物体三维图像运动重建算法仿真
胡成松
【期刊名称】《计算机仿真》 【年(卷),期】2013(030)002
【摘要】在运动图像准确识别问题的研究中,在物体运动过程和背景较为复杂的情况下,物体的形变程度存在很大的随机性和不确定性,很难运用精确的形状基数量对非确定的形变加以约束.传统的非刚体物体三维图像重建算法中均为假设物体的形状基个数已知,没有运动物体形变复杂时,形状基参数估计错误带来的重建特征点缺失,影响三维图像重建真实性和重建清晰度.为解决上述问题,提出了一种散乱特征点碰撞的物体三维运动重建算法,运用动态图像序列中恢复的特征残缺的三维物体结构,在极小的特征空间内,通过散乱特征点间的碰撞联系,建立较为准确的补偿特征,对形状基估计不准进行补偿,弥补由于形状基估计不准带来的物体重建特征缺失、物体重建效果差、鲁棒性不强,不清晰的弊端.%This paper proposed an object three - dimensional movement reconstruction algorithm based on scattered feature points collision. Using the three - dimensional objects structure with incomplete characteristics and recovered from the dynamic image sequence, in the tiny feature space, through the collision contact of the scattered feature points, more accurate compensation characteristics were established to compensate the object reconstruction characteristics loss, poor object reconstruction effect, weak robustness andvague, due to inaccurate estimation of shape base.
物体三维图像运动重建算法仿真
