摘要
传统的三维重建一般是由待重建物体的图像信息求解空间参数,但实际中待重建物体由于其复杂结构很难得到准确的结果。本文提出一种由图像中特征显著的参照物作为标尺进而对待测物体进行重建的策略。该策略首先采用水平集分割算法分割出作为参照物物体特征点集,然后采用GASAC遗传算法估计基础矩阵,从而完成目标物体或全景的三维重建。文中对相关算法进行了改进和调整,提出了参照物的选取算法。最后实验证明,该策略避免了误差特征点的干扰,可以应用于一些复杂情况下的重建,且能保证获得重建理想效果。
The traditional three-dimensional reconstruction is to solve space parameter by the image information of reconstruction object,but the actual reconstruction of the object is difficult to get the accurate space parameter because of its complex structure.This paper presents a reconstruction strategy by the significant features of the image.This strategy first extracts feature point by level set segmentation algorithm and set about reference,then uses GASAC genetic algorithm to estimate fundamental matrix,thus completes the reconstruction of the objects.Correlation algorithms have been improved in this paper,this paper also proposes a selection of reference algorithm,finally,the experiments prove the feasibility of this method,which avoids the interference of error feature points,and is usually used in reconstruction in complex cases and receives a satisfactory result.
出处
《微计算机信息》
2011年第8期234-236,共3页
Control & Automation
关键词
基础矩阵
水平集分割
参照物
GASAC
三维重建
fundamental matrix
level set segmentation
reference
GASAC
three-dimensional reconstruction
