Deep Learning Segmentation and Classification for Urban Village Using a Worldview Satellite Image Based on U-Net

Abstract

Unplanned urban settlements exist worldwide. The geospatial information of these areas is critical for urban management and reconstruction planning but usually unavailable. Automatically characterizing individual buildings in the unplanned urban village using remote sensing imagery is very challenging due to complex landscapes and high-density settlements. The newly emerging deep learning method provides the potential to characterize individual buildings in a complex urban village. This study proposed an urban village mapping paradigm based on U-net deep learning architecture. The study area is located in Guangzhou City, China. The Worldview satellite image with eight pan-sharpened bands at a 0.5-m spatial resolution and building boundary vector file were used as research purposes. There are ten sites of the urban villages included in this scene of the Worldview image. The deep neural network model was trained and tested based on the selected six and four sites of the urban village, respectively. Models for building segmentation and classification were both trained and tested. The results indicated that the U-net model reached overall accuracy over 86% for building segmentation and over 83% for the classification. The F1-score ranged from 0.9 to 0.98 for the segmentation, and from 0.63 to 0.88 for the classification. The Interaction over Union reached over 90% for the segmentation and 86% for the classification. The superiority of the deep learning method has been demonstrated through comparison with Random Forest and object-based image analysis. This study fully showed the feasibility, efficiency, and potential of the deep learning in delineating individual buildings in the high-density urban village. More importantly, this study implied that through deep learning methods, mapping unplanned urban settlements could further characterize individual buildings with considerable accuracy.

Deep Learning-Based Semantic Segmentation of Urban Features in Satellite Images: A Review and Meta-Analysis

“…Finally, a softmax function generates the segmentation results. The model soon became widely used for medical image segmentation and also some studies can be found in urban images segmentation [100][101][102]. Yi et al [103] proposed DeepResUnet for effective urban building segmentation at pixel-scale from VHR imagery.…”

“…[106,121,140,141]. Some classified buildings to their utilities [102,113], smaller features inside roads [142] and slum area [135]. Besides the segmentation of features, the studies use different approaches to improve semantic segmentation of urban features, which are shown in Table 1.…”

“…The most commonly used datasets are ISPRS Vaihingen 2D Semantic Labeling dataset [166] (38 papers) and ISPRS Potsdam 2D Semantic Labeling dataset [167] (27 papers) and IEEE GRSS Data Fusion Contest [168] (5 papers) collected their images using UAV. Other than UAV-based images, satellite images were obtained from RADARSAT-2 [136], Worldview-2 [102,134,156], Worldview-3 [149], Landsat-8 [96], SPOT [137,155], Gaofen [138,169], Quickbird, Sentinel-1 and 2 [147], Sentinel-2 and TerraSAR-X [135], Plan-etScope (Dove constellations) [164]. A particular research collected data from a plane [140].…”

“…Pre-processing methods change the characteristics of images on pixel or spectral level. Some of these methods include image processing methods, image normalization by mean value subtraction on images [31,96,158,161], k-means clustering [134], relative radiometric calibration strategy [79], super-pixel segmentation [93], Lidar to digital surface model (DSM) [163], satellite image correction and pan-sharpening [102,164,169], normalized DSM (nDSM) and calculation of vegetation index such as normalized difference vegetation index (NDVI). Twenty-two papers performed such pre-processing methods.…”

Change Detection in Unlabeled Optical Remote Sensing Data Using Siamese CNN

“…A good review of these traditional techniques can be found in [4], [5], [1]. Although these methods use the contextual spatial and spectral information of remote sensing images, the traditional manual design of image features is a tedious and complex task and it is mostly performed on the basis of domainspecific knowledge [6], [7], [8], [9]. The explosive growth of available remote sensing data and the increased processing power afforded by graphical processing units (GPUs) has led to the rise of advanced techniques based on deep learning (DL) [10], [11].…”