Raspberry Pi + Webcam + Python + OpenCV = Weird Problems + No Solutions

Update,

Had a stroke of genius, I changed the end of the code to be as follows:

img3=TakePic(cam)
ShowImage('img3',img3)

#AddWeighted(img2,0.4,img3,0.6,20,img1)
ShowImage('img1 again',img1)

k=WaitKey(100000)
DestroyAllWindows()
del(cam)

So now it doesnt do the AddWeighted, it is just meant to show img1 again, however when I run this it shows img3, so for some unknow reason you can only ever access the last image created... (unless someone thinks otherwise)

Anyone have any idea how to get around this?

Wrote two functions (below). I capture an image, convert to mat, capture an image, convert to mat, add the two mats, convert to image and show..... now this works, it shows overlapped images however this is awful coding and terribly slow, but proves the Add function works, and the theory is there there is jsut something wrong with the IplImage's

def Img2Mat(img):
     width=img.width
     height=img.height
     depth=img.depth
     mat=CreateMat(height,width,depth)
     for w in range(0,width):
          for h in range(0,height):
               value=Get2D(img,h,w)
               Set2D(mat,h,w,value)

     return mat

def Mat2Img(mat):
     width=mat.cols
     height=mat.rows
     size=(width,height)
     img=CreateImage(size,8,3)
     for w in range(0,width):
           for h in range(0,height):
                value=Get2D(mat,h,w)
                Set2D(img,h,w,value)
     return img

Daniel,

  I think your assignments to img1, img2, and img3 are just copying pointers,

not the full images, so all the pointers end up pointing to the last image taken

by TakePic. 

Ah I thought I had escaped the clutches of Pointers moving from C to Python, but then python openCV is C underneath yeah?

Anyway I'll try looking into making sure everything is pointing in the right direction tomorrow.

Thanks heaps for your help.

Dan.

Guido explains here:

http://www.python.org/doc/essays/ppt/acm-ws/sld039.htm

coder27 you legend, you have fixed all my problem and I wish to send you flowers.... or maybe thanks is just enough?

Here is the easy fix,

cam=CaptureFromCAM(-1)

img=QueryFrame(cam)

img1=CloneImage(img)

img=QueryFrame(cam)

img2=CloneImage(img)

img=Add(img1,img2,img)

img3=CloneImage(img)

Cheers everyone,

Dan.

hello Dan

we write python code

import improc

from improc import*

camera= Camera (320,240)

viewer = Viewer (640,480,"pic.")

img= camera.grabImage()

viewer.displayImage (img)

for x in range  (0, img.width) :

           for y in range(0, img.height) :

                      red, green, blue = img [x, y]

if red > green  and  red >blue :

   img [x, y]  = 255, 0, 0

elif green > red and green > blue :

img[x, y] =0, 255, 0

else

img[x, y] = 0, 0, 255

viewer.displayImage (img)

waitTime (1000)

we take the image  viewer,but we want to  save image on raspberry pi

how can i do this ? what does it comment ?

Hi osman,

Sorry but I'm not familar with the python package improc. Is it a standard package or did you create it? I'm assuming you want to save "img" as a JPEG, PNG or other format?

I could do this in openCV but not sure about improc sorry.

I would use:

#Start Example 1

from cv import *

#All your capture and image processing here

SaveImage("img.png",img)

#End Example 1

This will work using PIL, if you don't have the Python Image Library, use the command:

     sudo apt-get install python-imaging

Or if you have PIP:

    sudo pip install PIL

Anyway here is the code

#Start Example 2

import improc

from improc import *

from PIL import Image #Importing the PIL package

camera=Camera(320,240)

viewer=Viewer(640,480,"pic.")

img=camera.grabImage()

viewer.displayImage(img)

img2=Image.new('RGB',(320,240),(0,0,0)) #Creates a blank PIL Image

pixels=img2.load() #Create the pixel map

for x in range (0,img.width):

    for y in range(0,img.height):

        red,green,blue=img [x, y]

        if red>green and red>blue:

            img[x,y]=(255,0,0)

            pixels[x,y]=(255,0,0)

        elif green>red and green>blue:

            img[x,y]=(0,255,0)

            pixels[x,y]=(0,255,0)

        else

            img[x,y]=(0,0,255)

            pixels[x,y]=(0,0,255)

viewer.displayImage(img)

waitTime(1000)

img2.save('img2.png') #Save the img as a PNG file

#End Example 2

Note I just change the pixels variable, I am pretty sure that pixels points to the pixel values of img2 therefore if you change pixels, img2 changes.

