Several problems are encountered when data recording is concerned and more specifically video ones. Kinect sensor can record both RGB and Depth data simultaneously but their storage to hard disk has been proved quite tricky, as for 10 minutes of recording might need 3G of required space. Furthermore, an appropriate encode is required for saving Depth maps without losing any information, considering that 16-bit integer number should be converted to 8-bit ones.
Note: The following response is only addressing the Depth stream
I would prefer more of a programmatic approach to this issue. 3D depth-map data can be efficiently stored on PCs in a loss-less manner via an appropriately done frame matrix parsing algorithm. I'd recommend using a point cloud generation algorithm (e.g. the PCL), serialising the data coordinates and saving it to the drive. In this way, as you are only saving pixel coordinate (x,y,z) information, possibly via one of the Collection data structures available in all the high-level languages, I am sure the overall file size would drop substantially when the underlying Object is deserialised in order to reconstruct the underlying depth-map.
I would suggest please use the AFORGE dll and coding4fun dll they will help you to save the RGB , Depth and Skeleton data and practically saying you do no need that much space . I am using it and have saved lot of videos but nothing of that size.
- Badges
- Science method
- Similar topics
- Methodology