| 摘要: |
| 城市街道景观指数提取为城市街景研究量化途径之一。结合已有指数，提出用全景静态图的不同数据类型，分类计算视域景观指数的优化方法；提出基于计算机视觉尺度不变特征转换（SIFT）的关键点邻域尺度区间频数；采用层次聚类分析指数不同的邻里尺度最优簇数、贡献度，确定分布特征重要指数组成；探索不同指数作用于不同邻里尺度的特征效应。研究发现视域景观指数具有不同邻里尺度效应。研究指数中城市街道空间的绿视率和天空开阔度（空间组成层面）、关键点邻域尺度（0，10]和（10，20]区间频数（对象尺度层面），以及色彩丰富度指数（颜色层面）是构成城市街道分布特征的重要特征指数。通过确定不同邻里尺度城市街道特征分布，可以为城市街道空间的景观质量提升、量化管理和城市微更新提供参照。 |
| 关键词: 风景园林 城市街道空间 特征指数 全景静态图 计算机视觉 层次聚类 |
| DOI：10.14085/j.fjyl.2022.09.0041.07 |
| 分类号:TU981 |
| 基金项目: |
|
| Urban Street View Landscape Indices and Neighborhood-Scale Feature Effects |
|
BAO Ruiqing1, (ECU) Alexis Arias Betancourt2
|
|
1.Xi’an University of Architecture and Technology;2.Illinois Institute of Technology
|
| Abstract: |
| The extraction of urban streetscape index is a quantitative approach for urban streetscape research. In combination with existing indices, this research proposes an optimization method for calculating view landscape indices by category using different types of data shown in the panoramic static image, and proposes the frequency of the keypoint neighborhood-scale interval based on scale-invariant feature transform (SIFT) from computer vision. Moreover, the research adopts the hierarchical clustering method to analyze the optimal number of clusters, the degree of contribution of different neighborhood scales, and determine the composition of important indices constituting distribution features. Finally, the research explores the feature effects of different indices on various neighborhood scales. Research results show that, view landscape indices have different neighborhood scale effects. Furthermore, for the indexes studied, the green view index and sky view factor (spatial composition level), the frequency of the keypoint neighborhood-scale intervals (0, 10] and (10, 20] (object scale level), and the color richness index (color level) are important feature indices that constitute urban street distribution features. Determining the distribution of urban street features at various neighborhood scales can provide reference for landscape quality improvement, quantitative management, and urban micro-renewal of urban street spaces. |
| Key words: landscape architecture urban street space feature indices panoramic static image computer vision hierarchical clustering |
