Exploring Urban Street Green Perception from the Perspective of Combining GVI and NDVI: A Case Study of Zhongshan City, Guangdong Province

Abstract

Urban street greening has a positive impact on the health of citizens and the urban environment. This study takes the representative streets in the main urban area of Zhongshan City, Guangdong Province as an example to explore urban street greening perception from the perspective of combining Green visual index (GVI) and Normalized difference vegetation index (NDVI). This study uses a deep learning based image semantic segmentation method to analyze Baidu Street View to calculate the GVI of the street, and uses GF-1 satellite data to calculate NDVI to compare and analyze the characteristics and correlation of GVI and NDVI of urban streets. The results show that: 1. The GVI of streets in the central urban area of Zhongshan varies from 8.06% to 36.00%, with Xingzhong Road in Shiqi District Street having the highest GVI; 2. The mean value of NDVI of each street shows different changes with the increase of buffer scale, and the mean value of NDVI has a strong scale sensitivity; 3. The highest Pearson correlation coefficient between GVI and 25m DNVI mean value was 0.862; 4. The GVI prediction model based on NDVI is: y=0.8249x+0.0181, R2=0.7433. On this basis, the shortcomings of street landscape are analyzed and optimization suggestions are given, providing reference for urban street landscape evaluation, spatial optimization, and landscape improvement.