Object extraction of remote sensing image by agent and group collaboration technology

摘要

Because different types of ground objects have different spatial and spectrum characteristics, automatic and accurate extraction of different thematic objects from remote sensing without spatial conflict and topology conflict is still a challenge. The design of special thematic intelligent Agents to recognize special objects in high-resolution remote sensing images is useful for object extraction and thematic geo-database updating. This paper gives a framework about object extraction of remote sensing image by using Agent and group collaboration technology. A single Agent generates a thematic map of a single type of ground objects by using prior knowledge, and then the thematic map is compared with the thematic maps generated by other Agents. An algorithm is used to detect and define conflict objects. The conflict objects will be classified again. Repeatedly checking the classification results to reduce the phenomenon of incorrect and conflict classification and to achieve the purpose of improving recognition accuracy.