Design of Iterative Reconstruction Method of Landscape Environment Based on Deep Belief Network

Abstract

The 3D reconstruction system can reproduce the spatial pattern of urban landscape ecotone and provide an effective design method for urban ecological environment planning and urban spatial layout. Using the technology of combining computer aided design (CAD) with geographic information system (GIS), digitizing the components of landscape architecture can not only realize the virtual reproduction of landscape architecture, but also store and manage the artistic details such as the structure and function of landscape architecture. This article constructs a 3D reconstruction system of landscape environment based on CAD technology driven by artificial intelligence (AI), expounds the structure of deep belief network and the stage of automatically extracting landscape image features, and how to eliminate the data redundancy of original CAD data sources while ensuring the authenticity of the scene, so as to reproduce the accurate, realistic and visual spatial pattern of urban landscape ecotone. The simulation results show that compared with the control scheme, the accuracy of this method is improved by 25.72%, and the feature information of terrain image can be well restored and the background information can be suppressed. CAD 3D structure construction drawing can not only accurately show the shape change law of the structure, but also observe the shapes of various parts of the structure in various directions, and intuitively see the mutual relationship and interference of the components, which is more conducive to accurately guiding the construction.