Any issues let me know

Daniel thanks for codes

but i want to know how can i compile this code and runn i could not find these thinks

Hi Osman,

I still don't know what the improc libaray is so wouldnt be able to say how to compile or run.

Assuming you don't have OpenCV which is what I use to do my image processing on the Raspberry Pi in Python, you can follow my steps below (but it takes 5 hours)

or there is this guide which I haven't done before

http://www.technolabsz.com/2013/03/how-to-easily-install-opencv-on.html

Once that is installed you can use this code:

#START EXAMPLE CODE

from cv import *

SIZE=(HEIGHT,WIDTH)=(320,240) # Image capture size
CAMERA_INDEX=-1 # -1 means any camera attached

cam=CaptureFromCAM(CAMERA_INDEX) # Initializes capturing a video from a camera

SetCaptureProperty(cam,CV_CAP_PROP_FRAME_WIDTH,WIDTH) # Set the capture image width

SetCaptureProperty(cam,CV_CAP_PROP_FRAME_HEIGHT,HEIGHT) # Set the capture image height

NamedWindow('image_before',CV_WINDOW_AUTOSIZE) #Create the before window

NamedWindow('image_after',CV_WINDOW_AUTOSIZE) #Create the after window

img1=QueryFrame(cam) #capture the frame from the camera

img2=CloneImage(img1) #Copy the image to be edited

W,H=GetSize(img1)

for x in range (0,W):

    for y in range(0,H):     

        red,green,blue=get2D(img1,x,y)

        if red>green and red>blue:

            Set2D(img2,x,y,(255,0,0) )

        elif green>red and green>blue:

            Set2D(img2,x,y,(0,255,0) )

        else

            Set2D(img2,x,y,(0,0,255) )

SaveImage('img2.png',img2)

ShowImage('image_before',img1) #Show before image

ShowImage('image_after',img2) #Show after image

k=WaitKey(100000) #keep the images up
DestroyAllWindows() #clean up
del(cam) #clean up
#END EXAMPLE CODE

Cheers,

Dan.

1 Install OpenCV for Python TAKES APPROX 5 HOURS!!!

  1.1 Install the dependant packages using the command:

          $sudo apt-get -y install build-essential cmake cmake-qt-gui pkg-config libpng12-0 libpng12-dev libpng++-dev libpng3 libpnglite-dev zlib1g-dbg zlib1g zlib1g-dev pngtools libtiff4-dev libtiff4 libtiffxx0c2 libtiff-tools

        and some more:

          $sudo apt-get -y install libjpeg8 libjpeg8-dev libjpeg8-dbg libjpeg-progs ffmpeg libavcodec-dev libavcodec53 libavformat53 libavformat-dev libgstreamer0.10-0-dbg libgstreamer0.10-0 libgstreamer0.10-dev libxine1-ffmpeg libxine-dev libxine1-bin libunicap2 libunicap2-dev libdc1394-22-dev libdc1394-22 libdc1394-utils swig libv4l-0 libv4l-dev python-numpy libpython2.6 python-dev python2.6-dev libgtk2.0-dev pkg-config

  1.2 Get the source for OpenCV using the command:

          $wget http://sourceforge.net/projects/opencvlibrary/files/opencv-unix/2.4.3/OpenCV-2.4.3.tar.bz2

  1.3 Extract the archive using the command:

          $tar -xvjpf OpenCV-2.4.3.tar.bz2

  1.4 Remove the file (space saving)using the command:

          $rm OpenCV-2.4.3.tar.bz2

  1.5 Open the directory using the command:

          $cd OpenCV-2.4.3

  1.6 Make a new directory using the command:

          $mkdir build

        and open it:

          $cd build

  1.7 Configure the build using the command:

          $cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local -D BUILD_PYTHON_SUPPORT=ON -D BUILD_EXAMPLES=ON ..

  1.8 Now install (TAKES 5 HOURS!!!! have a coffee) using the commands:

          $make

        and then:

          $sudo make install

  1.9 Open and configure the OpenCV files using the command:

          $sudo nano /etc/ld.so.conf.d/opencv.conf

        then add the following line to the end (fle may be empty, its all good):

          /usr/local/lib

  1.10 Open and edit the system bashrc file using the command:

           $sudo nano /etc/bash.bashrc

         and add to the end of the file these lines:

           PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/lib/pkgconfig export

           PKG_CONFIG_PATH

  1.11 Reference:

           http://mitchtech.net/raspberry-pi-opencv/