基于图像语义分割的关中“文商旅街区”30年景观视觉变迁

30-year Visual Changes of Landscape in Guanzhong Cultural and Commercial Tourist Blocks Based on Image Semantic Segmentation

1990年以来,大批传统的历史文化街区向新时代“文商旅街区”转型。文章从“景观视觉吸引”原理出发,采用全卷积神经网络技术,以西安三学街、韩城古城街区、三原县东大街为研究区域,对其街道景观变迁(1990—2020年)进行调查研究。结果表明:30年的发展过程中,关中“文商旅街区”景观的“服务性”意识逐渐增强,其中,“建筑/构筑物、道路系统”的景观视觉面积占比持续减少,“广告与装饰物、销售物品、人群分布、基础设施”等景观视觉面积持续增长,“植物”的景观视觉面积呈现由少到多的波动变化;植物景观的密集程度与人群分布特征具有显著相关性。基于研究结果,提出管控“商业”类型景观视觉刺激物、减少“交通车辆”的出现比例、增加“植物”景观视觉刺激物类型等优化建议,以期为“文商旅街区”景观的可持续发展提供参考。

Since 1990,a large number of historically and culturally traditional blocks have been transformed into“Cultural and Commercial Tourist Blocks”.In order to understand the development of these Cultural and Commercial Tourist Blocks over these years,the paper starts from the principle of“visual attraction of landscape”,and takes Sanxue Street,blocks within Hancheng Ancient City and East Street in Sanyuan County to investigate and study the changes in street landscape from 1990 to 2020 by using the technology of fully convolutional network.The results show that during the 30 years development,the sense of“service”in Guanzhong“Cultural and Commercial Tourist Blocks”landscape has been gradually increasing.Among them,the proportion of the visual area of“buildings/structures and road system”displays a trend of continuous decrease,the proportion of the visual area of“advertisements and decorations,sales items,crowd distribution,infrastructure”shows a continuous increase,and the proportion of“plants”landscape presents a fluctuating growth.At the same time,the density of plant landscape has a significant correlation with the distribution characteristics of crowds.Based on the results,it is proposed to control the increase of“commercial”landscape visual stimuli,reduce the proportion of“traffic vehicles”,and increase the“plant”landscape visual stimuli.The study is aimed at providing references for the sustainable development of cultural,business and tourism block landscape.