Abstract:Image segmentation aims to partition an image according to the objects in the scene and is a fundamental step in analysing very high spatial-resolution (VHR) remote sensing imagery. Current methods struggle to effectively consider land objects with diverse shapes and sizes. Additionally, the determination of segmentation scale parameters frequently adheres to a static and empirical doctrine, posing limitations on the segmentation of large-scale remote sensing images and yielding algorithms with limited interpretability. To address the above challenges, we propose a deep-learning-based region merging method dubbed DeepMerge to handle the segmentation of complete objects in large VHR images by integrating deep learning and region adjacency graph (RAG). This is the first method to use deep learning to learn the similarity and merge similar adjacent super-pixels in RAG. We propose a modified binary tree sampling method to generate shift-scale data, serving as inputs for transformer-based deep learning networks, a shift-scale attention with 3-Dimension relative position embedding to learn features across scales, and an embedding to fuse learned features with hand-crafted features. DeepMerge can achieve high segmentation accuracy in a supervised manner from large-scale remotely sensed images and provides an interpretable optimal scale parameter, which is validated using a remote sensing image of 0.55 m resolution covering an area of 5,660 km^2. The experimental results show that DeepMerge achieves the highest F value (0.9550) and the lowest total error TE (0.0895), correctly segmenting objects of different sizes and outperforming all competing segmentation methods.

Another Look on Region Merging Procedure from Seed Region Shift for High-Resolution Remote Sensing Image Segmentation

Decompose Image into Meaningful Regions Based on Contour Detector and Watershed Algorithm

Multiscale segmentation of high-resolution remote sensing images based on region merging

Improvement of Region-Merging Image Segmentation Accuracy Using Multiple Merging Criteria

Hybrid Region Merging Method for Segmentation of High-Resolution Remote Sensing Images

Local Scale-Guided Hierarchical Region Merging and Further Over- and Under-Segmentation Processing for Hybrid Remote Sensing Image Segmentation

Segmentation Quality Evaluation Using Region-Based Precision and Recall Measures for Remote Sensing Images

Accuracy Evaluation of Segmentation for High Resolution Imagery and 3D Laser Point Cloud Data

Interactive Color Image Segmentation by Region Growing Combined with Image Enhancement Based on Bezier Model

Research on multi-scale segmentation algorithm of high-resolution remote sensing imagery based on improved statistical region merging

DeepMerge: Deep-Learning-Based Region-Merging for Image Segmentation

Toward Evaluating Multiscale Segmentations of High Spatial Resolution Remote Sensing Images

An Improved Hierarchical Segmentation Method for Remote Sensing Images

Supervised Region Merging Based on Binary Classification and Active Learning Strategy for Segmentation of Very High Resolution Imagery

Hierarchical Segmentation Evaluation of Region-Based Image Hierarchy

Semi-supervised Remote Sensing Image Segmentation Using Dynamic Region Merging.

Improved Mean-Shift Segmentation Approach for Remote Sensing Images

Segmentation for High-Resolution Optical Remote Sensing Imagery Using Improved Quadtree and Region Adjacency Graph Technique

A Fast and Effective Method for Unsupervised Segmentation Evaluation of Remote Sensing Images

Segmentation of Very High Resolution Imagery Based on Multichannel Watershed Transformation and Region Merging

A new segmentation method for very high resolution imagery using spectral and morphological information