道路绿视率不同识别方法比较研究——以徐州市为例

A Comparative Study on Different Identification Methods for Road Green Wiew Index——A Case Study of Xuzhou City

探究不同识别方法对绿视率的影响,对于提高绿化识别准确率以及绿视率标准的制定有重要意义。以徐州市3类特征用地为研究对象,运用3种不同识别方法对道路移动采样数据集进行绿视率提取,进而分析不同方法对道路绿视率识别值以及准确率的影响。结果表明,1)基于RGB、HSL颜色的方法对于绿视率值存在低估,机器学习SegNet方法在识别精度上优于基于颜色提取的RGB和HSL方法;2)不同用地分组和图像亮度与绿视率差异和精度差异相关性不显著,色偏系数与绿视率差异呈显著正相关。研究结果为政府绿视率标准制定和绿视率监测工作提供了方法论的实证参考及建议。

It is of great significance to explore the influence of different identification methods on green view index(GVI)to improve the accuracy rate of GVI and the formulation of GVI standard.Taking three types of characteristic sections in Xuzhou City as the research objects,three different recognition methods were used to extract road GVI in local mobile datasets.The influence of different methods on the recognition value and accuracy of road GVI was analyzed.The results showed that 1)the method based on RGB and HSL color underestimated the value of green apparent ratio.The SegNet method based on machine learning was superior to the RGB and HSL methods based on color extraction in recognition accuracy.2)The group of different sections and image brightness had no significant correlation with the difference of GVI and the difference of accuracy,while the color bias coefficient had a significantly positive correlation with the difference of GVI.The results of this study provide methodological empirical reference and suggestion for the government GVI standard formulation and GVI monitoring.