Integrated method for boundary delineation of agricultural fields in multispectral satellite images

Most agricultural statistics are calculated per field, and it is well known that classification procedures for homogeneous objects produce better resu lts than per-pixel classification. In this study, a multispectral segmentat ion method for automated delineation of agricultural field boundaries in re motely sensed images is presented. Edge information from a gradient edge de tector is integrated with a segmentation algorithm. The multispectral edge detector uses all available multispectral information by adding the magnitu des and directions of edges derived from edge detection in single bands. Th e addition is weighted by edge direction, to remove "noise" and to enhance the major direction. The resulting edge from the edge detection algorithm i s combined with a segmentation method based on a simple ISODATA algorithm, where the initial centroids are decided by the distances to the edges from the edge detection step. From this procedure, the number of regions will mo st likely exceed the actual number of fields in the image and merging of re gions is performed. By calculating the mean and covariance matrix for pixel s of neighboring regions, regions with a high generalized likelihood-ratio test quantity will be merged. In this way, information from several spectra l hands (and/or different dates) can be used for delineating field borders with different characteristics. The introduction of the ISODATA classifier compared with a previously used region growing procedure improves the outpu t. Some results are compared with manually extracted field boundaries.