Title :
Shadow detection and removal by object-wise segmentation
Author :
K A Divya;K I Roshna;Shelmy Mathai
Author_Institution :
Computer Science, FISAT, Ernakulam, Kerala, India
Abstract :
Traditional pixel level shadow detection methods cause loss of information in high resolution images. Here present an object wise methodology which can automatically detect and remove shadows from satellite images. In this method using image parameters, image segmentation is done. For seperating shadow region threshold values are used, thereby shadows are detected. Based on grayscale values Some dark objects which are mistakenly classified as shadows are ruled out and then Image featurs are taken by support vector machine for effective classification of data. Using morphological operation inner outer outline profile line (IOOPL) are created for shadow removal. Relative Radiometric Correction(RRN) is performed over each object using IOOPL sections. The application shows that the new method can effectively detect shadows from urban high-resolution remote sensing images and can accurately restore shadows.
Keywords :
"Image segmentation","Remote sensing","Image resolution","Gray-scale","Vegetation mapping","Satellites","Feature extraction"
Conference_Titel :
Computational Intelligence and Computing Research (ICCIC), 2015 IEEE International Conference on
Print_ISBN :
978-1-4799-7848-9
DOI :
10.1109/ICCIC.2015.7435784