Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

How to Identify Good Superpixels for Deforestation Detection on Tropical Rainforests

Isabela Borlido, Eduardo Bouhid, Victor Sundermann, Hugo Resende, Alvaro Luiz Fazenda, Fabio Faria, Silvio Jamil F. Guimarães

TL;DR

According to the experiments, superpixel methods with better tradeoffs among delineation, homogeneity, compactness, compactness, and regularity are more suitable for identifying good superpixels for deforestation detection tasks.

Abstract

The conservation of tropical forests is a topic of significant social and ecological relevance due to their crucial role in the global ecosystem. Unfortunately, deforestation and degradation impact millions of hectares annually, requiring government or private initiatives for effective forest monitoring. However, identifying deforested regions in satellite images is challenging due to data imbalance, image resolution, low-contrast regions, and occlusion. Superpixel segmentation can overcome these drawbacks, reducing workload and preserving important image boundaries. However, most works for remote sensing images do not exploit recent superpixel methods. In this work, we evaluate 16 superpixel methods in satellite images to support a deforestation detection system in tropical forests. We also assess the performance of superpixel methods for the target task, establishing a relationship with segmentation methodological evaluation. According to our results, ERS, GMMSP, and DISF perform best on UE, BR, and SIRS, respectively, whereas ERS has the best trade-off with CO and Reg. In classification, SH, DISF, and ISF perform best on RGB, UMDA, and PCA compositions, respectively. According to our experiments, superpixel methods with better trade-offs between delineation, homogeneity, compactness, and regularity are more suitable for identifying good superpixels for deforestation detection tasks.

How to Identify Good Superpixels for Deforestation Detection on Tropical Rainforests

TL;DR

According to the experiments, superpixel methods with better tradeoffs among delineation, homogeneity, compactness, compactness, and regularity are more suitable for identifying good superpixels for deforestation detection tasks.

Abstract

The conservation of tropical forests is a topic of significant social and ecological relevance due to their crucial role in the global ecosystem. Unfortunately, deforestation and degradation impact millions of hectares annually, requiring government or private initiatives for effective forest monitoring. However, identifying deforested regions in satellite images is challenging due to data imbalance, image resolution, low-contrast regions, and occlusion. Superpixel segmentation can overcome these drawbacks, reducing workload and preserving important image boundaries. However, most works for remote sensing images do not exploit recent superpixel methods. In this work, we evaluate 16 superpixel methods in satellite images to support a deforestation detection system in tropical forests. We also assess the performance of superpixel methods for the target task, establishing a relationship with segmentation methodological evaluation. According to our results, ERS, GMMSP, and DISF perform best on UE, BR, and SIRS, respectively, whereas ERS has the best trade-off with CO and Reg. In classification, SH, DISF, and ISF perform best on RGB, UMDA, and PCA compositions, respectively. According to our experiments, superpixel methods with better trade-offs between delineation, homogeneity, compactness, and regularity are more suitable for identifying good superpixels for deforestation detection tasks.
Paper Structure (8 sections, 3 figures, 2 tables)

This paper contains 8 sections, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Superpixel segmentation
  3. Experimental setup
  4. Results and Discussion
  5. Superpixel segmentation
  6. Classification
  7. Discussion
  8. Conclusions

Figures (3)

  • Figure 1: Example of satellite image for deforestation. In (a) the original image and (b) the ground truth mask, in which forest regions are in green, recent deforestation in red, and old deforestation in black. Most classification methods incorporate a (c) SLIC segmentation to reduce workload, while recent methodologies, such as (d) DISF, remain unexplored.
  • Figure 2: Superpixel results for object delineation (BR and UE) on top, color homogeneity (EV and SIRS) in the middle, and compactness (CO) and regularity (Reg) at the bottom.
  • Figure 3: Original image with PCA composition and segmentation with 6000 superpixels for SLIC, ERS, SH, ETPS, DISF, GMMSP, and ISF. In superpixel segmentations, the superpixel borders are indicated in yellow and recently deforested regions are colored in red.