本文選題：Delaunay + 柵格　； 參考：《西北農(nóng)林科技大學(xué)》2017年碩士論文

【摘要】：隨著對(duì)計(jì)算機(jī)圖形學(xué)領(lǐng)域研究的進(jìn)一步深入,其所衍生出的虛擬現(xiàn)實(shí)以及增強(qiáng)現(xiàn)實(shí)等研究方向吸引了眾多的研究學(xué)者。同時(shí)也對(duì)計(jì)算機(jī)圖形學(xué)的發(fā)展提出了更大的需求,相關(guān)產(chǎn)業(yè)所涉及的快速建模也一度成為研究學(xué)者所熱衷的研究方向。點(diǎn)云數(shù)據(jù)的獲取主要通過(guò)掃描設(shè)備或者從圖片中提取特征點(diǎn),點(diǎn)云數(shù)據(jù)獲取方式的不同所得到的點(diǎn)云密度也不盡相同,研究者針對(duì)不同密度點(diǎn)云數(shù)據(jù)提出了基于稀疏或稠密點(diǎn)云數(shù)據(jù)的網(wǎng)格重建方法,現(xiàn)有的方法對(duì)不同密度點(diǎn)云數(shù)據(jù)魯棒性不強(qiáng),本研究針對(duì)點(diǎn)云分布密度不均勻以及無(wú)固定的點(diǎn)云密度界定等問(wèn)題提出了一種提高三維重建普適性的基于點(diǎn)云泊松分布的自適應(yīng)三維重建方法。本研究基于上述說(shuō)明問(wèn)題,提出了改進(jìn)方法。主要完成了以下工作:(1)獲取到的點(diǎn)云數(shù)據(jù)集呈離散分布,首先對(duì)點(diǎn)云數(shù)據(jù)進(jìn)行預(yù)處理操作和數(shù)據(jù)分析,為下一步柵格化做論證。進(jìn)而根據(jù)柵格中進(jìn)行曲面擬合所需最小三維點(diǎn)數(shù)及泊松分布適用條件確定柵格閾值范圍,然后在閾值范圍內(nèi)動(dòng)態(tài)調(diào)整閾值大小并迭代柵格劃分步驟,直到閾值可以使得柵格劃分單元中點(diǎn)云分布符合泊松分布,從而保證算法處理數(shù)據(jù)的穩(wěn)定性和自適應(yīng)特性的魯棒性。(2)對(duì)確定閾值大小后的點(diǎn)云數(shù)據(jù)進(jìn)行柵格劃分并建立相應(yīng)的鄰域關(guān)系與拓?fù)潢P(guān)系。在不考慮點(diǎn)云密度的情況下,通過(guò)迭代過(guò)程中動(dòng)態(tài)調(diào)整后的閾值對(duì)三維數(shù)據(jù)點(diǎn)進(jìn)行空間劃分,使得劃分之后的柵格單元中點(diǎn)云分布趨于穩(wěn)定;然后對(duì)劃分得到的柵格單元中的三維數(shù)據(jù)點(diǎn)構(gòu)建鄰域的拓?fù)溧徑P(guān)系,同時(shí)構(gòu)建柵格單元的26鄰域關(guān)系,從而保證三角剖分生成網(wǎng)格過(guò)程中搜尋擴(kuò)展點(diǎn)線的范圍,提高擴(kuò)展搜尋速度。(3)選取最佳候選點(diǎn)不斷擴(kuò)展生成三角網(wǎng)格。以柵格劃分和建立的拓?fù)潢P(guān)系為基礎(chǔ),選擇合適的初始三角形作為種子三角形遍歷點(diǎn)云數(shù)據(jù)集進(jìn)行擴(kuò)展,通過(guò)對(duì)柵格中點(diǎn)的鄰域搜索和對(duì)候選點(diǎn)所在柵格鄰域范圍的搜索查找,直到所有的三維點(diǎn)和生成的線及三角面片都加入網(wǎng)格中,以此來(lái)驗(yàn)證自適應(yīng)柵格劃分的效果和算法的魯棒性。實(shí)驗(yàn)結(jié)果表明,對(duì)比分析三角面片數(shù)和運(yùn)行耗時(shí)兩個(gè)參數(shù),本研究方法在三角剖分面片數(shù)增加了2.5%~27.9%,在運(yùn)行耗時(shí)降低了5.0%~31.6%。實(shí)驗(yàn)表明在大部分情況下本文算法在運(yùn)行效率和三角形面片的交叉和誤配方面魯棒性較強(qiáng)。
[Abstract]:With the further development of computer graphics, the virtual reality and augmented reality have attracted a lot of researchers. At the same time, the development of computer graphics has put forward a greater demand, and the rapid modeling involved in related industries has once become a hot research direction for researchers. The point cloud data is obtained mainly by scanning equipment or extracting feature points from the image, and the point cloud density is different from the different acquisition methods of point cloud data. Based on sparse or dense point cloud data, researchers proposed a mesh reconstruction method for different density point cloud data. The existing methods are not robust to point cloud data with different densities. In this paper, an adaptive 3D reconstruction method based on Poisson distribution of point cloud is proposed to improve the universality of 3D reconstruction. Based on the above problems, an improved method is proposed. Firstly, the point cloud data is preprocessed and analyzed to prove the rasterization of the next step. Then the threshold range is determined according to the minimum number of 3D points needed for surface fitting and the applicable conditions of Poisson distribution in the grid, and then the threshold value is dynamically adjusted and the grid partition steps are iterated within the threshold range. Until the threshold value enables the point cloud distribution in the grid partition unit to conform to the Poisson distribution, Therefore, the stability of the data and the robustness of the adaptive property of the algorithm are guaranteed. The point cloud data after determining the threshold value are raster partitioned and the corresponding neighborhood and topology relations are established. Without considering the point cloud density, the point cloud distribution in the grid cell is stable by dividing the 3D data points by the dynamically adjusted threshold in the iterative process. Then, the topological proximity of the three dimensional data points in the grid cells is constructed, and the 26 neighborhood relations of the grid cells are constructed, so as to ensure the range of the extended point lines in the process of triangulation generation. Improve the speed of extended search. 3) select the best candidate point and expand to generate triangular mesh. Based on the topological relation of grid partition and establishment, the suitable initial triangle is selected as the seed triangle to traverse the point cloud data set. By searching the neighborhood of the middle point of the grid and searching the range of the candidate point in the grid neighborhood, Until all the 3D points and generated lines and triangles are added to the mesh to verify the effectiveness of the adaptive grid partition and the robustness of the algorithm. The experimental results show that the number of triangles and the running time are compared and analyzed. The number of triangulated slices is increased by 2.5% and 27.9%, and the running time is reduced by 5.0% and 31.6%. Experiments show that in most cases, the proposed algorithm is robust in terms of running efficiency and cross and mismatch of triangular surfaces.
【學(xué)位授予單位】：西北農(nóng)林科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.41

