CODEWITHSUNDEEP
pythonc++opencvnumpy
anti
anti

Reputation: 3125

numpy python openCv remap

I need to switch a function from opencv c++ to opencv python. The c++ version is: (just the part i am having problems with)

map_x.at<float>(j,i) = pc.x;
map_y.at<float>(j,i) = pc.y;


remap(frame, unDistFrame, map_x, map_y, CV_INTER_LINEAR, 0, Scalar(0, 0, 0));

In python, I have:

    rows,cols,channels = frame.shape

    map_x = np.array((rows,cols, channels), np.uint8) # (that is: height, width,numchannels)
    map_y = np.array((rows,cols, channels), np.uint8)
    frameUnDist = np.array((rows,cols, channels), np.uint8)



    for i in xrange(rows):
        for j in xrange(cols):
            p1 = [i, j]

            p1_a = sendToFunction(params, p1)

           np.insert(map_x, p1_a[0])
           np.insert(map_y, p1_a[1])

          cv2.remap(frame, frameUndist, map_x, map_y, flags=CV_INTER_LINEAR+CV_WARP_FILL_OUTLIERS, fillval=(0, 0, 0, 0))

The insert line is clearly wrong, as I don't specify where to insert the value. How should I be doing this?

Thanks.

Upvotes: 0

Views: 971

Answers (2)

anti
anti

Reputation: 3125

Solved, thanks. I needed itemset.

map_x.itemset((j,i),p1_a[0])
map_y.itemset((j,i),p1_a[1])

Upvotes: 0

Rosa Gronchi
Rosa Gronchi

Reputation: 1911

In python OpenCV Mats are actually numpy arrays so you can just use map_x[j,i] = pc.x

Upvotes: 1

Related Questions