cv2.drawKeypoints not drawing keypoints on the outImage

Question

I am working with OpenCV in Python and trying to diplay the keypoints recognized by SIFT from gray image to the original color image. Here is my code

import cv2 as cv

org_img = cv.imread('/content/Sunset_org.jpg')
gray_img = cv.cvtColor(org_img,cv.COLOR_BGR2GRAY)

sift = cv.SIFT_create()
kp = sift.detect(gray_img,None)
sift = cv.drawKeypoints(gray_img, kp, outImage=org_img, flags=cv.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)
cv.imshow('Sift Img', sift)
cv.waitKey(0)

I want the keypoints to be drawn on the original image, but instead it is being diplayed on the gray image. Did I misunderstand the use of the third argument, i.e. outImage? What am I doing wrong?

Asked By: soucs

||

Source

Answer 1

To see where keypoints are located, you need to use original image as a background. This is the first argument of cv.drawKeypoints function. The third one is the output image (keypoints + original image). Output image is returned. This API is strange but to draw keypoints properly use following syntax:

sift = cv.drawKeypoints(org_img, kp, 0, flags=cv.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)

Answered By: Piotr Siekański

cv2.drawKeypoints not drawing keypoints on the outImage

Question:

Answers: