Contents I. Introduction
Remote sensing change detection identifies changes occurring on the earth’s surface by jointly processing multitemporal images acquired from the same geographical area at different times [1]–[3]. This technique has become an increasingly popular research topic due to its relevant and practical applications, including deforestation, damage assessment, disaster monitoring, and urban expansion [4]. In the past decades, a variety of change detection approaches have been developed. Generally, these approaches can be divided into two steps: difference generation and analysis. The difference generation methods include image differencing, image ratio and change vector analysis (CVA) [5], while the analysis step adopts threshold-based methods [6], [7] and clustering-based methods [8], [9]. With the emergence of very high resolution (VHR) remote sensing images, more detailed ground information can be obtained. However, increasing spatial resolution always leads to a reduction in the ability to distinguish spectral statistics between different classes. The conventional change detection methods only utilize spectral information regardless of spatial information, which is inadequate for VHR remote sensing image change detection. Nonetheless, VHR remote sensing images contain abundant spatial and contextual information, which can aid in accurate change detection.
