On Adaptive Image Segmentation of Remotely Sensed Imagery

A critical step in object-oriented geospatial analysis (OBIA) is image segmentation. A single set of parameters is often not effective segmenting an image. To solve this problem, an adaptive approach to image segmentation has been proposed, which utilizes segments determined from a lower-spatial resolution image as the context to analyse a corresponding image at a higher-spatial resolution to create multiple sets of segmentation parameters to address the needs of different parts of the image. However, due to inherent differences in perceptions of a scene at different spatial resolutions and co-registration, segment boundaries from the low spatial resolution image need to be adjusted before they are applied to the high-spatial resolution image. This is a non-trivial task due to considerations such as noise, image complexity, and determining appropriate boundaries. Accordingly, an innovative method was developed. Adjustments were executed for each boundary pixel based on the minimization of an energy function characterizing local homogeneity. Adjustments are based on a structure which rewarded movement towards edges, and superior changes towards homogeneity. The adjusted segments act as the basis for the determination of segmentation parameters through a variogram based method. The developed method was tested on a set of Quickbird, and ASTER images, from a study area in Ontario, Canada. Results showed that the adjusted segmentation boundaries obtained from the lower resolution imagery were aligned well with the features in the Quickbird imagery, and segmentation maps determined using the adaptive segmentation method were superior to those created by a non-adaptive approach. This work will allow users to more easily and quickly segment large high resolution images.