Optical and Radar Image Fusion for Urban Landuse Mapping

نوع: Type: thesis

مقطع: Segment: masters

عنوان: Title: Optical and Radar Image Fusion for Urban Landuse Mapping

ارائه دهنده: Provider: mina peyrovedin

اساتید راهنما: Supervisors: Dr. Morteza heidari Mozzafar

اساتید مشاور: Advisory Professors:

اساتید ممتحن یا داور: Examining professors or referees: Dr. Hassan Khotanlou - Dr.hossein Torabzadeh khorasani

زمان و تاریخ ارائه: Time and date of presentation: 2023-01-25 - 14:00

مکان ارائه: Place of presentation: Seminar2

چکیده: Abstract: Knowing about the types of land surfaces and human activities as a basis for different planning is one of the most important special information. Maps that represent such activities are called land use maps. In other words, these maps show the current conditions and geographical distribution of land use in activities such as agriculture, urban management, forestry, etc. They also represent the cover that covers the surface of the earth. One of the most important information needed by urban management is the use of urban and regional land use maps. Land use maps are different both in terms of the number of tax layers and in terms of details. Preparing such maps using traditional methods and interpretation of aerial photographs will require spending a lot of time and money. With remote sensing satellite images, some layers of the land use map can be prepared in general. These layers can be vegetation, soil, construction effects, etc. Remote sensing data provide the possibility of wide visibility with low cost and time. The use of different spectrums of electromagnetic waves have a great potential for preparing up-to-date urban land use maps. Remote sensing can be used as a powerful tool by using data from different sources and fusion them with each other to classify land use types. In this research, the fusion of Sentinel-1 and Sentinel-2 satellite images was used to prepare urban land use maps. The aim of this research was to achieve the appropriate accuracy in land use map classification by fusion optical and radar images. Classification of Sentinel-1 single image, Sentinel-2 single image, fusion of optical and radar image (by Gram _Schmidt (G.S) method along with Normalized Difference Vegetation (NDVI)) and fusion of optical and radar image (by Discrete Wavelet Transform (DWT) method) Done. Also, in this research, different classification algorithms Support Vector Machine (SVM), Neural Network (NN) and Minimum Distance (MD) were used. The results show that the classification accuracy with a single radar image has the lowest accuracy. Also, the accuracy of classification using single optical image was better than using single radar image. However, the accuracy of classification by fusion optical and radar image using DWT method was lower than that of single optical image. This method has a poor performance in revealing border areas among phenomena and complications, especially in urban areas. Finally, the fusion of optical and radar images using the G.S. method resulted in the best accuracy among the different classifications performed in this research. Its overall accuracy and kappa coefficient by SVM classification method were 95.52% and 0.93%, respectively. This research showed that using the fusion of optical and radar images can have a very good effect on the results of land use classification.