矩阵分解的二次曲线分层重建计算方法研究  

Computational Methods of Conics Stratified Reconstruction Based on Matrix decomposition

作　　者：张政武[1] 张国海[1] 

机构地区：[1]陕西理工学院机械工程学院,陕西汉中723003

出　　处：《计算机仿真》2014年第1期268-271,共4页Computer Simulation

基　　金：陕西省教育厅科研计划项目资助(12JK0687)

摘　　要：三维重建足计算机视觉领域中一个至关重要的经典问题,传统采用三维重建主要是利用点集、直线集以及点线组合实现对场景物体的三维重建误差较大,目前采用二次曲线的三维重建研究受到更多关注。为解决实际计算的快速性和准确性,提出了采用矩阵分解的二次曲线分层重建计算方法,利用分层重建算法,能够实现不同变换层次上的图像矫正。改进算法是将射影变换分解为相似变换、仿射变换和约束透视变换,然后根据各分解矩阵表达式,以圆为例给出了分层重建矩阵算法。最后通过算法举例进行了算法验证。The 3D reconstruction is a crucial classic problem in computer vision. To attain a quick calculation speed and a high accuracy, a computational method of conics stratified reconstruction based on matrix decomposition was studied from the perspective of computational processes in this paper. The image rectification can be achieved in different transform levels with this method. The modified algorithm was used to decompose the projective transforma- tion into a similarity transformation, an affine transformation and a constraint perspective transformation. Then, ac- cording to the decomposed matrix expressions the computational methods of conics stratified reconstruction was pro- posed with rotundity. The computational process and verification were presented at last.

关 键 词：计算机视觉 二次曲线 分层重建 算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]