项目名称: 运动图搜索的DNA计算模型研究
项目编号: No.61300015
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 周东生
作者单位: 大连大学
项目金额: 25万元
中文摘要: 本项目以运动捕捉中的数据重用为研究背景,针对数据合成中大规模运动图的最优路径搜索问题,结合具有高度并行性特质的DNA计算,探寻新的高效计算模型。项目首先研究运动数据的特征空间提取与表征方法,基于降维、聚类等算法思想完成运动捕捉数据的关键帧提取和关键运动片段切割,并在此基础上构建大规模运动图,将动作合成问题转化为图的最优路径搜索问题。然后综合考虑个性化、结构化、语义化以及运动学性能等约束特性,设计可有效将运动图转换为DNA编码的规则与方法。进而设计基于DNA表面计算的运动图搜索生化操作算法,并在传统电子计算机上完成运动图搜索的DNA计算模型仿真。本项目将DNA计算引入到人体运动数据合成重用领域,有助于拓展DNA计算的应用范围,推动基于DNA计算的数据处理理论方法的进展。
中文关键词: 运动捕捉;数据重用;运动合成;运动图;DNA计算
英文摘要: This project takes motion capture reuse as research background, attempts to deal with the problem of searching optimal path in a large-scale motion graph, and aims to structure a new effective computing model by employing DNA computing. First, the project intends to study on the method of feature extraction from the original motion data, and plans to extract the key frames and sequence fragments to construct motion graph, which based on the methods dimension reduction and clustering. It is intends to covert motion synthesis into combinatorial optimization problem. Then, based on the different constraints of personalization, structure, semantic and kinematics of the motion capture data, the project will study on the DNA code model for expressing the motion graph, and focus on the establishment of DNA surface computing model. The project also plans to design a series of biology and chemical experiments to examine the DNA computing. Compared to the traditional ideas of 3D human motion data resue, the methods with DNA computing have the potential ability to meet the real-time mass data processing, and could help to promote the progress of the data processing theory in DNA computing.
英文关键词: Motion Capture;Data Reuse;Motion Synthesis;Motion Graph;DNA Computing