Mapping between scatter plot points and image pixels

Question

I started with a set of points from a laser scan which I plotted as a scatter plot using matplotlib. I then used plt.savefig to be able to open the plot as an image and use openCV to find contours around the points. Now, I want to be able to find the centers of the contours and plot them as points in the original scatter plot. The problem is I don't know how to create a mapping between the original scatter plot points and the image pixels. Is there a way to do this? Or another way to mark the center of the contours in matplotlib?

Note: the reason I need to draw contours is that later I need to use openCV's matchShapes function to compare the contours.

Here are the images from each step:

scatter plot

scatter plot ,

contours with centers marked in red

Now I basically want to be able to add the red markings from the image to the scatter plot.

Here is my code:

    plt.scatter(X[:,0], X[:,1], s=2)
    plt.axis('equal')
    plt.axis('off')
    plt.savefig(name)
    plt.clf()
    img = cv2.imread(name)
    imgGray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    ret, thresh = cv2.threshold(imgGray, 127, 255, 0)

    im2, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)

    height = img.shape[0]
    width = img.shape[1]
    blank_image = np.zeros((height,width,3), np.uint8)

    cv2.drawContours(blank_image, contours, -1, (255,0,0))

    for contour in contours:
        M = cv2.moments(contour)
        cX = int(M["m10"] / M["m00"])
        cY = int(M["m01"] / M["m00"])
        cv2.circle(blank_image, (cX, cY), 2, (0, 0, 255), -1)
    cv2.imwrite(name, blank_image)

UPDATE: Based on suggestions I looked at matplot's transforms function and tried the following:

fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(x_coords, y_coords, 'bo', markersize=2)
ax.axis('equal')
ax.axis('off')
height1 = fig.get_figheight()*fig.dpi
width1 = fig.get_figwidth()*fig.dpi
inv = ax.transData.inverted()

plt.savefig('img.png')

img = cv2.imread('img.png')
imgGray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
ret, thresh = cv2.threshold(imgGray, 127, 255, 0)
im2, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)

height = img.shape[0]
width = img.shape[1]
blank_image = np.zeros((height,width,3), np.uint8)

centers_x = []
centers_y = []

for contour in contours:
    M = cv2.moments(contour)
    cX = int(M["m10"] / M["m00"])
    cY = int(M["m01"] / M["m00"])

    centers_x.append(inv.transform((cX, height1-cY))[0])
    centers_y.append(inv.transform((cX, height1-cY))[1])

    cv2.drawContours(blank_image, [contour], -1, (255,0,0),1)
    cv2.circle(blank_image, (cX, cY), 2, (0, 0, 255), -1)

cv2.imwrite("test.png", blank_image)

ax.plot(centers_x, centers_y, 'ro', markersize=4)

plt.show()

This got me close but it seems like the x coordinates are still slightly off

new result]

new result .

I also tried

centers_x.append(inv.transform((width1-cX, height1-cY))[0])
centers_y.append(inv.transform((width1-cX, height1-cY))[1])

but that also didn't work.

FINAL UPDATE: adding

plt.tight_layout()

solved the problem.

Possible duplicate of [How to get pixel coordinates for Matplotlib-generated scatterplot?](https://stackoverflow.com/questions/13662525/how-to-get-pixel-coordinates-for-matplotlib-generated-scatterplot) — Diziet Asahi, Jun 13 '19 at 19:58
@DizietAsahi that helps a little, but how would I do the opposite (get point coordinates from pixel coordinates)? — ria, Jun 13 '19 at 20:30

score 0 · Accepted Answer · answered Jun 13 '19 at 21:47

0

x = np.linspace(0,1,10)
y = 5*x+2

fig, ax = plt.subplots()
ax.scatter(x,y)

height = fig.get_figheight()*fig.dpi
width = fig.get_figwidth()*fig.dpi

# the coordinates in pixel
cX = 147
cY = 142

# we need to invert the y value as matplotlib considers (0,0) to be
# on the lower left, while opencv uses upper left as origin
tX,tY = ax.transData.inverted().transform([cX,height-cY])
ax.scatter(tX,tY, s=50, c='r')
fig.savefig('test.png', dpi=fig.dpi)

answered Jun 13 '19 at 21:47

Diziet Asahi

38,379
7
60
75

thank you @DizietAsahi. This got me closer but it's still a bit off. I added an update to my original question so I could include and image of the new output. I appreciate the help! – ria Jun 13 '19 at 22:35

Mapping between scatter plot points and image pixels

1 Answers1