Pipeline of rgb-d data collection:
- refer to blog, collect data in real-time. Power on = k4a::device::open() -> dev.start_cameras() Capture RGB-D data = k4a_capture_t capture_handle (A capture represents a set of images that were captured by a device at approximately the same time. A capture may have a color, IR, and depth image, link) -> Write color image and depth image = capture.get_color_image() / capture.get_depth_image() -> cv::imwrite() -> Power off = dev.close() (The code is in file ManipulateAK.cpp, but it is not complete)
- refer to official example, using playback() function to read mkv file. (The code is in file mkv_Process.cpp, but the result is strange)
- Recording rgbd data by using Azure Kinect SDK -> .mkv file
- Using API (azure_kinect_mkv_reader.py) of python package Open3D to process mkv file.
- Using reconstruction module in Open3D. Open3D scripts will generate fragments and scene folder under the path_dataset (which is specified in the config.json file) For reconstruction, tsdf-fusion also can be used, but need to have related RT. The module in Open3D is recommended.