Reconstruction of fruits surface models based on simple constrained triangulation | 基于简约束三角形剖分的果实表面模型重建

Chinese. 研究轮廓线三维表面模型重建中存在的嵌套、分叉和投影交叉等复杂轮廓的轮廓匹配问题。在定义轮廓距离基础上，制定了轮廓匹配的2条准则，构成拓扑结构简单的轮廓对，针对一对多且投影交叉的轮廓对，采用等比放大方法进行预处理，利用简约束算法对放大后的轮廓对进行剖分。用实际算例对提出的重建方法进行了验证。算例验证表明，提出的重建方法能快速进行拓扑结构复杂的轮廓剖分，重建合理的表面网格模型，重建只存在轮廓嵌套的南瓜模型需19.673 ms，重建最复杂的莲藕模型需51.845 ms。通过对果实MRI图像序列提取的轮廓进行剖分，能够快速进行果实表面重建。

English. This paper attempted to remove the difficulties in contours matching due to some complex contour pairs with nests, branches, and projective intersections in the process of the contour-based three dimensional (3D) surface model reconstruction. On the basis of two defined contours distances, two criteria were proposed for contour matching to construct the simple contours pairs. A novel geometric amplification method was put forward to preprocess one-to-many contour pairs with projective intersections; the simple-constrained algorithm was applied to triangulate the amplified contour pairs. The proposed reconstruction method was validated with practical examples. The experimental results verify that the proposed method is able to achieve faster contours triangulation of the complex topologies, and reasonable reconstruction of mesh surface models. 19.673 ms was needed for the reconstruction of pumpkin model with contour nesting, and 51.845 ms for reconstruction of the most complex lotus model. This paper proposed a novel method for fast surface models reconstruction of fruits by triangulating the contour pairs extracted from MRI image sequence.