Deep Learning Approach for Building Detection in Satellite Multispectral Imagery

Building detection from satellite multispectral imagery data is being a fundamental but a challenging problem mainly because it requires correct recovery of building footprints from high-resolution images. In this work, we propose a deep learning approach for building detection by applying numerous enhancements throughout the process. Initial dataset is preprocessed by 2-sigma percentile normalization. Then data preparation includes ensemble modelling where 3 models were created while incorporating OpenStreetMap data. Binary Distance Transformation (BDT) is used for improving data labeling process and the U-Net (Convolutional Networks for Biomedical Image Segmentation) is modified by adding batch normalization wrappers. Afterwards, it is explained how each component of our approach is correlated with the final detection accuracy. Finally, we compare our results with winning solutions of SpaceNet 2 competition for real satellite multispectral images of Vegas, Paris, Shanghai and Khartoum, demonstrating the importance of our solution for achieving higher building detection accuracy.

[1]  C. Heltin Genitha,et al.  Super resolution mapping of satellite images using Hopfield neural networks , 2010, Recent Advances in Space Technology Services and Climate Change 2010 (RSTS & CC-2010).

[2]  Sunitha Abburu,et al.  Satellite Image Classification Methods and Techniques: A Review , 2015 .

[3]  Yang Wang,et al.  MARTA GANs: Unsupervised Representation Learning for Remote Sensing Image Classification , 2016, IEEE Geoscience and Remote Sensing Letters.

[4]  Trevor Darrell,et al.  Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation , 2013, 2014 IEEE Conference on Computer Vision and Pattern Recognition.

[5]  Yoshihiko Mochizuki,et al.  Detection by classification of buildings in multispectral satellite imagery , 2016, 2016 23rd International Conference on Pattern Recognition (ICPR).

[6]  Bertrand Le Saux,et al.  Joint Learning from Earth Observation and OpenStreetMap Data to Get Faster Better Semantic Maps , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).

[7]  Antoine Manzanera,et al.  Binary Distance Transform to Improve Feature Extraction , 2016, ArXiv.

[8]  Liu Pin,et al.  Super-resolution Reconstruction of Satellite Video Images Based on Interpolation Method , 2017 .

[9]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[10]  Vladislav Ostankovich,et al.  Illegal Buildings Detection from Satellite Images using GoogLeNet and Cadastral Map , 2018, 2018 International Conference on Intelligent Systems (IS).

[11]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[12]  C. Tappert,et al.  A Survey of Binary Similarity and Distance Measures , 2010 .

[13]  Michael S. Bernstein,et al.  ImageNet Large Scale Visual Recognition Challenge , 2014, International Journal of Computer Vision.

[14]  Fadi Dornaika,et al.  Building detection from orthophotos using a machine learning approach: An empirical study on image segmentation and descriptors , 2016, Expert Syst. Appl..

[15]  DaoYu Lin,et al.  Deep Unsupervised Representation Learning for Remote Sensing Images. , 2016 .

[16]  Trevor Darrell,et al.  Fully Convolutional Networks for Semantic Segmentation , 2017, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[17]  Xin Huang,et al.  A multi-index learning approach for classification of high-resolution remotely sensed images over urban areas , 2014 .

[18]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[19]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[20]  Luc Van Gool,et al.  The Pascal Visual Object Classes (VOC) Challenge , 2010, International Journal of Computer Vision.