الفهرس 
AbstractOnly 14 pages are availabe for public view
 |  | from | 108 |  |  |
| | | from | 108 | | |
Abstract
the availability of the new generation high spectral and spatial resolution satellite images has opened a new era for producing and updating accurate maps Urban landscapes have different land cover classes include buildings roads vegetation water and soil the high and accelerating rate of urban changes and extensions in developing countries calls for an efficient technique for updating building areas the main problem lies in the highly heterogeneous nature of the building and the spectral similarities between building and some other land cover classes Moreover the acquired images are subject to the presence of shadows Shadow is one of the main factors that negatively affect the ability to extract information from the image Otherwise shadows can provide additional semantic clues about their casting object the aim of the present study is to develop new techniques to optimize the features extraction process from new generation remote sensing satellite images for the purpose of mapping For achieving this aim WorldView-2 (WV-2) remote sensing satellite images with eight spectral bands were used A spectral reflectance analysis for the main ground features has been studied along the eight spectral bands to determine the most effective bands for shadow and building extraction processes These bands are employed with invariant color models for producing the new proposed shadow and building detection indices Both shadow and building pixels are separated from the remained features using selected automatic thresholding methods A comparative study is applied to the proposed indices with standard indices for validation purpose the developed indices are then integrated into a complete approach to extract borders between buildings and their corresponding shadows the building’s rooftops boundaries are determined using a novel algorithm the performance of the proposed approach has been visually and quantitatively compared with recent approaches for building extraction Results of this work show that Coastal-Blue (CB) Red Edge (RE) and
Near-Infrared 1 (NIR1) bands are the most effective bands for shadow detection the experimental results demonstrate the efficiency and feasibility of the proposed Saturation-Intensity Shadow Detection Index (SISDI) with an overall accuracy of 97.80% In the SISDI index image the shadow can be separated at zero threshold value directly and the index has the ability to detecting small shadow areas In regard to building class Green(G)Yellow(Y)Red(R)and RE bands are the optimal bands for extraction Also it has been seen that the overall accuracy of the proposed Transform Building Index (TBI) and Building Detector Index (BDI) up to 89.31% and 90.52%, respectively The Neighborhood valleyemphasis thresholding method can separate buildings from remaining features on the building indices histograms effectively The two proposed building indices have the ability to reduce misclassification between buildings and spectral similar objects Also the experimental results show the efficiency of the proposed approach in building borders extraction, with an average overall accuracy of 92.20%, and has the ability for detecting building rooftops separated from its façades All thresholding processes are carried out automatically through the workflow of the proposed approach to be completely
automatic.