The property of maintaining the lens state of the liquid crystal (LC) lens during the switching between positive and negative lens states is made use of in the fast acquirement of multi-focus images without magnification change. A depth from focus (DFF) pipeline that can generate a low-error depth map and an all-in-focus image is proposed. The depth of the scene is then obtained via DFF pipeline from the captured images. The depth sensor proposed in this paper has the advantages of simple structure, low cost, and long service life.

针对双Kinect扫描大型物体过程中双Kinect外参标定问题,利用改进的随机采样一致性方法提取了点云球体及平面模型的关键参数,分析了不同Kinect坐标系下的球心坐标及平面法向量间的关系,提出了一种基于球体及平面模型的双Kinect空间位置的标定方法,实现了视角位移变化较大的平移旋转矩阵的标定。相对于利用传统迭代最近点(ICP)算法及改进的ICP算法标定两个Kinect外参的方法,所提方法的速度、精度均有较大幅度提高。最后,利用实际模型的配准结果验证了所提方法的可行性,所提方法为视角位移变化较大的双Kinect外参的标定提供了一种快捷、准确的标定方案。

基于深度传感器骨骼追踪提出了一种新的快速人体测量方法。通过深度传感器获取骨骼关节信息和人体深度数据, 完成了人体特征尺寸中部分长度和围度的测量。实验结果表明, 该方法能在短时间内获得较准确的结果, 能用低成本设备在狭小场地完成部分人体特征尺寸的获取。