Object-oriented image classification recognizes that important semantic information is not represented in single pixels but in meaningful objects in an image and their mutual relations, i.e. the context. This process feeds specialized software with imagery from satellites or aircraft. The software then tries to find and report significant features in the images. Object-oriented classification software assumes that related pixels are actually part of objects, and assigns properties and relationships to the whole object rather than individual pixels.

The object-oriented approach can speed up and increase the accuracy of a process that automatically extracts information from very high resolution aerial and satellite imagery like IKONOS and QuickBird. Automatic classification of digital images has been traditionally carried out pixel-by-pixel. Pixel-by-pixel algorithms are slow when applied to high resolution images.

An urban image is also better classified by the objects in it, rather than by pixels. Many urban land cover types, such as roads, buildings, parking lots, etc., are spectrally similar. So, spatial information such as texture and context must be exploited to produce more accurate urban maps. Since these are inherently properties that cover several pixels, object-oriented classification can use them more easily than pixel-by-pixesl methods. Pixel-by-pixel methods are slower and less accurate when classifying images of urban environments, which consist of a mosaic of small-scale features made up of different materials.

Digital images can be classified by themes such as context, shape, form, texture, color and topological attributes. With the right selection of themes, object-oriented image classification can be applied to studies of geography, military intelligence, ecology, geology, and marine science.

The only software currently available for object oriented classification is called eCognition. eCognition is a very powerful software developed by Definiens AG, a multinational company headquartered in Munich, Germany and founded by Physics Nobel Prize Laureate, Dr. Gerd Binnig.

Earth sciences | Ecology | Satellites | Earth observation satellites