Barry M. Evans

Synergism between human interpretation and digital pattern recognition in preparation of thematic maps

Abstract Recent history of thematic mapping from remotely sensed data has been characterized by a dichotomy between human photointerpretation methods and digital pattern recognition methods. Despite a decade of effort to achieve completely automated delineation of thematic classes, photointerpretive techniques remain the standard of comparison for detailed thematic mapping. Computerized pattern recognition techniques handle certain thematic categories reasonably well, but fail to produce satisfactory results for categories having substantial elements of spatial context inherent in their definitions. An integrated manual/digital approach to thematic mapping has been developed that permits utilization of computers and pattern recognition in conjunction with human contextual analysis capabilities. The human analyst remains in control throughout the mapping process, being able to interact with and override algorithmic designations at any stage. Interactive workstations and integration of digital image analysis systems with advanced geographic information systems permit compositing of data sources while promoting partnership between man and machine.