Numpy gradient of 2D array

Question

I have a 2D array sinusoidal pattern and want to plot both the x and y gradient of that function. I have a 2D array image_data:

def get_image(params):
    # do some maths on params
    return(image)

params = (a, b, c)
image_data = get_image(params)

I use numpy.gradient to get the gradients:

gradients = numpy.gradient(image_data)
x_grad = gradients[0]
y_grad = gradients[1]

Plotting all three looks like:

This pattern is not at 45 degrees. I'd expect x and y gradients to be different. In my mind x_gradient[i][j] should be the gradient of image_data[i][j] with respect to the indexes either side and y_gradient[i][j] the gradient with respect to indexes above and below. Because the pattern is angled slightly, in image_data the gradient changes at different rates.

Am I misinterpreting my data or not understanding numpy.gradient output?

My guess is you're doing everything correctly. And while it's possible that the two gradients are the same, from the pictures I don't think they are (try print(np.all(x_gradient == y_gradient))). But even if they were, it's certainly possible for the two gradients to be the same depending on your generating function (it looks like sin(a*x + b*y) - which has periodic gradients) — jedwards
– jedwards, Commented Jun 15, 2018 at 12:10

Community · Accepted Answer · 2020-06-20 09:12:55Z

5

I'm not sure why you think the gradients should be different. But the gradient shapes for the given function should have the same shape with different levels. Let's do it analytically:

So you can see they both have the same shape. Add the color levels and you'll see the difference.

edited Jun 20, 2020 at 9:12

CommunityBot

11 silver badge

answered Jun 15, 2018 at 12:58

anishtain4

2,4102 gold badges18 silver badges22 bronze badges

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

Collectives™ on Stack Overflow

Numpy gradient of 2D array

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related