Capturing what human eyes perceive: A visual hierarchy generation approach to emulating saliency-based visual attention for grid-like urban street networks

https://doi.org/10.1016/j.compenvurbsys.2019.101454Get rights and content

Highlights

  • Hierarchies of grid-like streets is prone to cause visual cognitive confusion.

  • Using saliency-based visual attention greatly improves visual discrimination.

  • The proposed indicator help forming scaling of the city image in human minds.

  • The head/tail breaks scheme is advised to fit visual hierarchies of grid-like streets.

Abstract

Visual hierarchy is an important notion in urban imagery research. As the skeletons of cities, urban streets attract more attention from urban residents and street network hierarchies are important references for urban planning and urban studies. However, due to the characteristic of over-regularization, it is often difficult for humans to differentiate visual salience for grid-like street networks, resulting in the hierarchies of grid-like streets yielded by existing methods being prone to cause visual cognitive confusion. Therefore, in this study, we proposed a novel model to quantify the extent to which a street attracts human visual attention through emulating the visual attention mechanism that can capture the focus of relatively significant elements at different levels of perception. Using the natural street (also known as the stroke) as the sensor unit, the comprehensive visual salience (CVS) index combining the geometric competitive factors of natural streets at the local scale and psychological competitive factors of natural streets at the global scale is designed. Finally, the visual salience of the urban natural streets is ranked by these CVS scores and the visual hierarchy is derived by the head/tail breaks scheme. The model was applied to eight typical grid-like street networks and the results show that the performance of visual discrimination on street hierarchies is greatly improved. Our hierarchy generation method could effectively detect visually prominent streets for grid-like street networks and generate the visual hierarchies of grid-like street networks that conform to the hierarchies perceived by human eyes. These results would provide helpful suggestions in practical urban street network applications.

Introduction

Visual hierarchy is an important notion in many areas of urban research that can greatly help understand people's cognition of urban morphologies (Jiang, 2012a; Lynch, 1960). As the skeletons of cities, urban streets attract considerable attention from urban residents and are regarded by Lynch (1960) as one of the five main city elements. Urban streets tend to have a hierarchical structure in the human mind due to their enormous imbalance (Brassel & Weibel, 1988; Jiang, 2008; Jiang, 2012b; Jiang, Liu, & Jia, 2013; Li, Fan, Luan, Yang, & Liu, 2014; Weibel & Jones, 1998). From the perspective of visual perception, those streets in a city that are distinctive and easy to attract people's visual attention are recognized first by the human mind (Tomko, Winter, & Claramunt, 2008). These prominent streets could be extracted at different levels to form a visual hierarchy, which represents the arrangement of streets in a way that implies visual importance or visual salience and influences the order in which the human eye perceives what it sees. Accordingly, the visual hierarchy of streets is an essential component for understanding and perceiving a city and could further shed light on human wayfinding and urban morphology (Car, Taylor, & Brunsdon, 2001; Huang, Schmidt, & Gartner, 2012).

Essentially, visual hierarchy is not only an external structural representation of the street network but also an intrinsic requirement of its structural organization, which is closely related to the characteristics of the street pattern. The patterns of urban streets are the result of many interacting phenomena over time, such as geography, natural setting, socioeconomic transformation, and so on (Feng & Sun, 2013). In this process, several typical street patterns are gradually formed: grid-like (planned) street networks, irregular (self-evolved) street networks, and hybrid street networks (planned and self-evolved structures coexist) (Serge, 2012). In the context of the sustainable development of transportation infrastructure, and driven by the dual requirements of street network formation and densification, the increasing grid-like characteristics of modern urban street networks are significant. Evidently, the grid-like street network becomes an indispensable part of a modern street network and will gradually be developed into the main style of future urban traffic environment (Fawaz & Newell, 1976; Hu, Lu, Wang, & Ye, 2015; Miyagawa, 2009; Tanner, 1966). Formally, a grid-like street network consists of a large number of straight streets intersecting at right angles, showing extremely regular characteristics (Serge, 2012).

However, due to the characteristic of over-regularization, it is difficult to differentiate visual salience of different streets on grid-like street networks, resulting in the cognitive confusion of the visual hierarchies in grid-like street networks. Existing methods usually evaluate the importance of streets by functional indices or visual indices (mainly length), also known as salience indices (Chaudhry & Mackaness, 2005; Huang, Zhu, Ye, Guo, & Wang, 2016). These salience indices are usually self-salience indicators that focus on the unique characteristics of streets themselves and neglect the visual competition relationship. They, therefore, lead to the low levels of visual discrimination on grid-like streets, and further cause the visual hierarchies of these grid-like streets being poorly captured and inconsistent with that perceived by human eyes. Taking two classical cities (London and San Francisco) as examples, as shown in Fig. 1(a) and (b), the hierarchies generated from these two cities' street networks are visualized using a spectral color legend with smooth transition from red to blue, where red and blue respectively represent the highest importance (Level 1) and lowest importance (Level 5) and using length as the evaluation indicator. Compared to the clear and vein-like hierarchical structure captured from the London street network (an irregular street network), the hierarchical structure of San Francisco (a grid-like street network) is not clearly captured. Owing to the special characteristics of grid-like street patterns, the limitations of the current salience indices when measuring the contribution of visual significance on grid-like street networks need to be addressed.

