Research on the Fusion of Superpixels in High-Resolution Remote Sensing Images

With the development of small UAV remote sensing technology and the gradual improvement of observation system, high-resolution remote sensing images become easy to obtain, and China has entered the era of application of high-resolution remote sensing images. Traditional image segmentation methods treat pixels as processing units, focusing solely on spectral information while neglecting the influence of other image features. This approach fails to meet the accuracy and efficiency requirements for practical applications in high-resolution remote sensing image segmentation. Representing a high-resolution remote sensing image with superpixels aligns better with human visual perception, reduces information redundancy, and decreases data processing volume. Therefore, superpixel-based segmentation methods for high-resolution remote sensing images have become an important research direction. This paper primarily investigates the fusion of superpixels in high-resolution remote sensing images, elaborating on the principles of the fuzzy C-means clustering algorithm and explaining a proposed spatially constrained fuzzy clustering method that integrates spectral differences. Qualitative and quantitative analyses of the experimental results are conducted to validate the effectiveness and reliability of the proposed algorithm.