Vista scenic beauty estimation model: An application of integrating neural net and geographic information system

Abstract

There are some issues that have to be addressed for further understanding and improving scenic beauty management. First, the conventional model, preference rating based on fixed scene and direction, may not sufficiently reflect the reality of visual experience. Rather, visual and scenic preference is construed of a spatial experience. Second, the predictors are chosen based on measuring the composition of landscape features shown in the image. The measurement may not necessarily represent the contents of the physical environment. Third, judgements of scenic preference are complicated tasks. Simple linear regression analysis, with limited degree of freedom and some statistical constraints, may not represent the complexity of human judgments. An integrated model was developed by integrating the Scenic Beauty Estimation (SBE) model (Terry, 1976), the geographic information system (GIS) and, the artificial neural network (ANN). The results suggested the integrated model might be utilized as an automatic scenic preference mechanism for policy making. Implications for future research are also suggested.