Sobel Edge detection in Python and Opencv

Question

the following code in python detects edge using sobel operator in horizontal as well as vertical direction

import cv2
import numpy as np

img = cv2.imread('image.bmp', cv2.IMREAD_GRAYSCALE)
rows, cols = img.shape

sobel_horizontal = cv2.Sobel(img, cv2.CV_64F, 1, 0, ksize=5)
sobel_vertical = cv2.Sobel(img, cv2.CV_64F, 0, 1, ksize=5)

cv2.imshow('Original', img)
cv2.imshow('Sobel horizontal', sobel_horizontal)
cv2.imshow('Sobel vertical', sobel_vertical)

cv2.waitKey(0)

is there any logic to detect the edge from left to right and vice versa?

Solution

When you use double (CV_64F) as a destination type, you can distinguish between left/right (or up/down) edges by the sign of pixel value in the output image (remember that sobel is a smoothed numerical approximation of derivative, so this is quite natural)

Comments

[Jeru Luke]

Yes, but does it use the same kernel as that of Sobel operator for convolution?

[alexisrozhkov]

I can't say for sure right now, but I don't see a reason to think otherwise

[alexisrozhkov]

According to the doc, for kernel sizes > 1 a separable kernel is used (faster than naive approach, yields same result), also it is possible to use Scharr kernel(more accurate derivative approximation), but in your snippet it's not used

[m3h0w]

I don't see why this answer is downvoted. It answers the question. Sobel result from right to left is the result from left to right taken with the opposite sign.