Interest Points and corners

Question 1

What are some applications of using Interest points and keypoint matching?

Answer 1

• Alignment
• Recognition
• Reconstruction

Question 2

What is the fundamental matrix of an image?

Answer 2

Its the “standard” form of the image. Like when you are looking straight ahead on a face, it has a fundamental matrix. From other angles it looks different, but it is still the same face with the same fundamental matrix.q

Question 3

Describe the steps of keypoint matching.

Answer 3

• Find a set of keypoints
• Define a region around each point
• Compute local descriptor
• Match local descriptors

Question 4

What are local features?

Answer 4

Features in certain points/regions of the image. They should be salient, repeatable and effective in use.

Question 5

How does corner detection work?

Answer 5

If you shift a window IN ANY DIRECTION, it should give a large change in the intensity. So we can use that to find these points.

Question 6

Explain Harris corner detection and its method.

Answer 6

You compute a M matrix for each image window which gives you its score. Then you find the surrounding points where the window gets a large response. Lastly you perform non-max suppression to remove bad detections.

Question 7

How does corner detection respond to different image changes?

Answer 7

Works with translation (moving the object in the scene) and rotation. Does nOT work with scaling.