Classes and functions to manage video streams and present them as images. More...
|
Modules | |
| Video buffers | |
| All classes and functions relating to video buffers (as opposed to video frames) | |
| Video frames | |
| All classes and functions relating to video frames (as opposed to video buffers) | |
Classes | |
| class | CVD::V4LControl |
| exposes the V4L2 API to set parameters on a capture device. More... | |
Functions | |
| template<class T > | |
| VideoBuffer< T > * | CVD::open_video_source (std::istream &in) |
| opens a video device described by a video source url given from an input stream. More... | |
| template<class T > | |
| VideoBuffer< T > * | CVD::open_video_source (const std::string &src) |
| opens a video device described by a video source url. More... | |
Detailed Description
Classes and functions to manage video streams and present them as images.
Function Documentation
◆ open_video_source() [1/2]
| VideoBuffer< T > * CVD::open_video_source | ( | const std::string & | src | ) |
opens a video device described by a video source url.
This allows to decide at runtime what kind of video input your program is using. Basic use is to call open_video_source<T>(url) to get a VideoBuffer<T>*.
In many cases, the VideoBuffer returned by open_video_source() is a wrapper around the video buffer dealing with the hardware and so does not provide access to the controls. The underlying buffer can be accessed with VideoBuffer::root_buffer(). Threre are also several pseudo buffers (such as deinterlacing and colorspace conversion) which are chained to other buffers by taking a URL as an argument. The url syntax is the following:
url := protocol ':' [ '[' options ']' ] // identifier protocol := "files" | "file" | "v4l2" | "jpegstream" | "dc1394" | "qt" | "colourspace" | "deinterlace" options := option [ ',' options ] option := name [ '=' value ]
identifier and values can be quoted literals with escapes, all other text is unquoted. Some Examples:
Open a DiskBuffer2 for *.pgm in /local/capture/:
files:///local/capture/ *.pgm
Open a V4L2 device at /dev/video0:
v4l2:///dev/video0
Open a V4L2 device with fields on input 2:
v4l2:[input=2,fields]///dev/video0
Open firewire camera 1 with the default fps:
dc1394://1
Open firewire camera 1, capturing in YUV411, at 30fps: colourspace:[from=yuv411]//dc1394:[fps=30]//1
Open an avi file relative to the current directory:
file://../../stuff/movie.avi
Open the first QuickTime camera and show the settings dialog
qt:[showsettings=1]//0
Open an HTTP camera. First create a named pipe from the shell, and start grabbing video:
mkfifo /tmp/video wget http//my.camera/file_representing_video -O /tmp/video
then open a source with:
jpegstream:///tmp/video
If the argument is provided from a shell such as BASH, then then redirection can be used:
jpegstream://<(wget http//my.camera/file_representing_video -O - )
Fields are: bool = true | yes | 1 | false | no | 0 offset = <width>x<height> size = <offset> | qvga | vga | pal | ntsc | xga
Options supported by the various protocols are:
'files' protocol (DiskBuffer2): identifier is glob pattern
fps = <number>
read_ahead [= <number>] (default is 50 if specified without value)
on_end = repeat_last | unset_pending | loop (default is repeat_last)
'file' protocol (VideoFileBuffer): identifier is path to file, or device name
read_ahead [= <number>] (default is 50 if specified without value)
on_end = repeat_last | unset_pending | loop (default is repeat_last)
format = <ffmpeg format name>
<ffmpeg option> = <value>
'v4l2' protocol (V4LBuffer): identifier is device name
size = <size> (default vga)
input = <number>
interlaced | fields [= <bool> ]
verbose [ = <bool> ]
'uvc' protocol (UVCBuffer): identifier is device name
size = <size> (default vga)
fps = <number>
mjpeg [ = <bool> ]
verbose [ = <bool> ]
'dc1394' protocol (DVBuffer): identifier is camera number
fps = <number> (default is camera default)
size = <size>
offset = <offset>
verbose [ = <bool> ]
reset [ = <bool> ]
mode | format7 | format7_mode = <number>
'qt' protocol (QTBuffer): identifier is camera number
size = vga | qvga | <width>x<height> (default vga)
showsettings [ = <bool> ] (default 0)
verbose [ = <bool> ] (default 0)
'jpegstream' protocol (ServerPushJpegBuffer): identifier is path to file
read_ahead [= <number>] (default is 50 if specified without value)
'deinterlace' protcol (DeinterlaceBuffer): identifier is a video URL
oddonly [ = <bool> ]
evenonly [ = <bool> ]
oddeven [ = <bool> ] (default)
evenodd [ = <bool> ]
double [ = <bool> ] (perform line doubling by averaging)
'colourspace' protcol (ColourspaceBuffer): identifier is a video URL
from = byte | mono | gray | grey | yuv411 | yuv422 | rgb<byte>
| rgb | bayer_bggr | bayer_gbrg | bayer_grbg | bayer_rggb (default mono)
'skip' protocol: identifier is a video URL
seek = <double>
drip = <int>
◆ open_video_source() [2/2]
| VideoBuffer<T>* CVD::open_video_source | ( | std::istream & | in | ) |
opens a video device described by a video source url given from an input stream.
See open_video_source(const std::string &) for details on the url syntax.
