A quantitative spatial model was made for visual - aesthetic assessment of cultural landscape in parts of Notranjska and Primorska regions. The main input variables were spatial patterns of forest edge and spontaneous afforestation of abandoned farmland. The model is intended for regional-scale analyses and as a tool for focusing more detailed assessment to particular problem areas. The model combines the expert approach (a priori definition of assessment criteria based on expert opinion) with a statistical approach (a regression model, explaining existing assessment of other authors with GIS-based physical landscape indices). The map of spontaneous afforestation was made by intersecting legacy forest map with an up-to-date forest edge map.The latter was made by numerical interpretation of a Landsat TM satellite image, where unsupervised classification was combined with decision tree basedclassification. The decision trees were generated by machine learning with the CORINE Land Cover database representing the target classes. The visual - aesthetic assessment model was used for defining areas of preferential protection against spontaneous afforestation.