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 胡志勝;于敬武;束濤;;一種結(jié)合了柵格化和特征判斷的點(diǎn)云壓縮方法[J];遼寧工程技術(shù)大學(xué)學(xué)報(bào)(自然科學(xué)版);2015年08期

2 劉慶科;;淺談利用三維掃描拓展數(shù)字媒體藝術(shù)[J];新西部(理論版);2014年14期

3 紀(jì)志浩;于明旭;;基于點(diǎn)云數(shù)據(jù)三維重建方法的研究[J];黑龍江工程學(xué)院學(xué)報(bào);2014年03期

4 徐丹;;論三維掃描打印技術(shù)對(duì)文物保護(hù)的作用[J];數(shù)位時(shí)尚(新視覺(jué)藝術(shù));2014年02期

5 邱春麗;許宏麗;;一種散亂點(diǎn)云空間直接剖分算法[J];計(jì)算機(jī)科學(xué);2014年02期

6 周婷;彭正洪;密新武;;Delaunay三角網(wǎng)生長(zhǎng)算法改進(jìn)與實(shí)現(xiàn)[J];圖學(xué)學(xué)報(bào);2013年05期

7 段紅娟;;簡(jiǎn)述三維點(diǎn)云處理技術(shù)的研究[J];電子技術(shù)與軟件工程;2013年14期

8 張春成;;一種改進(jìn)的點(diǎn)云數(shù)據(jù)柵格算法[J];兵工自動(dòng)化;2013年06期

9 段瑩;管濤;盧金婷;;OpenGL在虛擬現(xiàn)實(shí)技術(shù)中的應(yīng)用研究[J];軟件導(dǎo)刊;2012年06期

10 譚志輝;黃運(yùn)保;王啟富;陳立平;;基于移動(dòng)最小二乘曲面的點(diǎn)云拼接算法[J];計(jì)算機(jī)輔助工程;2011年04期

相關(guān)博士學(xué)位論文 前1條

1 韋爭(zhēng)亮;基于彩色編碼的結(jié)構(gòu)光動(dòng)態(tài)三維測(cè)量及重構(gòu)技術(shù)研究[D];清華大學(xué);2009年

相關(guān)碩士學(xué)位論文 前8條

1 田改改;基于三維點(diǎn)云數(shù)據(jù)的表面重建技術(shù)研究[D];北京工業(yè)大學(xué);2015年

2 董天琪;散亂點(diǎn)云的三角網(wǎng)格重建關(guān)鍵技術(shù)研究[D];西北農(nóng)林科技大學(xué);2015年

3 蔡志敏;基于點(diǎn)云數(shù)據(jù)的精簡(jiǎn)算法研究[D];北京建筑大學(xué);2014年

4 廖家志;基于曲面約束的空間Delaunay三角剖分方法研究[D];電子科技大學(xué);2013年

5 譚小俊;大規(guī)模點(diǎn)云的處理及曲面重建技術(shù)[D];浙江工業(yè)大學(xué);2012年

6 劉俊;移動(dòng)最小二乘散點(diǎn)曲線曲面擬合與插值的研究[D];浙江大學(xué);2011年

7 宮世偉;Delaunay三角剖分方法在三維地形可視化中的應(yīng)用[D];遼寧科技大學(xué);2011年

8 王秀梅;點(diǎn)云數(shù)據(jù)的三角剖分及模型簡(jiǎn)化[D];南京理工大學(xué);2009年

，

本文編號(hào)：1963514