Revision history [back]

Hello,

I am trying to do some template matching with OpenCV. The templates in the new image could be warped and rotated. However, we know the rotation matrix before-hand.

1) Create ROI from source image 2) Warp image to new image. 3) Have resultant bounding box containing border with correct pixels.

I have seen the reverse of this process done, when used in "Deskew" methods.

Question: How do i perform the perspective transform so that all pixels in the bounding box of the warped ROI are included?

Regards,

Daniel

Hello,

I am trying to do some template matching with OpenCV. The templates in the new image could be warped and rotated. However, we know the rotation matrix before-hand.

1) Create ROI from source image image.

2) Warp image to new image. image.

3) Have resultant bounding box containing border with correct pixels.

I have seen the reverse of this process done, when used in "Deskew" methods.

Question: How do i perform the perspective transform so that all pixels in the bounding box of the warped ROI are included?

Regards,

Daniel

Hello,

I am trying to do some template matching with OpenCV. The templates in the new image could be warped and rotated. However, we know the rotation matrix before-hand.

1) Create ROI from source image.

2) Warp image to new image.

3) Have resultant bounding box containing border with correct pixels.

I have seen the reverse of this process done, when used in "Deskew" methods.methods. This is generally achieved using getPerspectiveTransform and is pretty standard opencv. However, the reverse seems a little more difficult if using a pure OpenCV approach.

Question: How do i perform the perspective transform so that all pixels in the bounding box of the warped ROI are included?

Regards,

Daniel

Hello,

I am trying to do some template matching with OpenCV. The templates in the new image could be warped and rotated. However, we know the rotation matrix before-hand.

1) Create ROI from source image.

2) Warp image to new image.

3) Have resultant bounding box containing border with correct pixels.

I have seen the reverse of this process done, when used in "Deskew" methods. This is generally achieved using getPerspectiveTransform and is pretty standard opencv. However, the reverse seems a little more difficult if using a pure OpenCV approach.

Question: How do i perform the perspective transform so that all pixels in the bounding box of the warped ROI are included?

Regards,

Daniel

Hello,

I am trying to do some template matching with OpenCV. The templates in the new image could be warped and rotated. However, we know the rotation matrix before-hand.

1) Create ROI from source image.

2) Warp image template ROI to new image.

3) Have resultant bounding box containing border with correct pixels.

I have seen the reverse of this process done, when used in "Deskew" methods. This is generally achieved using getPerspectiveTransform and is pretty standard opencv. However, the reverse seems a little more difficult if using a pure OpenCV approach.

Question: How do i perform the perspective transform so that all pixels in the bounding box of the warped ROI are included?

Regards,

Daniel