River and Sand Bank Regions Segmentation from Google Earth Imagery

Nowadays, extracting river information accurately from remotely sensed images plays a vital role in policy planning, hydrological planning and environmental planning. So, change detection from remotely sensed data is an essential for environmental monitoring. In change detection process, object segmentation from satellite imagery is a primary important step to obtain accurate classification results. Thus, this paper is proposed to segment river regions and sand bank regions from Google earth images. For these regions segmentation, heuristic rules are applied. In this system, Sobel operator is used to calculate the gradient value for the input grey scale image and then the river regions and sand bank regions are segmented by applying the thresholding approach and morphological approaches such as dilation, closing and extracted connected components. The proposed system obtained the 97.7 % accuracy for river regions segmentation and 72.1 % accuracy for sand bank regions extraction. Experimental results approve the effectiveness of the proposed system. Thus, this system can provide the river detection, change detection and policy making.