Hierarchical semantic cognition for urban functional zones with VHR satellite images and POI data

Multiscale Geoscene Segmentation for Extracting Urban Functional Zones from VHR Satellite Images

“…Firstly, low-level features, such as spectral, geometrical, and textural image features, are widely used in image analyses [11], but they are weak in characterizing functional zones which are usually composed of diverse objects with variant characteristics [12]. Then, middle-level features, including object semantics [4,8], visual elements [7], and bag-of-visual-word (BOVW) representations [13], are more effective than low-level features in representing functional zones [7], but they ignore spatial and contextual information of objects, leading to inaccurate recognition results. To resolve this issue, Hu et al (2015) extracted high-level features using convolutional neural network (CNN) [10], which could measure contextual information and were more robust than visual features in recognizing functional zones [14,16].…”

“…To resolve this issue, Hu et al (2015) extracted high-level features using convolutional neural network (CNN) [10], which could measure contextual information and were more robust than visual features in recognizing functional zones [14,16]. Zhang et al (2017) had a different opinion on the relevance of deep-learning features and stated that these features rarely had geographic meaning and were weak for the purpose of interpretability [4]. Additionally, the size and shape of the convolution window can influence deep-learning features.…”

“…Additionally, the size and shape of the convolution window can influence deep-learning features. Recently, spatial-pattern features measuring spatial object distributions are proposed to characterize functional zones and produce satisfactory classification results [4]. However, these spatial-pattern features are measured based on roads blocks which cannot necessarily be applied to zone segmentation [36].…”

“…In most cities, they consist of commercial zones, residential districts, industrial zones, shanty towns, and parks ( Figure 1). Essentially, functional zones are spatially aggregated into regular patterns of diverse geographic objects, and their categories are semantically abstracted from these objects' land-use classes [4].…”

“…Finally, functional zones can be labeled with categories based on their features. Previous efforts at functional-zone analysis focus mainly on feature representations [9][10][11][12][13][14] and classification methods [4,15,16], but ignore zone segmentation. This is unfortunate because zone segmentation is an essential precursor to the other two steps of functional-zone analysis and is hence fundamental to the entire undertaking.…”