A Study of Fast Change Detection Algorithm Based on Feature Library of Remote Sensing Imagery

The features of remote sensing imagery like the large amount of data and the changeful of terrain features decide the complexity of the change detection technology. In allusion to the requirement of the efficiency and the precision for large-scale detection, at the base of the conclusions of the change detection technology all over the world, the key point is to study an algorithm of fast change detection based on feature library of remote sensing imagery. Under the high performance colony parallel processing environment, taking the incorporate change detection technical flow and the system frame as the backing, this paper analyzed the features of the representative terrain feature, established a feature library, and studied an algorithm based on the feature library of remote-sensing imagery. With the help of typical terrain samples feature library and classify system, matched the typical samples of the feature library with the object features by full wave bands segmentation and vector analysis, namely, automatically change detection. At last, it realized the change information vector output using the storied overlay strategy. Taking QUICKBIRD of Shenzhen area as the tentative - data, used this method to carry on change detection. The results indicate that this method enhances in the precision and the efficiency obviously compared to the traditional methods. Therefore, it may apply in the practice for the large-scale imagery of change detection.