期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2006
卷号：XXXVI-4/C42
出版社：Copernicus Publications
摘要：The key step in object-oriented image classification is the segmentation of the image into discrete meaningful objects. Generally the relation between the segmentation parameters and the corresponding segmentation outcome is far from being obvious, and the definition of suitable parameter values is usually done through a troublesome and time consuming trial and error process. This paper proposes a method for the automatic adaptation of segmentation parameters based on Genetic Algorithms. The intuitive and computationally uncomplicated fitness function proposed expresses the similarity of the segmentation result with a reference provided by the user. The method searches the solution space for a set of parameter values that minimizes this fitness function. A prototype including an implementation of a widely used segmentation algorithm was developed to assess performance of the method. A set of experiments on two pairs of LANDSAT and IKONOS images was carried out and the method was able in most cases to come close to the ideal solution
