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

Agricultural field boundary information is important in many agricultural management practices and remotely sensed imagery can present valuable, timely, information about land parcels. This paper presents a multispectral segmentation method for automated delineation of agricultural field boundaries in digital imagery. Edge information from a gradient edge detector is integrated with region growing, where the multispectral information from all selected image bands is utilized in both procedures. In this way, information from several spectral bands (and/or different dates) can be used for delineating field boundaries with different characteristics. This is particularly important for agricultural applications, where multidate information is necessary to differentiate between, for instance, crops at different phenological growing stages. The results show that Landsat TM images are easier to delineate than SPOT images due to lower spectral variance between adjacent pixels.