Pass C array to python via ctypes

Question

I have a C function which controls a camera, I am trying to send the image buffer to Python via Ctypes to be used in a GUI.

I can grab the buffer, but I am stuck on basically how to get this to work, I suppose I could look into numpy-c api but it looks very confusing.

Thus far I have tried the following:

Create a C array (I guess it is by pointer reference), something like:

   UINT16 * MyFunction()
    {

        ...grab image... and return Buffer

        size = 2088 * 2080 * sizeof( UINT16);
        array = (BYTE*) malloc(size);
        API_Write_to_buffer(Buffer, 0, array, size);

        API_Free_Buffer(Buffer);


        return array;
    }

I can try to get the return of the array in python:

Data = MyFunction() 
array = (c_ubyte * size).from_address(addressof(Data.contents))

where MyFunction() looks like the following:

def MyFunction():

    MyFunction= lib.MyFunction
    MyFunction.restype = POINTER(c_ubyte)
    img_arr = MyFunction()
    return img_arr

Another option is to read line by line using pointers:

for(i=0;i<Width;i++)
{
    Ptr = (UINT16*)(bufferPointer + i*bufferWidth);
    for(j=0;j<Height;j++)
    {
    ...
    }
}

UPDATE:

It turns out I need to assign a pointer to my array, but it is a UINT16 in C. When I try to get the array into a numpy array, python crashes. I can get this as a return from the wrapper function:

def Grab(nframes):

    Grab= 2112 * 2088 * 8   (size of frame width * height * pixelDepth)
    Grab= lib.MyFunction
    Grab.argtypes = [c_uint32]
    Grab.restype = POINTER(c_uint16 * array_length)
    r = Grab(nframes)
    return r

Calling the function looks like this:

Data = Grab(1)
print Data

Which returns this:

<__main__.LP_c_ushort_Array_4409856 object at 0x000000000436FE48>

You freed the array before returning it? Wouldn't you get an invalid address at that point, then? — Santa
– Santa, Commented Sep 23, 2014 at 19:09
Indeed, that was a typing mistake on this post, good catch. But I do not have that problem in the actual code. Thanks! — user2221667
– user2221667, Commented Sep 23, 2014 at 19:13
What does not work? What is the error you get? What are you trying to do that you don't know how to do? The conversion to c_ubyte array looks correct to me. From the question here, it is unclear what is the problem that you have. — pv.
– pv., Commented Sep 23, 2014 at 19:40

derricw · Accepted Answer · 2014-10-02 01:13:15Z

1

Lets say you have a c_ubyte array img_buffer and it has a width and a height, because it is an image. I'm going to assume that you have that already because it looks like you do in your code.

Your cast for the c_ubyte array should look something like this:

img_buffer = (c_ubyte * (height * bytes_per_pixel) *
         (width * bytes_per_pixel)).from_address(addressof(data.contents))

You can use numpy to convert it to an ndarray using:

import numpy as np

img = np.ndarray(buffer=img_buffer, dtype=np.uint16, shape=(width, height, 1))

Once you have the ndarray you can use PIL or matplotlib or OpenCV to show it.

EDIT: Saw that your image is 16-bit so I changed the data type.

edited Oct 2, 2014 at 1:13

answered Sep 23, 2014 at 20:29

derricw

7,0643 gold badges33 silver badges36 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

user2221667 Over a year ago

I tried that, but I get : TypeError: buffer is too small for requested array.

derricw Over a year ago

Then your height, width, or data_type doesn't match your c_ubyte buffer size. Check your buffer size and compare it to what you expect for your image and ensure that they match.

Collectives™ on Stack Overflow

Pass C array to python via ctypes

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related