Correspondence: matching points, patches, edges, or regions across images

Invariant Local Features

Image content is transformed into local feature coordinates that are invariant to translation, rotation, scale, and other imaging parameters

Interest Point = Keypoint

interest points: Suppose you have to click on some point, go away and come back after I deform the image, and click on the same points again.

Keypoint Matching: Detect points that are repeatable and distinctive

1. Find a set of distinctive keypoints

2. Compute a local descriptor from the region around each keypoint

3. Match local descriptors

Local features: main components

1. Detection: Identify the interest points
   1. Corner Detection

shifting window in any direction should give a large change in intensity

Let’s look for patches that have high gradients in the x and y directions.

For a patch to be a corner, the gradient distribution needs to be full rank

• We should check more than 2 pixels

Corners as distinctive interest points