contained in the red square at different intensity scales. On the bottom left of the figure we show the sub-image with 2% of the outliers eliminated, whereas on the right 5% of them have been removed. In addition to a great number of compact sources, several extended structures and their surrounding emission have appeared (the zoomed area contains Lynds Bright Nebula 679). In previous work we focused our attention on the performance of techniques suitable for the detection of faint compact objects (see for example Peracaula et al. 2008). In the present work we focus on the automatic detection of extended and irregular structures for further cataloguing and morphological analysis. In this context, wavelet image decomposition has been proven as a tool that can detect and separate objects represented at different spatial frequencies. However, the high dynamic range in intensity of this kind of image diminishes the performance
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it