Image captured by satellite will be degraded due to scattering of the light by the atmospheric particles under challenging environmental conditions like fog, haze, smoke, etc. Hence this will seriously affect the performance of computer vision system. In this paper an image dehazing based on Quad tree subdivision and convolution neural network(CNN) transmission map is developed to provide end to end dehazing. This algorithm will help to recover the image clearly and accurately. Road extraction plays a significant role in traffic management, city planning road monitoring map updating, GPS navigation, etc. After analyzing various road models and features, this paper also presents an effective method for road extraction based on morphological operation and canny edge detection from the dehazed image. Hence provide a fast, simple and accurate method of dehazing and road extraction.
Dehazing and Road Feature Extraction from Satellite Images
2019-04-01
828315 byte
Conference paper
Electronic Resource
English
Image Dehazing Network Based on Multi-scale Feature Extraction
TIBKAT | 2022
|Image Dehazing Network Based on Multi-scale Feature Extraction
Springer Verlag | 2021
|Automatic archaeological feature extraction from satellite VHR images
Online Contents | 2010
|