Home | History | Annotate | Download | only in highgui 
      1 Using Kinect and other OpenNI compatible depth sensors {#tutorial_kinect_openni}
      2 ======================================================
      3 
      4 Depth sensors compatible with OpenNI (Kinect, XtionPRO, ...) are supported through VideoCapture
      5 class. Depth map, BGR image and some other formats of output can be retrieved by using familiar
      6 interface of VideoCapture.
      7 
      8 In order to use depth sensor with OpenCV you should do the following preliminary steps:
      9 
     10 -#  Install OpenNI library (from here <http://www.openni.org/downloadfiles>) and PrimeSensor Module
     11     for OpenNI (from here <https://github.com/avin2/SensorKinect>). The installation should be done
     12     to default folders listed in the instructions of these products, e.g.:
     13     @code{.text}
     14     OpenNI:
     15         Linux & MacOSX:
     16             Libs into: /usr/lib
     17             Includes into: /usr/include/ni
     18         Windows:
     19             Libs into: c:/Program Files/OpenNI/Lib
     20             Includes into: c:/Program Files/OpenNI/Include
     21     PrimeSensor Module:
     22         Linux & MacOSX:
     23             Bins into: /usr/bin
     24         Windows:
     25             Bins into: c:/Program Files/Prime Sense/Sensor/Bin
     26     @endcode
     27     If one or both products were installed to the other folders, the user should change
     28     corresponding CMake variables OPENNI_LIB_DIR, OPENNI_INCLUDE_DIR or/and
     29     OPENNI_PRIME_SENSOR_MODULE_BIN_DIR.
     30 
     31 -#  Configure OpenCV with OpenNI support by setting WITH_OPENNI flag in CMake. If OpenNI is found
     32     in install folders OpenCV will be built with OpenNI library (see a status OpenNI in CMake log)
     33     whereas PrimeSensor Modules can not be found (see a status OpenNI PrimeSensor Modules in CMake
     34     log). Without PrimeSensor module OpenCV will be successfully compiled with OpenNI library, but
     35     VideoCapture object will not grab data from Kinect sensor.
     36 
     37 -#  Build OpenCV.
     38 
     39 VideoCapture can retrieve the following data:
     40 
     41 -#  data given from depth generator:
     42     -   CAP_OPENNI_DEPTH_MAP - depth values in mm (CV_16UC1)
     43     -   CAP_OPENNI_POINT_CLOUD_MAP - XYZ in meters (CV_32FC3)
     44     -   CAP_OPENNI_DISPARITY_MAP - disparity in pixels (CV_8UC1)
     45     -   CAP_OPENNI_DISPARITY_MAP_32F - disparity in pixels (CV_32FC1)
     46     -   CAP_OPENNI_VALID_DEPTH_MASK - mask of valid pixels (not ocluded, not shaded etc.)
     47         (CV_8UC1)
     48 
     49 -#  data given from BGR image generator:
     50     -   CAP_OPENNI_BGR_IMAGE - color image (CV_8UC3)
     51     -   CAP_OPENNI_GRAY_IMAGE - gray image (CV_8UC1)
     52 
     53 In order to get depth map from depth sensor use VideoCapture::operator \>\>, e. g. :
     54 @code{.cpp}
     55     VideoCapture capture( CAP_OPENNI );
     56     for(;;)
     57     {
     58         Mat depthMap;
     59         capture >> depthMap;
     60 
     61         if( waitKey( 30 ) >= 0 )
     62             break;
     63     }
     64 @endcode
     65 For getting several data maps use VideoCapture::grab and VideoCapture::retrieve, e.g. :
     66 @code{.cpp}
     67     VideoCapture capture(0); // or CAP_OPENNI
     68     for(;;)
     69     {
     70         Mat depthMap;
     71         Mat bgrImage;
     72 
     73         capture.grab();
     74 
     75         capture.retrieve( depthMap, CAP_OPENNI_DEPTH_MAP );
     76         capture.retrieve( bgrImage, CAP_OPENNI_BGR_IMAGE );
     77 
     78         if( waitKey( 30 ) >= 0 )
     79             break;
     80     }
     81 @endcode
     82 For setting and getting some property of sensor\` data generators use VideoCapture::set and
     83 VideoCapture::get methods respectively, e.g. :
     84 @code{.cpp}
     85     VideoCapture capture( CAP_OPENNI );
     86     capture.set( CAP_OPENNI_IMAGE_GENERATOR_OUTPUT_MODE, CAP_OPENNI_VGA_30HZ );
     87     cout << "FPS    " << capture.get( CAP_OPENNI_IMAGE_GENERATOR+CAP_PROP_FPS ) << endl;
     88 @endcode
     89 Since two types of sensor's data generators are supported (image generator and depth generator),
     90 there are two flags that should be used to set/get property of the needed generator:
     91 
     92 -   CAP_OPENNI_IMAGE_GENERATOR -- A flag for access to the image generator properties.
