In recent years, Deep Learning (DL) is proving very successful set of tools for several image analysis, segmentation, and classification tasks. In this paper an automated Deep Learning Architecture (DLA) called the Deep Belief Neural Networks (DBN) stacked by Restricted Boltzmann Machines (RBMs), is designed, implemented, and experimentally evaluated for extracting semantic maps of roads in Remote Sensing (RS) images. Representative features are extracted by unsupervised pre-training of DBN and supervised fine-tuning phase. A Logistic Regression (LR) is added to the end of feature learning system to constitute a DBN-LR architecture. This LR classifier is employed to fine-tune the whole pre-trained network in a supervised way and classifies the patches from RS images. The features extracted from the image patches are fed to the architecture as input and it produces the class labels as a probability matrix as either a positive sample (road) or a negative sample (non-road). A math morphology algorithm is used to improve DBN performance during post processing. Experiments are conducted on a dataset of 970 RS scene images of urban and suburban areas to demonstrate the performance of the proposed network architecture. The proposed deep model resulted in an Overall Accuracy (OA) of 96.57% and F1-score of 0.9552. The results of the proposed architectures are compared with those of other network architectures. Experimental results demonstrate the effective performance of the proposed method for extracting roads from a complex scene.

Md. Abdul Alim Sheikh1, Tanmoy Maity2, Alok Kole3
Aliah University, India1, Indian Institute of Technology, Dhanbad, India2, RCC Institute of Information Technology, India3

Remote Sensing Imagery, Road Networks Extraction, Deep Learning, Deep Belief Network, Restricted Boltzmann Machine
Published By :
Published In :
ICTACT Journal on Soft Computing
( Volume: 13 , Issue: 2 , Pages: 2879 - 2889 )
Date of Publication :
January 2023

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.