Determining patterns of land degradation, and by inference the ecological processes at play, is critical to designing and implementing strategies and tactics to halt degrading practices and restore already degraded areas. In this paper we describe the application of a new image analysis technique to determine the presence and density of shrubs in a grassland matrix and evaluate its use for assessing rates and patterns of shrub increase, a common form of land degradation in rangelands throughout the world. By analysing a series of images derived from aerial photography to identify and separate objects (shrubs) in progressive steps, we were able to detect accurately individual shrubs with canopies greater than 9 m²(about 85% of the sample population). The method also was moderately successful in predicting canopy area of individual trees. This approach offers land managers and policy-makers a tool for defining both temporal patterns and rates of shrub increase as well as for defining the current extent of an invasion as a basis for targeting control technologies.