Font Size: a A A

Application And Research Of Natural Interaction Technology In Virtual Simulation

Posted on:2022-01-29Degree:MasterType:Thesis
Country:ChinaCandidate:L ZhuFull Text:PDF
GTID:2518306485980679Subject:Control Engineering
Abstract/Summary:PDF Full Text Request
The natural human-computer interaction method is simpler and more efficient than the traditional keyboard and mouse interaction method.Excellent interactive application scenarios should have the advantages of accurate human expression,accurate machine recognition,and clear scene display.The application under the virtual simulation environment has the advantages of low cost,high repeatability,safe engineering operation,transparent engineering structure,etc.,and is suitable for research on various complex machinery and human body observation.The traditional keyboard and mouse cannot meet the requirements of these complex interactive environments.Therefore,the application of natural interaction technology in the virtual simulation environment is more in line with the requirements of the new environment.This paper takes image information processing technology as the core,based on the cross integration of interactive technology,virtual simulation,deep learning and other technologies,analyzes and compares common natural interaction technologies,realizes gesture recognition and application in virtual simulation,and designed the natural interaction of 3D virtual simulation scene based on depth information.Specifically,computer vision-based gesture recognition,human posture,depth information,and other interactive methods are hotspots of natural human-computer interaction.This paper establishes an evaluation model for the integration of natural interaction and virtual simulation based on scene design,interaction constraints,and introduction of recognition technology;through the completion of data collection,data processing,data set production,network construction and other steps,a comprehensive comparison experiment with Visual Gesture Builder(VGB),Deep Convolutional Encoder-Decoder Network(Deep Convolutional Encoder-Decoder Network,DCEDN)gesture recognition technology and put forward a method to increase the naturalness of the interaction mode in virtual simulation scenes;Perform depth information acquisition based on Time-OF-Flight,improved "Zhang’s calibration method" and calibration of depth images and color images,and established a weighted average image filter to complete image denoising and other aspects,realizing natural interaction based on depth information in 3D virtual simulation scenes.To strengthen the fusion application effect of human body posture and depth information in virtual simulation scenes.Through specific experimental research,this paper effectively demonstrates the feasibility of fusion application of natural interaction technology such as 3D depth human body information and virtual simulation scene,and the effect is fantastic.
Keywords/Search Tags:natural interaction technology, virtual simulation, gesture recognition, Kinect, depth information
PDF Full Text Request
Related items