【摘要】：醫(yī)療三維幾何模型的對稱平面定位適用于從生物識(shí)別到口腔正畸科和整形外科的醫(yī)學(xué)面部重建等各種應(yīng)用場景。對稱面定位的關(guān)鍵通常是三維幾何模型表面的配準(zhǔn)技術(shù),其中使用最廣泛且最具代表性的是迭代最近點(diǎn)算法,基于迭代最近點(diǎn)配準(zhǔn)技術(shù)的對稱面定位避免了傳統(tǒng)點(diǎn)構(gòu)建法生成對稱平面容易產(chǎn)生隨機(jī)偏差的問題,該研究對于臨床醫(yī)學(xué)領(lǐng)域具有十分重要的意義。本文首先對基本的迭代最近點(diǎn)算法做了詳細(xì)的介紹,主要包括算法的基本原理、算法步驟和坐標(biāo)變換參數(shù)的求解等,接著研究了迭代最近點(diǎn)算法的時(shí)間復(fù)雜度和特性等問題,并通過實(shí)驗(yàn)對算法的理論進(jìn)行了驗(yàn)證,分析得到算法的待改進(jìn)點(diǎn)。本文提出了一種基于快速近似最近鄰搜索和多重曲率約束的迭代最近點(diǎn)算法,該算法從標(biāo)準(zhǔn)算法的不足之處入手,將主成分分析應(yīng)用于初始配準(zhǔn),在算法的點(diǎn)對匹配問題上,一方面用改進(jìn)的最近鄰搜索方法進(jìn)行加速,另一方面則通過建立多重曲率特征的約束對匹配的精度進(jìn)行優(yōu)化,并通過實(shí)驗(yàn)證實(shí)了改進(jìn)算法的確具備更好的配準(zhǔn)效果。本文將上述改進(jìn)的迭代最近點(diǎn)配準(zhǔn)技術(shù)應(yīng)用于醫(yī)學(xué)顱骨模型的對稱面定位功能,通過點(diǎn)云配準(zhǔn)技術(shù)對三維模型進(jìn)行鏡像、配準(zhǔn)和對稱平面定位等操作,并從量化的角度分析了提取平面的對稱效果,實(shí)驗(yàn)證明定位的對稱面比較精確,對于臨床醫(yī)學(xué)領(lǐng)域具有較好的實(shí)用價(jià)值。
[Abstract]:The symmetrical plane localization of 3D medical geometry model is suitable for various applications from biometric recognition to orthodontic and orthodontic medical facial reconstruction. The key of symmetrical surface location is usually the registration technique of 3D geometric model surface. The most widely used and most representative is iterative nearest point algorithm. The symmetric plane location based on iterative nearest point registration technique avoids the problem that the symmetric plane generated by the traditional point construction method is prone to generate random deviations. This research is of great significance in the field of clinical medicine. In this paper, the basic iterative nearest point algorithm is introduced in detail, including the basic principle of the algorithm, the steps of the algorithm and the solution of coordinate transformation parameters, and then the time complexity and characteristics of the iterative nearest point algorithm are studied. The theory of the algorithm is verified by experiments, and the points for improvement are obtained. In this paper, an iterative nearest point algorithm based on fast approximate nearest neighbor search and multiple curvature constraints is proposed. Starting with the shortcomings of the standard algorithm, principal component analysis is applied to the initial registration. On the one hand, the improved nearest neighbor search method is used to accelerate the algorithm; on the other hand, the matching accuracy is optimized by establishing the constraints of multiple curvature features, and the experimental results show that the improved algorithm does have better registration effect. In this paper, the improved iterative nearest point registration technique is applied to the symmetrical plane location function of the medical skull model. The 3D model is mirrored, registered, and symmetrical plane location is performed by the point cloud registration technique. From the point of view of quantification, the symmetry effect of extracting plane is analyzed. The experimental results show that the symmetrical plane is more accurate and has good practical value in the field of clinical medicine.
【學(xué)位授予單位】：浙江大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.41;TP18

【相似文獻(xiàn)】

相關(guān)期刊論文 前2條

1 ;謹(jǐn)防“畫病”[J];新聞通訊;1987年06期

2 張寶貴;劉露;宋明;蔣田仔;;一種精確提取三維腦圖像對稱面的方法[J];中國醫(yī)學(xué)影像技術(shù);2011年08期

相關(guān)碩士學(xué)位論文 前2條

1 程小富;醫(yī)療三維幾何模型的對稱面定位研究[D];浙江大學(xué);2017年

2 羅莉;具有一個(gè)彈性對稱面的各向異性介質(zhì)橢圓夾雜反平面問題[D];同濟(jì)大學(xué);2008年

，

本文編號(hào)：2240618