Urban Expansion Pattern Analysis and Planning Implementation Evaluation Based on using Fully Convolution Neural Network to Extract Land Range

Abstract

In recent years, due to the rapid development of China’s urban, it is significant for effective implementation of urban science development and planning that grasp the process of urban development, analyze the potential of subsequent development, and evaluate the matching degree of the development status and the planning. Thereinto, an effective way we exercise today is to evaluate urban expansion pattern analysis and planning implementation. According to research results of the urban land range extraction method based on the support vector machine (SVM) and fully convolution neural network (FCN) of the depth learning method for the night light image data, this paper describes an integration of remote sensing (RS) and geographic information system (GIS) and analyzes the urban expansion pattern of Beijing based on the computed results of landscape pattern indices. The results unveil that from 1990s to 2010s, Beijing took on a circle expansion mode on the ground the spatial agglomeration degree gradually increases and the expansion potential has spatial distinctions, which basically meets the requirements of the overall planning.