Abstract:
This study sets out from the extension of environmental comfort theory and the practical application of big data approach using city-scale streetscape images. The streetscape comfort in Beijing and Shanghai, and the comfort-relevant street indicators are investigated. The difference of streetscape comfort between cities were discussed in this paper as well. The image semantic segmentation was adopted to classify the street images according to their constituent attributes, and then the subjective evaluation of streetscape comfort was collected by online questionnaire. Through the correlation analysis, the relationship between users' perceived comfort and objective street features was successfully established. This study also managed to identify street indicators with positive and negative effects on streetscape comfort. By mapping the evaluation result of comfort in urban space, the difference of comfort perception between Beijing and Shanghai, and between different regions within these cities were reflected. The result shows that the terrain can improve the streetscape comfort in both Beijing and Shanghai, while the wall shows different effects between these cities. The plant have no noticeable effect on improving the streetscape comfort which is different with previous studies. In terms of the distribution of comfort, the streetscape comfort in the old urban areas of Beijing and Shanghai is relatively low. Most of the streetscape with low comfort in Beijing are concentrated in the historic districts within the second Ring Road, while the distribution of comfort in Shanghai streets does not show an obvious rule. This study provides practical suggestions for optimizing the the streetscape comfort in Beijing and Shanghai and explored new possibilities for forming a more refined and multi-dimensional classification standard of street images in the future.