Knowledge-based texture image segmentation using iterative linked quadtree splitting

A knowledge-based texture image segmentation system is discussed in which knowledge is used at the segmentation level. The system is characterized by a control mechanism based on an iterative linked quadtree splitting scheme. The main advantages of this system include the possibility of incorporating knowledge from diverse sources and with different scales, and the classification process is balanced and less dependent on the order in which the image is processed. The performance of the system is illustrated on two test images from totally different applications. The first is a natural texture image of an outdoor scene, and the second is a seismic image of stacked seismic traces used in interpretation of the Earth´s subsurface geology