Object-oriented change detection for multi-source images using multi-feature fusion

With the development of remote sensing technology, the source of data is getting more abundant and the resolution is becoming higher. Consequently, conventional change detection method can't meet the application requirements any more. In this paper, an object-oriented change detection method for multisource remote sensing images using multi-feature fusion was proposed to solve this problem. On the basis of objects acquisition and multiple features extraction, SVM was adopted for its outstanding character in high dimensional data classification. Through the efficient combination of binary classification algorithm based on SVM and object-oriented change detection, the accuracy and reliability of change detection for multi-source images were increased. With manual visual judgment, a computing method for ground objects oriented evaluation index was designed. The experiments were conducted among multi-source and multi-temporal images, and the change detection accuracy of different ground objects were counted, which verified the effectiveness of this method.