     93 -   CAP_OPENNI_DEPTH_GENERATOR -- A flag for access to the depth generator properties. This flag
     94     value is assumed by default if neither of the two possible values of the property is not set.
     95 
     96 Some depth sensors (for example XtionPRO) do not have image generator. In order to check it you can
     97 get CAP_OPENNI_IMAGE_GENERATOR_PRESENT property.
     98 @code{.cpp}
     99 bool isImageGeneratorPresent = capture.get( CAP_PROP_OPENNI_IMAGE_GENERATOR_PRESENT ) != 0; // or == 1
    100 @endcode
    101 Flags specifing the needed generator type must be used in combination with particular generator
    102 property. The following properties of cameras available through OpenNI interfaces are supported:
    103 
    104 -   For image generator:
    105 
    106     -   CAP_PROP_OPENNI_OUTPUT_MODE -- Three output modes are supported: CAP_OPENNI_VGA_30HZ
    107         used by default (image generator returns images in VGA resolution with 30 FPS),
    108         CAP_OPENNI_SXGA_15HZ (image generator returns images in SXGA resolution with 15 FPS) and
    109         CAP_OPENNI_SXGA_30HZ (image generator returns images in SXGA resolution with 30 FPS, the
    110         mode is supported by XtionPRO Live); depth generator's maps are always in VGA resolution.
    111 
    112 -   For depth generator:
    113 
    114     -   CAP_PROP_OPENNI_REGISTRATION -- Flag that registers the remapping depth map to image map
    115         by changing depth generator's view point (if the flag is "on") or sets this view point to
    116         its normal one (if the flag is "off"). The registration processs resulting images are
    117         pixel-aligned,which means that every pixel in the image is aligned to a pixel in the depth
    118         image.
    119 
    120         Next properties are available for getting only:
    121 
    122     -   CAP_PROP_OPENNI_FRAME_MAX_DEPTH -- A maximum supported depth of Kinect in mm.
    123     -   CAP_PROP_OPENNI_BASELINE -- Baseline value in mm.
    124     -   CAP_PROP_OPENNI_FOCAL_LENGTH -- A focal length in pixels.
    125     -   CAP_PROP_FRAME_WIDTH -- Frame width in pixels.
    126     -   CAP_PROP_FRAME_HEIGHT -- Frame height in pixels.
    127     -   CAP_PROP_FPS -- Frame rate in FPS.
    128 
    129 -   Some typical flags combinations "generator type + property" are defined as single flags:
    130 
    131     -   CAP_OPENNI_IMAGE_GENERATOR_OUTPUT_MODE = CAP_OPENNI_IMAGE_GENERATOR + CAP_PROP_OPENNI_OUTPUT_MODE
    132     -   CAP_OPENNI_DEPTH_GENERATOR_BASELINE = CAP_OPENNI_DEPTH_GENERATOR + CAP_PROP_OPENNI_BASELINE
    133     -   CAP_OPENNI_DEPTH_GENERATOR_FOCAL_LENGTH = CAP_OPENNI_DEPTH_GENERATOR + CAP_PROP_OPENNI_FOCAL_LENGTH
    134     -   CAP_OPENNI_DEPTH_GENERATOR_REGISTRATION = CAP_OPENNI_DEPTH_GENERATOR + CAP_PROP_OPENNI_REGISTRATION
    135 
    136 For more information please refer to the example of usage
    137 [openni_capture.cpp](https://github.com/Itseez/opencv/tree/master/samples/cpp/openni_capture.cpp) in
    138 opencv/samples/cpp folder.
    139