MORPHOLOGICAL PARTITIONING OF MULTISPECTRAL IMAGES

The partitioning of an image may be defined as the division of its ima ge plane into image objects. This paper presents a new methodology for partitioning multispectral images. It combines the analysis of the sp ectral properties of the pixels with the analysis of their spatial pro perties. Spectral properties are studied in the multivariate histogram of the image, while spatial properties are analyzed in the multispect ral gradient of the image. The histogram of the image is first segment ed by a nonparametric algorithm, The segmented histogram allows the cl assification of all image pixels. Each resulting class is then separat ely filtered in order to remove all classified pixels having a high pr obability of being misclassified when considering spatial criteria. Th e filtered classes are used as seeds for boundary detection on the gra dient of the original image. The class of the resulting regions is giv en by the class of the seeds that created those regions. (C) 1996 SPIE and IS&T.