Abstract: Change detection of construction land is of great significance to the sustainable development of cities. To extract the change information of construction land accurately, this paper proposes a novel construction land change detection method by combining old-time land use vector data, random forest and fuzzy C-means clustering algorithm. According to the prior knowledge provided by the old land use vector data, the construction land is first divided into built-up area construction land and non-built-up area construction land for change detection. Then, object-oriented segmentation is used and various features of the objects, including spectral, GLCM texture, and shape are extracted from high spatial resolution remote sensing images. After applying a local mean assignment and a normalization, multiple feature sets of objects are obtained. Finally, random forest or fuzzy C-means clustering algorithm is employed to achieve the final change detection result in different scene. Our experiments demonstrate that, the proposed method can effectively improve the extraction of change information of construction land in complex scenes with high performance.

Key words: construction land, change detection, object-oriented, random forest, fuzzy C-means