This paper describes an approach for automatic forest boundary delineation in aerial images with special emphasis on spatially contiguous and reproducible results. The JSEG segmentation method has been found to produce meaningful image segmentation results. Nevertheless, its outcome turned out not to be sufficient to match the semantic meaning required for forest vegetation. To overcome this conceptual limitation the combination with wavelet transformation gives access to additional texture features and leads to robust forest boundaries. Preliminary results are encouraging, especially without any subjective decisions. The ongoing research focuses on the integration of digital terrain and surface data and other texture features in order to increase accuracy and robustness of forest delineation.