Author(s)

A. TSOUCHLARAKI

Abstract

This study aims to investigate indices for the automatic evaluation and classification of landscape quality using digital ground photographs. Research efforts to date are scarce on automated extraction of qualitative information based on photographs and therefore this study contributes in this respect, i.e. the automated quantification of landscape qualitative characteristics. Based on the texture indices that are commonly used in landscape analysis, eight quantitative indices are selected and the results from the application of these indices to a sample of ground photographs are described in this paper. These indices are richness, fragmentation, diversity, dominance, grouping and complexity. Furthermore, we investigate the effectiveness of the indices selected as to the classification of the landscape’s qualitative characteristics, such as relief morphology, visibility, water existence, vegetation patterns, etc. The results are compared to the results derived from a research programme of the National Technical University of Athens, in which the qualitative characteristics of the landscapes depicted in the same samples of ground photographs have been manually assessed based on the scientific opinion of seven experts. Comments and suggestions are presented based on the comparison for further investigation. The main conclusion of the investigation is that the texture measurement indices are sensitised in the landscape’s qualitative characteristics, a fact that is positive and encouraging enough in order to pursue further research.

Keywords

landscape evaluation and classification, texture indices, digital ground photographs, geographical information systems.