现象描述
官方回复1
图像最左侧的噪声通常与称为“无效深度带”的现象有关,该现象会在相机靠近物体时扩大。
在当前版本的400系列摄像机的数据表文档的第59-60页中对此进行了描述。
The depth data generated with stereo vision uses the left imager as the reference for stereo matching resulting in a non-overlap region in the field of view of left and right imagers where we will not have depth data at the left edge of the frame.
Closer scenes result in a wider invalid depth band than scenes at larger distances.
用立体视觉生成的深度数据将左成像器用作立体匹配的参考,从而在左成像器和右成像器的视场中形成一个非重叠区域,在该区域中,帧的左边缘将没有深度数据。
与距离较远的场景相比,较近的场景会导致更宽的无效深度范围。
官方回复2
I thought very carefully about your question. In the past, some users have taken the approach of removing invalid depth pixels from the image by converting them to a value of zero.
I wonder if trying to use post-processing filters to adjust invalid pixels may be easier.
https://dev.intelrealsense.com/docs/post-processing-filters
A tutorial for using post-processing filters in Python code is in the link below.
https://github.com/IntelRealSense/librealsense/blob/jupyter/notebooks/depth_filters.ipynb
我非常仔细地考虑过您的问题。 过去,一些用户采用了通过将无效深度像素转换为零值来从图像中删除无效深度像素的方法。
我不知道尝试使用后处理滤镜来调整无效像素是否更容易。
https://dev.intelrealsense.com/docs/post-processing-filters
下面的链接中有一个在Python代码中使用后处理过滤器的教程。
https://github.com/IntelRealSense/librealsense/blob/jupyter/notebooks/depth_filters.ipynb