Design And Implementation Of Intelligent Home Improvement Sales Management System

With the advancement of science and technology and the improvement of living standards,people have put forward a higher level of customization requirements for the living environment,and Internet home decoration platforms have emerged.In recent years,home decoration online design software has emerged in an endless stream,but the display on the design software interface cannot take into account the immediate response requirements and rendering effect requirements,so the rendering system has become an indispensable part of the online home decoration platform.The indoor home decoration scenes are complex and changeable,and the automation of the rendering process faces many challenges.Under this background,it is extremely necessary to design a complete rendering system.Based on the design software of an Internet home decoration design platform in Beijing,the subject designed and implemented a set of efficient indoor scene rendering system based on computer clusters.This article conducts precise research on the status of the home decoration industry and the development of the Internet home decoration,and deeply analyzes the actual needs of the indoor scene rendering system.At the business processing level,a fully-automatic rendering system was designed in conjunction with functional requirements.The system can automatically generate cenes,adaptively adjust lighting,and automatically set rendering parameters.At the system design level,a complete rendering task scheduling system and task monitoring system are designed for situations where there are many rendering tasks and limited computing resources.Rendering speed is the bottleneck of the efficiency of this system.In order to improve the rendering speed,this paper investigates the literature related to rendering acceleration in recent years,and proposes a deep learning denoising algorithm based on kernel_splating combined with the actual situation of the system.The algorithm analyzes the auxiliary feature buffer of the rendering algorithm to obtain additional input features,uses the deep neural network based on the UNet network to calculate the filter kernel,and uses the kernel_splating strategy for image reconstruction.This paper uses real scenes in the system to conduct comparative experiments to verify the effectiveness of the algorithm.Experiments prove that the denoising effect of the algorithm in most scenes can take into account both the effect and the speed.To deal with the poor performance of the above algorithm in particularly complex lighting environment scenes,a rendering denoising algorithm based on feature decomposition is proposed.The algorithm divides the input features into four parts according to the rules,and denoising each part separately,and then use these four parts to synthesize the final result.Experiments show that the improved algorithm can also obtain better noise reduction results when dealing with scenes with complex lighting conditions.