Spectral unmixing is a challenging issue in hyperspectral image (HSI) processing, which aims at decomposing the mixed pixels of hyperspectral images to extract the end elements and calculate their corresponding abundances. Due to the limited spatial resolution of hyperspectral images, spectral unmixing is of great significance to the HSI understanding task, such as fine land cover classification, anomaly detection, object recognition, etc. Additionally, spectral unmixing also plays a positive role on HSI processing task. For instance, accurate endmember extraction contributes to enhancing the HSI super-resolution performance. Further, the development of deep learning techniques not only greatly improves the accuracy of various detection or classification tasks, but also puts forward requirements for the quantity and quality of training samples. In this case, spectral unmixing is very necessary to achieve data augmentation or few-shot learning encountering the insufficient or class-imbalanced training samples. In this special issue, we welcome the papers related to innovative spectral unmixing techniques, the applications of unmixing to classification, anomaly detection, recognition, band selection, super-resolution, data augmentation, etc.