Abstract
Remote sensing image segmentation with samples imbalance is always one of the most important issues. Typically, a high-resolution remote sensing image has the characteristics of high spatial resolution and low spectral resolution, complex large-scale land covers, small class differences for some land covers, vague foreground, and imbalanced distribution of samples. However, traditional machine learning algorithms have limitations in deep image feature extraction and dealing with sample imbalance issue. In the paper, we proposed an improved full-convolution neural network, called DeepLab V3+, with loss function based solution of samples imbalance. In addition, we select Sentinel-2 remote sensing images covering the Yuli County, Bayingolin Mongol Autonomous Prefecture, Xinjiang Uygur Autonomous Region, China as data sources, then a typical region image dataset is built by data augmentation. The experimental results show that the improved DeepLab V3+ model can not only utilize the spectral information of high-resolution remote sensing images, but also consider its rich spatial information. The classification accuracy of the proposed method on the test dataset reaches 97.97%. The mean Intersection-over-Union reaches 87.74%, and the Kappa coefficient 0.9587. The work provides methodological guidance to sample imbalance correction, and the established data resource can be a reference to further study in the future.
Highlights
Remote sensing images have become the main data source for obtaining land-use information at broad spatial scales [1]
The study area is located between the Taklamakan Desert and the Kumtag Desert, and it is of great significance to understand the ecological situation of the study area by mastering the distribution of ground features
In this paper, when considering high-resolution remote sensing image space has the characteristics of high resolution, low spectral resolution, large-scale size with complex land covers, and unclear image prospects resulting from large intra-category differences or small inter-category differences
Summary
Remote sensing images have become the main data source for obtaining land-use information at broad spatial scales [1]. The remote sensing image data that were obtained from high-resolution remote sensing satellite sensors have rich texture information, spatial information, and more obvious ground geometry. Remote sensing data obtained from hyperspectral remote sensing satellite sensors has rich spectral information, but the spatial resolution is often not high enough. Image segmentation algorithm has a strong ability to extract spectral and spatial features, which makes more researchers introduce it into remote sensing image classification [2]. The classification methods of remote sensing images mainly relied on prior knowledge and classification was always based on the differences in the spectral characteristics of ground features. The selected model is generally trained by selecting feature bands and regions of interest These traditional machine learning methods rely more on spectral features and ignore the spatial features and texture information of high-resolution remote sensing images
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call