2 minute read

Street View Imagery in Urban Research4

Lin Zhuowei, Wen Jing, Yin Yue, Kong Fanding

Liu Weiran, Xiao Di, Chen Mengdi, Wu Yingxian

Advertisement

For further details please get in touch with:

• Kong Fanding (kong3810@connect.hku.hk)

• Xiao Di (u3589430@connect.hku.hk)

• Lin Zhuowei (lzzzzzz@connect.hku.hk)

• Ghoshan Jaganathamani (gjaa@connect.ust.hk)

• Siddharth Khakhar (khakhar@hku.hk)

2.30 Introduction

The emergence of streetscape data technology has helped to address the previous relatively vague and subjective evaluation of urban spatial quality. Researchers can leverage the widespread application of open data in urban spatial analysis to change the status quo of over-scaled designs and poorly detailed indicators caused by the constraints of basic data conditions. As streetscape images are closer to the individual’s perspective, they can give users a 360° panoramic spatial information of the street and are gradually being applied to the study of the spatial quality of streets from a human perspective.

SVI has already been used to some extent in the field of urban street research. The use of computer technology can help people to process a huge number of street images. Green street coverage, the visual quality of street space, green visibility and even the impact of the natural environment on mental health all benefit from SVI. Google Street view has been used to refine 3D models of cities and to evaluate the perception of street safety, as shown in several studies.

By combining remote sensing and proximity sensing, it can help people to better understand the shape and quality of cities. With the development of computer technology and the introduction of a variety of machine learning algorithms, it has become possible to use deep convolutional neural network architecture to line-of-sight accurate deep bottom processing of street view images and to effectively identify multiple elements such as sky, pavement, lanes, buildings and greenery.

In addition, as the relationship between built environment features is not simple linear, the validity of sociological statistical analysis in built environment research is relatively limited, a few studies have started to try to apply machine learning techniques to research design and data analysis, which can better handle the interrelationship between complex built environment features, such as support vector machine, random forest, etc., and have achieved good results.

The combination of machine learning and streetscape images has changed the previous situation where basic street data was difficult to obtain and streetscape images were difficult to use efficiently. The use of machine learning algorithms can not only provide refined basic data for spatial quality research, but also ensure the refinement and rapid processing of large-scale data, solving the traditional data faced by the large-scale is difficult to refine, while the local refinement of the data is difficult to represent the global situation, making the measurement of the previous “unmeasurable” spatial quality in the technical sense. This makes it technically possible to measure the spatial quality of previously “unmeasurable” data.

SVIs can be applied in urban studies in the following areas

Neighbourhood evaluation:

Street view image processing can be used to obtain spatial elements of large scale urban neighborhoods, thus enabling rapid urban neighbourhood evaluation.

Urban greenery: Collect streetscape photos and calculate the green index for each photo, which can be used to evaluate urban greenery.

Transportation and mobility:

Considering that SVI is captured along streets, transportation and mobility studies are unsurprisingly another major application area. Most use cases in this domain revolve around traffic safety, as SVI provides a convenient source to conduct virtual street audits and extract characteristics of roads.

Visual quality:

The physical visual quality of street space is achieved automatically by combining 3-dimensional composition calculation of greenery, openness, enclosure using machine-learning segmentation method SegNet.

Urban morphology:

SVI is a powerful source to measure the urban form as perceived by a pedestrian in a street canyon. Most of the relevant studies focus on urban climate.

This article is from: