根据像素颜色的变化定位坐标

Question

I am trying to locate specific coordinates in an image. I have an image that contains only 2 colours, pink and black as shown in the image. If I know an (x,y) coordinate in the pink region (marked in yellow dot at the centre) how can I find the coordinates that are in the boundary of the pink region(as shown in yellow dots at the boundary).

NB: The yellow dots are not part of the image and I'm using this just to represent the region of interest.

I just want to know whether there is any fast and better approach for doing this other than nested for loops which may really slow down the process because I've to find the boundary coordinates in multiple regions of the image.

Thank you!

Answer 1

Here is one way using Python/OpenCV and Numpy.

• Read the input
• Convert to gray
• Otsu threshold
• Crop the row containing the center
• Get all the coordinates in the row that are white
• Print the first and last coordinate
• Draw line on input
• Save results

Input:

import cv2
import numpy as np

# load image
img = cv2.imread("pink_blob.png")
hh, ww = img.shape[:2]

# convert to gray
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# threshold
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY+cv2.THRESH_OTSU)[1]

# center coords
center = (115,82)
cy = center[1]

# crop row at y center
row = thresh[cy:cy+1, 0:ww]

# get coordinates along row where it is white
# swap x and y between numpy and opencv coords
coords = np.argwhere(row==255)
num_coords = len(coords)
start = coords[0]
end = coords[num_coords-1]
start_pt = (start[1],cy)
end_pt = (end[1],cy)

print(start_pt)
print(end_pt)

# draw line from start to end coordinates on input
result = img.copy()
cv2.line(result, start_pt, end_pt, (255,255,255), 1)

# save result
cv2.imwrite('pink_blob_line.png', result)

# show result
cv2.imshow('result', result)
cv2.waitKey(0)

Start and End Coordinates:

(67, 82)
(160, 82)

Line on input image:

Answer 2

First of all, find contour of the pink region in the image. You can do this by first applying Otsu's thresholding on the image and then find contours using cv2.findContours().

Then in the contour boundary points, find the points having the same y-coordinate as that of the center pixel.

Among these points, point with the maximum x-coordinate will be the point on the right and point with minimum x-coordinate will be the point on the left.