In this study, the human visual attention mechanism that can capture the focus of relatively significant elements at different levels of perception is brought in. A new salience measure model is proposed in this paper under the relative salience framework and by emulating the visual attention mechanism to increase visual discrimination of grid-like streets. The designed comprehensive visual salience (CVS) index combines the geometric competitive factors of natural streets at the local scale and psychological competitive factors of natural streets at the global scale to quantify the extent to which a street attracts human visual attention. The experimental results show that the performance of visual discrimination on the street hierarchy is greatly improved. Based on the CVS index, our hierarchy generation method could effectively detect visually prominent streets for grid-like street networks and generate the visual hierarchy of grid-like street networks that conform to the hierarchies perceived by human eyes.

The remainder of this paper is organized as follows. Section 2 provides a short overview of the hierarchy generation schemes. Section 3 elaborates on the hierarchy generation approach that is based on the human visual attention mechanism. Section 4 describes the study area and presents a comparative study between the proposed approach and two traditional approaches. Finally, conclusions are provided in Section 5.

Section snippets

Related work

The hierarchies of street networks have been widely studied in the fields of transportation, urban planning and geographic information science (Benz & Weibel, 2014; Carvalho & Penn, 2004; Huang et al., 2016; Jiang, 2008; Li & Dong, 2010). Hierarchy generation schemes in previous studies typically consist of two main components: basic structures and salience indices.

First, the Gestalt psychological principle which is the foundation of visual hierarchy is commonly employed to form the basic

Methodological framework

We present the basic framework of a hierarchy generation approach based on a new salience measure model which emulates the human visual attention mechanism to increase visual discrimination among street segments. The framework consists of three main elements as shown in Fig. 2:

(1) Generation of natural streets: The basic sensor unit first needs to be defined to quantify its features and generate visual hierarchy. In our study, an urban natural street is regarded as the basic sensor unit, which

Study area and data

The street network data sets of six cities were chosen in this study to evaluate the advantages of the proposed approach when compared to the conventional combination method. These cities are Manhattan in New York, Salt Lake City in Utah, San Francisco in California, Chicago in Illinois, Buffalo in New York and Santiago in Chile. All these street networks have typical grid-like characteristics and also are available in OpenStreetMap (OSM) databases (www.openstreetmap.org). In addition, to

Conclusion

In this paper, the proposed model simulates the human visual attention mechanism to evaluate the extent that a street attracts human visual attention. A novel visual salience indicator called comprehensive visual salience (CVS) is proposed and used as a basis for stratifying the hierarchy of urban street networks perceived by human eyes. For the calculation of the visual saliency indicator, the geometrical competitive factors at the local scale (the relative difference between each natural

Acknowledgements

This research was supported by the State Key Laboratory of Geo-information Engineering (SKLGIE2017-M-4-1) and the Open Fund of Key Laboratory of Urban Land Resources Monitoring and Simulation, Ministry of Land and Resources (KF-2018-03-038).

References (56)

  • T. Barkowsky et al.

    Cognitive requirements on making and interpreting maps

  • S.A. Benz et al.

    Road network selection for medium scales using an extended stroke-mesh combination algorithm

    Cartography and Geographic Information Science

    (2014)
  • K. Brassel et al.

    A review and conceptual framework of automated map generalization

    International Journal of Geographical Information Systems

    (1988)
  • O. Chaudhry et al.

    Rural and urban road network generalisation deriving 1: 250,000 from OS MasterMap

  • W. Dong et al.

    Eye tracking to explore the potential of enhanced imagery basemaps in web mapping

    The Cartographic Journal

    (2014)
  • W. Dong et al.

    Assessing the effectiveness and efficiency of map colour for colour impairments using an eye-tracking approach

    The Cartographic Journal

    (2016)
  • C.M. Feng et al.

    Developing urban roads and managing motorization

  • T.J. Hawbaker et al.

    Road density and landscape pattern in relation to housing density, and ownership, land cover, and soils

    Landscape Ecology

    (2005)
  • H. He et al.

    Road network selection based on road hierarchical structure control

    Acta Geodaetica Et Cartographica Sinica

    (2015)
  • B. Hillier et al.

    The social logic of space

    (1984)
  • X. Hu et al.

    Traffic signal synchronization in the saturated high-density grid road network

    Computational Intelligence and Neuroscience

    (2015)
  • H. Huang et al.

    Spatial knowledge acquisition in the context of GPS-based pedestrian navigation

  • L. Huang et al.

    Characterizing street hierarchies through network analysis and large-scale taxi traffic flow: A case study of Wuhan, China

    Environment and Planning. B, Planning & Design

    (2016)
  • L. Itti et al.

    Computational modelling of visual attention

    Nature Reviews Neuroscience

    (2001)
  • L. Itti et al.

    A model of saliency-based visual attention for rapid scene analysis

  • I. Jackson

    Gestalt-a learning theory for graphic design education

    International Journal of Art & Design Education

    (2010)
  • W. James

    The principles of psychology

    (1950)
  • B. Jiang

    Street hierarchies: A minority of streets account for a majority of traffic flow

    International Journal of Geographical Information Science

    (2008)
  • Cited by (0)

    View full text