Trip to the Linux with ‘hard workers and ‘lazy workers’
Visualizing Data

20th

09,07

Enabling OpenCV (and python) FFMPEG Video in Ubuntu 9.04

Uncategorized Taesoo Kim

Introduction

The OpenCV is one of the greates and widely using image library for all developers and students. However, I am not a OpenCC and Graphics guy, but I choose OpenCV for Video Recording purpose. Since I mostly live and develop within Ubuntu, I tried writing and testing a little python-opencv program, but not working. It is mainly because the problem of new ffmpeg in ubuntu 9.04 Jaunty – see, https://bugs.launchpad.net/ubuntu/+source/opencv/+bug/311188. While searching the Internet, I found an article to use video streaming of opencv(here, http://gijs.pythonic.nl/blog/2009/may/3/getting-video-io-working-opencv-and-ubuntu-jaunty-/). I tired because I don’t want to reinvent the wheel of course, but not working like many comments on ‘cvCreateVideoWriter‘ function.

Prepare to compile

Prepare to build

1
2
sudo apt-get build-dep python-opencv
sudo apt-get install libswscale-dev swig

Prepare to configure

Link avcodec.h, avformat.h, avio.h, avutil.h and swscale.h in the /usr/include/ffmpeg (you have to create) like below.

1
2
3
4
5
6
ls /usr/include/ffmpeg
avcodec.h -> ../libavcodec/avcodec.h
avformat.h -> ../libavformat/avformat.h
avio.h -> ../libavformat/avio.h
avutil.h -> ../libavutil/avutil.h
swscale.h -> ../libswscale/swscale.h

Checkout OpenCV

1
svn co https://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary/tags/latest_tested_snapshot

Configure & Make

1
2
cd latest_tested_snapshot/opencv
./configure --enable-shared --enable-swscale --enable-gpl --with-swig
  • –enabled-shared : as shared library
  • –enable-swscale : alternative work around for ffmpeg
  • –enable-gpl : opensource library
  • –with-swig : regenerate swig interfaces
...
    Video I/O ---------------------
    Use QuickTime / Mac OS X: no
    Use xine:                 no
    Use gstreamer:            no
    Use ffmpeg:               yes
    Use dc1394 & raw1394:     no
    Use v4l:                  yes
    Use v4l2:                 yes
    Use unicap:               no

Wrappers for other languages ========================

    SWIG                      /usr/bin/swig -c++
    Python                    yes
    Octave                    no
...

The highlighted configuration items are important, ffmpeg / swig / python. If everything is find, do make. The interface files in the checkout directory are not properly generated, so you have to generate yourself.

1
2
make
sudo make install

Test and Correct OpenCV Python files

1
2
3
4
5
6
7
8
9
10
11
$ cd /usr/local/lib/python2.6/site-packages/opencv
$ python
Python 2.6.2 (release26-maint, Apr 19 2009, 01:56:41)
[GCC 4.3.3] on linux2
Type "help", "copyright", "credits" or "license" for ...
>>> import cv.py
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "cv.py", line 8473, in <module>
    CV_SEQ_CONTOUR=CV_SEQ_POLYGON
NameError: name 'CV_SEQ_POLYGON' is not defined

Open cv.py and Edit

1
CV_SEQ_CONTOUR=CV_SEQ_POLYGON

(line 8473) to

1
CV_SEQ_CONTOUR=CV_SEQ_POLYGON_TREE

Everything is done.

Example of OpenCV and PyGame

Displaying two Video Cameras as one concatenated video and record it to a file.

screenshot_0551

screenshot_056

To make my program independent, I just copied the opencv python binding to the local directory.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
import sys
import pygame
import Image
import opencv

from PIL import Image
from pygame.locals import *

from opencv import highgui
from opencv import cv

def get_image( camera1, camera2 ) :
    im1 = highgui.cvQueryFrame( camera1 )
    im2 = highgui.cvQueryFrame( camera2 )
   
    pil1 = opencv.adaptors.Ipl2PIL( im1 )
    pil2 = opencv.adaptors.Ipl2PIL( im2 )

    img = Image.new( "RGB", (pil1.size[0] + pil2.size[0], pil1.size[1]) )
   
    img.paste( pil1, (0,0,pil1.size[0],pil1.size[1]) )
    img.paste( pil2, (pil1.size[0],0,pil1.size[0]+pil2.size[0], pil2.size[1] ) )

    return img

def CV_FOURCC( c1, c2, c3, c4 ) :
    return (((ord(c1))&255)             \
                + (((ord(c2))&255)<<8)  \
                + (((ord(c3))&255)<<16) \
                + (((ord(c4))&255)<<24))

if __name__ == "__main__" :

    camera1 = highgui.cvCreateCameraCapture( 0 )
    camera2 = highgui.cvCreateCameraCapture( 1 )

