CODEWITHSUNDEEP
pythonopencvhomography
coding4fun
coding4fun

Reputation: 3507

Homography matrix in OpenCV

I am creating an application where I'll find the bird view perspective of an image by multiplying it with a homography matrix. I have read a lot about it and it seems that the function to multiply the matrices is warpPerspective(). Although when I do so, the resulting image is an image where all pixels are zero and only the first row is not.

Please provide me with the way to multiply a homography matrix by an image whether is C++ or Python.

Here's the code I'm using:

import cv2
import numpy as np

tilt = np.pi * 35 / 180

a = np.zeros((3, 3))

a[0][0] = 1
a[1][1] = np.sin(tilt)
a[1][2] = -np.sin(tilt)
a[2][2] = np.cos(tilt)
a[2][2] = np.cos(tilt)

src = cv2.imread("S2.jpg")
width, height, depth = src.shape

output = cv2.warpPerspective(src, a, (width, height))

cv2.imwrite("results.png", output)

The image I'm applying the homography on:

Source Image

The result of using warpPerspective is:

Result Image

It can be seen that there is a small part on the top-left of resulting picture.

Your help is highly appreciated!!

Upvotes: 0

Views: 2455

Answers (1)

Arnaud P
Arnaud P

Reputation: 12607

I think it's got to do with your a matrix, since calling your code with

a = np.array([[1, 0, 0],
              [0, 1, 0],
              [0, 0, 1]],
             dtype=np.float32)

works fine.

Upvotes: 1

Related Questions