RS
Alireza Homayunfar; Khalil Valizadeh Kamran; Mostafa Mousapour
Abstract
The purpose of this research is to compare the efficiency of Pixel-Based kernel functions on the support vector machine algorithm and Object-Based fuzzy operators in the extraction of built-up land in Hamedan. For this purpose, the Sentinel 2 satellite multi-spectral image with a spatial resolution of ...
Read More
The purpose of this research is to compare the efficiency of Pixel-Based kernel functions on the support vector machine algorithm and Object-Based fuzzy operators in the extraction of built-up land in Hamedan. For this purpose, the Sentinel 2 satellite multi-spectral image with a spatial resolution of 10 meters has been used. ENVI software was used for image preprocessing and Pixel-Based classification, and eCognation software was used for Object-Based classification. In the processing stage, first, in the ENVI software environment, training data and ground truth points were determined, and then using support vector machine kernel functions, including linear, polynomial, radial, and sigmoid basis functions, the class process The pixel-based classification was done and then the classification accuracy of the pixel-based method was evaluated. In the environment of eCognation software, segmentation operation was done with a certain scale, shape factor, and compression factor, and then using object-oriented fuzzy operators including AND, OR, MGE, MAR, MGWE and ALP, the classification process was performed. The object-Based fuzzy classification was also performed and the accuracy of each of the maps produced by the Object-Based method was also calculated. Producing a map of built-up urban lands with better accuracy using satellite images justifies the innovative aspect of this research. In this research, the AND fuzzy operator had the highest amount of accuracy in the produced maps, which indicates that by using Object-Based processing of satellite images, more accuracy can be achieved in the production of urban built-up lands.