    ( width, height ) = ( 640, 480 )
    fps = 30.0
    pygame.init()
    window = pygame.display.set_mode( ( width * 2, height ) )
    pygame.display.set_caption( "WebCam Demo" )
    screen = pygame.display.get_surface()

    writer = highgui.cvCreateVideoWriter( "captured.avi",
                                          CV_FOURCC('M','J','P','G'),
                                          fps,  
                                          cv.cvSize(width * 2, height) )

    if not writer :
        print "Error on writer"
        exit(1)

    on_exit = False

    while not on_exit :
        events = pygame.event.get()
        for event in events:
            if event.type == QUIT or event.type == KEYDOWN:
                on_exit = True

        im = get_image( camera1, camera2 )

        highgui.cvWriteFrame( writer, opencv.adaptors.PIL2Ipl(im) )
       
        pg_img = pygame.image.frombuffer( im.tostring(), im.size, im.mode )
       
        screen.blit( pg_img, (0,0) )
        pygame.display.flip()
        pygame.time.delay( int( 1000 * 1.0 / fps ) )
1
2
3
python webcam.py
Output #0, avi, to 'captured.avi':
Stream #0.0: Video: mjpeg, yuvj420p, 1280x480, q=2-31, 78643 kb/s, 90k tbn, 30 tbc

Reference

  1. https://bugs.launchpad.net/ubuntu/+source/opencv/+bug/311188\
  2. http://ubuntuforums.org/showthread.php?p=7315433
  3. http://code.google.com/p/ctypes-opencv/source/checkout
  4. http://gijs.pythonic.nl/blog/2009/may/3/getting-video-io-working-opencv-and-ubuntu-jaunty-/
  5. http://www.pythonware.com/products/pil/

Disclamer

I copied all of ideas from the Internet and just do it myself for my purpose. However, if you want, feel free to use above example under GPL v.2. Thanks.

Trackback Rss 2.0

7 Responses

  1. Sam Cheng
    10 11th, 2009 at 1:17 pm

    Thank you for the guide. It is great!
    It works initially but it stops working two days later. I think maybe because I installed gstreamer-ffmpeg. But it still does not work after I uninstalled the package. I got the following error.

    ALSA lib confmisc.c:768:(parse_card) cannot find card ‘0′
    ALSA lib conf.c:3513:(_snd_config_evaluate) function snd_func_card_driver returned error: No such file or directory
    ALSA lib confmisc.c:392:(snd_func_concat) error evaluating strings
    ALSA lib conf.c:3513:(_snd_config_evaluate) function snd_func_concat returned error: No such file or directory
    ALSA lib confmisc.c:1251:(snd_func_refer) error evaluating name
    ALSA lib conf.c:3513:(_snd_config_evaluate) function snd_func_refer returned error: No such file or directory
    ALSA lib conf.c:3985:(snd_config_expand) Evaluate error: No such file or directory
    ALSA lib pcm.c:2205:(snd_pcm_open_noupdate) Unknown PCM default
    Output #0, avi, to ‘capture2.avi’:
    Stream #0.0: Video: mjpeg, xvmcmc, 640×480, q=2-31, 39321 kb/s, 90k tbn, 30 tbc
    [mjpeg @ 0x26e04d0]colorspace not supported in jpeg
    Traceback (most recent call last):
    File “./webcam1.py”, line 70, in
    cv.cvSize(width, height) )
    File “/usr/local/lib/python2.6/dist-packages/opencv/highgui.py”, line 303, in cvCreateVideoWriter
    return _highgui.cvCreateVideoWriter(*args)
    RuntimeError: openCV Error:
    Status=Bad argument
    function name=CvVideoWriter_FFMPEG::open
    error message=Could not open codec ‘mjpeg’: Unspecified error
    file_name=highgui/cvcap_ffmpeg.cpp
    line=1302

    Do you have any idea what may go wrong?

  2. admin
    10 11th, 2009 at 5:31 pm

    Could you check your “CV_FOURCC(’M',’J',’P',’G')” and try different codecs. The error msg said you don’t have mjpeg anymore. It should work, good luck!

  3. hsu
    10 11th, 2009 at 10:17 pm

    Hi Taesoo Kim;

    I’m a bit buffled at the Step 2

    Prepare to configure
    Link avcodec.h, avformat.h, avio.h, avutil.h and swscale.h in the /usr/include/ffmpeg (you have to create) like below.

    Could you please tell me what commands I should run?

    thanks
    Hsu

  4. hsu
    10 11th, 2009 at 11:16 pm

    Just ignore the question above.. I’ve got it working! thanks so much :)

  5. Sam Cheng
    10 12th, 2009 at 8:55 am

    MPEG1 (PIC1) works. Now I need to figure out why mjpeg dies… Thanks much for the reply!

  6. admin
    10 12th, 2009 at 12:13 pm

    I am happy that you are fine : )

  7. Vinz
    03 19th, 2010 at 11:03 pm

    Hey !

    Nice topic :-)
    If anyone search for an example with OpenCV and C++ there it is : http://www.geckogeek.fr/lire-le-flux-dune-webcam-camera-video-avec-opencv.html.

    Good Week !

Leave a comment ?