中山大学学报(自然科学版)
中山大學學報(自然科學版)
중산대학학보(자연과학판)
ACTA SCIENTIARUM NATURALIUM UNIVERSITATIS SUNYATSENI
2014年
4期
62-68
,共7页
周燕%曾凡智%卢炎生%周月霞
週燕%曾凡智%盧炎生%週月霞
주연%증범지%로염생%주월하
压缩感知%三维模型组件%三视图%组件检索%制造领域
壓縮感知%三維模型組件%三視圖%組件檢索%製造領域
압축감지%삼유모형조건%삼시도%조건검색%제조영역
compressed sensing%three-dimensional model%three-view%component retrieval%manufac-turing field
针对在工业设计海量组件库中快速检索特定三维模型组件的问题,文中结合压缩感知理论给出了一种基于行列与中心点的压缩感知测量过程,由此提出了一种面向制造领域基于三视图的模型组件快速检索方法。利用压缩感知方法提取各组件的三视图的颜色、纹理等测量特征并存入模型库;提取待检索组件的三视图压缩测量特征并与模型库中组件三视图的特征进行相似度计算;对组件三视图间的相似性与中心点的压缩测量特征进行融合,得出模型组件间的整体相似度并输出匹配检索结果。仿真实验表明,算法具有较好的检索性能,在查全率和查准率方面具有一定的优势。
針對在工業設計海量組件庫中快速檢索特定三維模型組件的問題,文中結閤壓縮感知理論給齣瞭一種基于行列與中心點的壓縮感知測量過程,由此提齣瞭一種麵嚮製造領域基于三視圖的模型組件快速檢索方法。利用壓縮感知方法提取各組件的三視圖的顏色、紋理等測量特徵併存入模型庫;提取待檢索組件的三視圖壓縮測量特徵併與模型庫中組件三視圖的特徵進行相似度計算;對組件三視圖間的相似性與中心點的壓縮測量特徵進行融閤,得齣模型組件間的整體相似度併輸齣匹配檢索結果。倣真實驗錶明,算法具有較好的檢索性能,在查全率和查準率方麵具有一定的優勢。
침대재공업설계해량조건고중쾌속검색특정삼유모형조건적문제,문중결합압축감지이론급출료일충기우행렬여중심점적압축감지측량과정,유차제출료일충면향제조영역기우삼시도적모형조건쾌속검색방법。이용압축감지방법제취각조건적삼시도적안색、문리등측량특정병존입모형고;제취대검색조건적삼시도압축측량특정병여모형고중조건삼시도적특정진행상사도계산;대조건삼시도간적상사성여중심점적압축측량특정진행융합,득출모형조건간적정체상사도병수출필배검색결과。방진실험표명,산법구유교호적검색성능,재사전솔화사준솔방면구유일정적우세。
For solving the problem to quickly retrieve a specific three-dimensional component from mas-sive component library in industrial design , it gives a measurement process based on row and center by combining the compressed sensing theory , and proposes a fast retrieval method for three-view model com-ponent in manufacturing field are studied .First, it extracts the measurement features of color and texture from the three-view of each component by compressed sensing method , and saves them into model librar-y.Then, it extracts the measurement features of the three-view component to be retrieved , and matches the similarity with the features of three-view components in model library .Finally, according to the simi-larity between the three-view of model components , it fuses the measurement features of the center and gets the overall similarity between model components and outputs the retrieval result .Experimental results show this method has better retrieval performance and has certain advantage in recall and precision rate .