Abstract:Abstract. Semantic segmentation of high-resolution remote sensing images based on deep learning has become a hot research topic and has been widely applied. At present, based on the structure of the convolutional neural network, when extracting target features through multiple layer convolutional layers, it is easy to cause the loss of small target features and fuzzy boundary of ground object classification. Therefore, we propose a remote sensing image semantic segmentation method P-Net to detect small target and enhance edge feature. The proposed network was based on an Encoder-Decoder structure. The decoder included the following components: a progressive small target feature enhancement network (IFEN), a boundary thinning module (BRM), and a feature aggregation module (FIAM). Firstly, the dense side output features of the encoder network were utilized to learn and acquired small target feature information and target edge features. Secondly, the pyramid segmentation attention module was introduced to effectively extract fine-grained and multi-scale spatial information. This module enhanced the feature expression of small targets and obtained high-level semantic feature information. The boundary refinement module was designed to refine the low-level spatial feature information extracted by the encoder. Finally, in order to improve the accuracy of remote sensing image object segmentation boundaries, skip connections were used to fuse high-level semantic information and low-level spatial information acrossed layers. These skip connections had the same spatial resolution but different semantic information. In this paper, six evaluation indices including mean intersection over union, frequency weighted intersection over union, pixel accuracy, F1, recall, and precision were used to verify on two public datasets of high-resolution remote sensing images, Gaofen image dataset (GID) and wuhan dense labeling dataset (WHDLD). In the GID dataset, each index reached 78.90%, 78.87%, 87.76%, 87.74%, 87.51%, and 88.04%, respectively; in the WHDLD dataset, each index reached 63.21%, 75.20%, 84.67%, 75.79%, 76.56%, and 75.45%, respectively. The results show that the performance of proposed method is better than that of DeepLabv3+, U-NET, PSPNet, and DUC_HDC methods. More precisely, the recognition performance of small target features is better, and the boundary obtained between object categories is clearer.

A New Instance Segmentation Model for High-Resolution Remote Sensing Images Based on Edge Processing

Query-Based Cascade Instance Segmentation Network for Remote Sensing Image Processing

HQ-ISNet: High-Quality Instance Segmentation for Remote Sensing Imagery

Multi-scale Segmentation of Remote Sensing Image Based on Watershed Transformation

Fuzzy neighbourhood neural network for high-resolution remote sensing image segmentation

Hi-ResNet: Edge Detail Enhancement for High-Resolution Remote Sensing Segmentation

Remote sensing image semantic segmentation method based on small target and edge feature enhancement

Deep Hierarchical Representation And Segmentation Of High Resolution Remote Sensing Images

Object Detection and Instance Segmentation in Remote Sensing Imagery Based on Precise Mask R-CNN

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

Instance Segmentation in Very High Resolution Remote Sensing Imagery Based on Hard-to-Segment Instance Learning and Boundary Shape Analysis

Accurate Instance Segmentation for Remote Sensing Images via Adaptive and Dynamic Feature Learning

An Object-Oriented Remote Sensing Image Segmentation Approach Based on Edge Detection

Boosting Semantic Segmentation of Remote Sensing Images by Introducing Edge Extraction Network and Spectral Indices

Enhanced Remote Sensing Instance Segmentation with Feature Fusion

High-Resolution Remote Sensing Image Segmentation Algorithm Based on Improved Feature Extraction and Hybrid Attention Mechanism

Spatial--spectral FFPNet: Attention-Based Pyramid Network for Segmentation and Classification of Remote Sensing Images

An Instance Segmentation Based Framework for Large-Sized High-Resolution Remote Sensing Images Registration