Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Learning for In-Orbit Cloud Segmentation and Classification in Hyperspectral Satellite Data

Daniel Kovac, Jan Mucha, Jon Alvarez Justo, Jiri Mekyska, Zoltan Galaz, Krystof Novotny, Radoslav Pitonak, Jan Knezik, Jonas Herec, Tor Arne Johansen

TL;DR

Results indicate that 1D-Justo-LiuNet achieves the highest accu-racy, outperforming 2D CNNs, while maintaining compactness with larger spectral channel sets, albeit with increased inference times, however, the performance of 1D CNN degrades with significant channel reduction.

Abstract

This article explores the latest Convolutional Neural Networks (CNNs) for cloud detection aboard hyperspectral satellites. The performance of the latest 1D CNN (1D-Justo-LiuNet) and two recent 2D CNNs (nnU-net and 2D-Justo-UNet-Simple) for cloud segmentation and classification is assessed. Evaluation criteria include precision and computational efficiency for in-orbit deployment. Experiments utilize NASA's EO-1 Hyperion data, with varying spectral channel numbers after Principal Component Analysis. Results indicate that 1D-Justo-LiuNet achieves the highest accuracy, outperforming 2D CNNs, while maintaining compactness with larger spectral channel sets, albeit with increased inference times. However, the performance of 1D CNN degrades with significant channel reduction. In this context, the 2D-Justo-UNet-Simple offers the best balance for in-orbit deployment, considering precision, memory, and time costs. While nnU-net is suitable for on-ground processing, deployment of lightweight 1D-Justo-LiuNet is recommended for high-precision applications. Alternatively, lightweight 2D-Justo-UNet-Simple is recommended for balanced costs between timing and precision in orbit.

Deep Learning for In-Orbit Cloud Segmentation and Classification in Hyperspectral Satellite Data

TL;DR

Results indicate that 1D-Justo-LiuNet achieves the highest accu-racy, outperforming 2D CNNs, while maintaining compactness with larger spectral channel sets, albeit with increased inference times, however, the performance of 1D CNN degrades with significant channel reduction.

Abstract

This article explores the latest Convolutional Neural Networks (CNNs) for cloud detection aboard hyperspectral satellites. The performance of the latest 1D CNN (1D-Justo-LiuNet) and two recent 2D CNNs (nnU-net and 2D-Justo-UNet-Simple) for cloud segmentation and classification is assessed. Evaluation criteria include precision and computational efficiency for in-orbit deployment. Experiments utilize NASA's EO-1 Hyperion data, with varying spectral channel numbers after Principal Component Analysis. Results indicate that 1D-Justo-LiuNet achieves the highest accuracy, outperforming 2D CNNs, while maintaining compactness with larger spectral channel sets, albeit with increased inference times. However, the performance of 1D CNN degrades with significant channel reduction. In this context, the 2D-Justo-UNet-Simple offers the best balance for in-orbit deployment, considering precision, memory, and time costs. While nnU-net is suitable for on-ground processing, deployment of lightweight 1D-Justo-LiuNet is recommended for high-precision applications. Alternatively, lightweight 2D-Justo-UNet-Simple is recommended for balanced costs between timing and precision in orbit.
Paper Structure (8 sections, 3 figures, 3 tables)

This paper contains 8 sections, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Dataset
  3. Methodology
  4. Principal Component Analysis
  5. Deep Learning Modeling
  6. Results
  7. Discussion
  8. Conclusion

Figures (3)

  • Figure 1: Histogram Illustrating the Dataset's Tile Count (Left Y-axis) and Class Distribution (Right Y-axis) Relative to Cloud Coverage (X-axis).
  • Figure 2: PCA and Correlation Matrix of the No Cloud Class: Identification of Correlation Clusters Highlighted in Black, with Emphasis on Peak Weights Highlighted in Red.
  • Figure 3: Inference Results For Two Tiles From the Commercial Ziyuan-1 02 Dataset. The Top Tile Depicts Varying Cloud Thickness in an RGB Composite, While the Bottom One Includes a Snow-covered Mountainous Terrain. Next, the Predicted Masks Generated by the CNN Models for 6-channel Inference.