城市园林景观设计过程模块化信息融合模型  

Modular Information Fusion Model of Urban Landscape Design Process

作　　者：冯瑞芳[1] 易晓园[1] FENG Rui-fang;YI Xiao-yuan(Jincheng College,Sichuan University,Chengdu Sichuan 611731,China)

机构地区：[1]四川大学锦城学院,四川成都611731

出　　处：《计算机仿真》2021年第5期163-167,共5页Computer Simulation

摘　　要：因传统的信息融合模型存在城市园林景观设计效率低的问题,基于遗传神经网络设计了新的城市园林景观设计过程模块化信息融合模型。在预处理传感器图像的基础上,利用有序数值序列创建数字高程模型。然后通过人工视差辅助机制获得立体正射影像对,结合蚁群算法生成三维园林景观。采用过程维和语境维描述园林设计规模和资源,搭建主体鲜明的模块化园林景观;最后基于实数编码的遗传算法优化神经网络初始权值,并挑选适当学习因子训练神经网络,从而完成信息融合任务。仿真结果表明,上述模型融合过程稳定性较好,信息融合能耗较低,可促进园林景观的高效率建设。Generally, the traditional information fusion model has the disadvantage of low efficiency in the design of the urban landscape. In this regard, this paper designed a new modular information fusion model of the urban landscape design process based on a genetic neural network. Based on the preprocessed sensor images, ordered numerical sequences were applied to create a digital elevation model. Stereo orthophoto pairs were obtained by using the artificial parallax assisted mechanism. A three-dimensional landscape was generated via the combination of the ant colony algorithm. Process dimension and context dimension were adopted to describe the scale and resources of landscape design, and the modular landscape architecture with the distinct subject was built. The initial weights of the neural network were optimized by a real coded genetic algorithm, and the appropriate learning factors were selected to train the neural network, completing the information fusion task. The simulation results show that the model fusion process has good stability and low energy consumption of information fusion, promoting the efficient construction of landscape architecture.

关 键 词：城市园林景观 信息融合 数字高程模型 三维景观 神经网络 

分 类 号：TP385[自动化与计算机技术—计算机系统结构]