Urban Expansion Classification using the Change Detection of High-Resolution Images, for Jeddah Province

Abstract

The Kingdom of Saudi Arabia is heading to catch up with civilization, to be one of the major international country in the first world. One of the important things is to identify the urban changes taking place in the city growth. Also, high resolution satellite images of Jeddah are available from various sources such as GeoEye (2010) and WorldView-3 (2015). Jeddah has different satellite images from different sources with different accuracy and therefore the spectral, spatial resolution and coverage area are different. So, the aim of this research work is to assess, compare, and describe the some ways of dealing with image of the GeoEye (2010) and image of the WorldView-3 (2015). To reach the desired uses of these methodologies and to achieve a degree of accuracy appropriate to the required results, formulating this comparison within the boundaries of the Jeddah Governorate to define the scope of the study and dealing with different geographical factors that characterize the province of Jeddah. In this context, in the current research work, the sample representing the urban area was chosen from 2010 and 2015. The quality control process was done on the product using different methods to detect automatic changes of the Study areas, like (area - geometric shape - number of changes). For Geometric shape of the image in 2015, it is noted that the best result was the ratio ranges between 50-60% in the following methods (IHS to RGB - Image Segmentation - Unsupervised (ISODATA) - Supervised (min)). As for the rest of the results, it ranges between 30% - 40%. As for the number of Vertex of the geometric shape of the image 2015, it is noted that the lowest number of points that make up the buildings is from the Image Segmentation method and the largest number of points that make up the buildings is from the Unsupervised (K-means) method. In the area of the building in 2015, is noted that the best result, where the ratio ranges between 65%-70% in the following methods (IHS to RGB - Image Segmentation - Unsupervised (ISODATA) and Supervised (Maximum)). On the contrary, the lowest result was the ratio reaches 59% in Supervised (Minimum) method. For the number of buildings in 2015, is noted that all the results of the supervised classification and the unsupervised classification are close to what was achieved by digitizing.