Application of Digital Terrain Editing Technology Based on Regular Grid Structure in Urban Landscape Design

Micro-terrain classification is an indispensable part of the regular grid digital elevation model and is important in urban landscape design. Given the shortcomings of the regular grid digital elevation model, a micro-terrain classification method integrating a GA-BPNN algorithm is proposed. Firstly, a micro-terrain classification model based on a regular grid DEM network is constructed by combining BPNN with the regular grid. Secondly, to address the issue of the slow convergence rate of BPNN, the GA is proposed as a means of optimization. On this basis, an improved GA-BPNN model is constructed and its efficacy is evaluated. The results demonstrated that the GA-BPNN model commenced convergence after 70 iterations, with a stable loss value of 0.4. The overall classification accuracy reached 95%, with the classification effect of the mountain shoulder being the best at 96.7%, representing a 6.4% improvement over the highest accuracy of the BPNN model. Therefore, the improved model reduces classification errors, improves the accuracy of micro-terrain classification, reduces time costs, and accelerates operational efficiency. It also improves the performance of the regular grid micro-terrain classification model, which is conducive to promoting the intelligence and modernization of urban landscape design.