一种基于图像的快速虚拟植物可视化重建

A Fast Visual Modeling Plant Based on Images

提出了一种视觉意义上快速植物三维重建方法,近似重构一株视觉上“相似植物”。该方法利用已有的植物结构信息,基于图像分割和三维重构技术,获取植物轮廓从而重构植物的骨架;结合植物真实图像修正生成的植物模型,最终生成视觉“真实”植物。该方法基于图像的三维重构技术,避开了植物复杂的形态结构规则提取,实现“快速意义”的重建,并采用真实图像验证技术,确保视觉上的近似,提高了对最终生成图像的控制度。

This paper proposes a fast visual modeling plant based on images, reconstruct an approximate plant model. Firstly, based on knowledge of growth patterns for plant, using image segmentation and 3D reconstruction to construct the plant skeleton, modeling leaves, and growing the remainder of the plant. Then the system produce the realistic plant model images based on image synthesis and validation. It is not necessary to acquire the complex production rules of L-systems. The proposed method provides a highly control over final shape, resulting in realistic reconstruction.