文章基本信息

标题：Real Time Dehazing System for Automobiles
本地全文：下载
作者：Basil Eldho ; Shruti Manmadhan ; Susan Jacob 等
期刊名称：International Journal of Computer Trends and Technology
电子版ISSN：2231-2803
出版年度：2020
卷号：68
期号：3
页码：99-102
DOI：10.14445/22312803/IJCTT-V68I3P120
出版社：Seventh Sense Research Group
摘要：Cars have been a common mode of transport ever since their innovation. Fog, smoke and heavy rains pose huge hindrances of sight for people when they drive. This has led to many dangerous accidents especially when it comes to driving at high altitudes or narrow roads. Hence, we propose a realtime image dehazing system using machine learning and convolutional neural networking concepts. It captures the path in front of the car as video which is then converted to frames and removes all the factors that reduce the clarity of the image. To do so, the loss per pixel is calculated. Here, training sets are utilized in order to obtain better outcomes. Hazed and dehazed images are analyzed and compared and then converted back to dehazed video. It requires a huge refresh rate to make it real time and finally achieve the output.
关键词：Convolutional neural networks; machine learning; digital image processing; training data set; layers; dehaze.