Real-time depth fusion using a low-cost depth sensor array
Real-time depth fusion using a low-cost depth sensor array
复制标题
DOI：
10.1109/3dtv.2014.6874758
复制
发表时间：
2014-07
期刊：
影响因子：
--
通讯作者：
Michael Schmeing;Elmar Krauskopf;Xiaoyi Jiang
中科院分区：
文献类型：
--
作者：
Michael Schmeing;Elmar Krauskopf;Xiaoyi Jiang
提出了一种基于多深度图融合的实时深度增强方法。我们使用低成本深度传感器的多传感器设置来获得可以在输入传感器之间任意定位的组合深度图。我们的方法填补了深度图中的漏洞并降低了传感器噪声。它通过OpenGL使用GPU加速，可作为各种需要实时深度图的应用程序的预处理步骤。我们提出了主观和客观的结果，并调查振动传感器对我们的方法的影响。
We present a method for real-time depth enhancement based on multiple depth map fusion. We use a multi-sensor setup of low-cost depth sensors to obtain a combined depth map that can be arbitrary positioned between the input sensors. Our method fills holes in the depth map and reduces sensor noise. It uses GPU acceleration via OpenGL and is applicable as a preprocessing step for a wide variety of applications where real-time depth maps are necessary. We present subjective and objective results and investigate the influence of vibrating sensors to our method.