TY - JOUR AU - Dr Gaurav Dubey, Sukanya, PY - 2020/03/28 Y2 - 2024/03/28 TI - Single Aerial Image based Road Region Segmentation using Deep Learning Algorithm JF - International Journal of Advanced Science and Technology JA - IJAST VL - 29 IS - 5s SE - Articles DO - UR - http://sersc.org/journals/index.php/IJAST/article/view/7163 SP - 315 - 321 AB - Mainly Road Region Detection and Segmentation is an important task for navigation, city planning, future enhancements. Due to noise and haze in aerial images due to pollution results in low accuracy based road detection which needs to be improved. Numerous segmentation techniques have been developed to improve the image segmentation accuracy. In this paper, a method is given to detect road region with high accuracy using deep learning with noise reduction in initial stages for the aerial images. There are three phases for this, one is removing noise using global filters, then training neural network with road region images, and then classifying the road regions for the final segmentation image. The deep learning algorithm for training and segmentation is CNN (Convolutional Neural Network) with Global Filtering technique. The accuracy of road region is increased by use of this technique as give clarity in the aerial images. The paper describes the global filter technique and Convolutional Neural Network based deep learning in detail. ER -