tubbz-alt / kinect_3d_dev

Scripts to read out the Kinect v1 (xbox 360) scanner and combine multiple scans into meshes, pointclouds or voxels.

Kinect 3d scanning and combining

Collection of scripts to acquire depth images from one or more Kinect v1 (xbox360) sensors and generate 3d models from them. The repository is divided in two parts, with different system and python requirements:

1. kinect_interface/
2. depth_image_processing/

The scripts in kinect_interface/ are only for acquiring depth images from the kinect sensor and storing them as .npy files. These scripts use pykinect with the Microsoft Kinect SDK and these dependencies also dictate the requirements, most importantly a Windows environment and Python 2.7. For Mac and Linux computers there are other ways of getting depth images from kinect sensors like OpenNI, see for instance the guide at [https://docs.opencv.org/2.4/doc/user_guide/ug_kinect.html].

The scripts in depth_image_processing/ constitute the main part of this repo. They are platform independent and require python 3.

1) kinect_interface/

1.1) Installation

(Windows only)

1. Download & install the Microsoft Kinect SDK v1.8 [https://www.microsoft.com/en-us/download/details.aspx?id=40278]

2. Optional but recommended because of the different python version: create a virtualenv with python 2.7 interpreter:

   virtualenv --python=<path/to/python2.7/interpreter> <path/to/new/virtualenv/>

   Example: virtualenv --python=D:/python27/python.exe kinect_interface/env

   Then, activate the environment. (Run <path/to/new/virtualenv>/scripts/activate)

3. Move to the kinect_interface/ folder and install dependencies:

   pip install -r requirements.txt

1.2) Usage

1. Connect Kinect scanners and position them pointing at the center of the stage/subject.
2. Run python acquire.py, enter a name for the session and the number of connected sensors.

2) depth_image_processing/

2.1) Installation

1. Optional: create a virtualenv with python 3.x interpreter:

   virtualenv --python=<path/to/python3/interpreter> <path/to/new/virtualenv/>

   Then, activate the environment. (Run <path/to/new/virtualenv>/scripts/activate)

2. Move to the kinect_interface/ folder and install dependencies:

   pip install -r requirements.txt

2) depth_image_processing/

2.1) Installation

1. Optional but recommended because of the different python version: create a virtualenv with python 2.7 interpreter:

   virtualenv --python=<path/to/python3.x/interpreter> <path/to/new/virtualenv/>

   Example: virtualenv --python=D:/python37/python.exe depth_image_processing/env

   Then, activate the environment. (Run <path/to/new/virtualenv>/scripts/activate)

2. Move to the depth_image_processing/ folder and install dependencies:

   pip install -r requirements.txt

2.2) Usage

1. Run calibrate_manually.py:

• in the parameters in the top of the file you can specify the session name, number of sensors and the snapshot index to use for the calibration.
• Use the specified keys to position the views, then press u to save.

2. Run test_calibration.py (with the right parameters for session name and number of sensors) to test if the calibration was saved right.

3. Run combine_continuous.py (with the right parameters for session name and number of sensors)