Object identification and classification in a high resolution satellite data using data mining techniques for knowledge extraction

The interpretation of remotely sensed images in a spatiotemporal context is becoming a valuable research topic. Today's optical sensor systems on satellite provide large-area images with 1-m resolution and better, which can deliver complement information to traditional acquired data. However, the constant growth of data volume in remote sensing imaging makes reaching conclusions based on collected data a challenging task. One of the main problems that arise during the data mining process is treating data that contains temporal information. However, two important issues must be considered in order to provide more accurate decisions on object identification and pattern recognition. First, the continuous growth of the dataset storage space and the advances in remote sensing sensors which generate a huge amount of satellite images making the manual image interpretation a difficult task. Second, the space/time components are inherent to satellite images; systems being developed to identify objects must take into account the spatiotemporal context to better interpret the collected image data. Spatial relations between objects are widely used in context-based image retrieval. This paper outlines the challenges and proposes in creation of a data mines capable of supporting the requirements of the system, which, inevitably demand a high level of cooperation between many disparate sources of spatial data.