Abstract

The responsibility to manage vineyards in the European Union belongs to a large range of organizations, which need detailed information about geographic data. For this purpose, most member states have developed vineyard registers. This paper has explored an inferential system for vineyard detection using digital aerial photographs. The system has been inspired by a recent memory prediction theory and models the high-level architecture of the human neocortex. In this study, the hierarchical architecture and recognition performance of this Bayesian model were described and applied. Using a photogram received by a photogrammetric UltraCamD® sensor of Vexcel, 96% of the parcels has been detected. The automatic process developed can be easily integrated into the final user’s geographical information system and produces useful information for vineyard management.

Key words: Memory-prediction theory, nupic, ultracamd sensor, hierarchical temporal memory, vineyards.