本研究引入空間包絡理論以頻率域視覺訊號的統計特徵來建構景觀偏好預測模型，嘗試為景觀研究領域引入新的探討觀點與工具。本研究收集了八種景觀類別的影像共480張（公路、高樓、街道、市區、海岸、森林、鄉村、高山），對頻率域訊號進行抽取，並以主成分分析提取出三個主要的空間包絡特徵，分別命名為空間質感、空間方向與空間深度，以此三個特徵構成預測景觀偏好的計算模型。結果顯示空間質感可以預測市區類型的景觀偏好，而空間方向與空間深度可以預測公路與森林兩種類型的景觀偏好，其餘類別的預測結果未達顯著效果。雖然本研究之模型預測能力尚有改善空間，然而空間包絡理論確實在景觀偏好研究提供了一個新的理論與方向。

This research establishes a prediction model of landscape preference using spatial envelope theory and provides a new perspective and novel tool for landscape research. The 480 images used in this research comprised eight types of scenes, including highways, tall buildings, streets, inner cities, coasts, forests, the countryside, and mountains. Principal component analysis extracted three important features-spatial texture, spatial direction, and spatial depth—from the frequency domain of these images, which were used to construct the prediction model of the landscape preference. The results show that the spatial texture could predict the preference for inner-cities, while spatial direction and spatial depth could predict the preference for roads and forests. Although future research could improve the prediction ability of this model, special envelope theory provides a new direction for landscape research.