A Change Detection Method with High Resolution Images Based on Polygon Automatic Validating

In Beijing, the earth surface changes has reached 5% each year, and artificial detection with high resolution images has been applied to peccancy building monitoring and road change detection. The main problem of artificial detection is its low efficiency and existing omitted changes. Therefore, a practicable automatic or semiautomatic change detection method with high resolution images is needed.Because of abundant details and coarse textures in high resolution images, traditional automatic or semiautomatic pixel-based change detection method can not get satisfiable detection results. Therefore, in this paper, considering the characters of high resolution images and algorithmic automation, an object-oriented change detection method with high resolution images based on polygon automatic validating is presented. The main technical flow of the new method includes five steps, which are new image preprocessing, change detection and extraction, polygon processing, polygon automatic validating, artificial polygon edit and statistics output. The principle and main steps of the method will be detailedly introduced.In order to evaluate this change detection method, the relative experiments are carried out. The main image sources are IKONOS and QuickBird of urban and suburban in Beijing. The detection results are compared with practical change results by artificial interpretation. According to experimental analysis, the advantages of this method are concluded: firstly, this method can keep polygons completeness and have less omitted polygons. Secondly, according to statistics of operational time, this method can save artificial edit workload and improve the whole detection efficiency. Thirdly, this method can eliminate lots of wrong detection polygons and improve the overall accuracy.