CS194-26 Project 4 - Yin Deng

Part 1.1: Shoot and digitize pictures

Part 1.2: Recover homographies

We first define correspondence between two images manually as shown by the red dots.

We then follow instruction in this link and solve the following system of equations using least squares:

Part 1.3: Warp the images

We show the original images on the left and the rectified images on the right.

Part 1.4: Blend images into a mosaic

We show original images on the first row and blended image on the second row.

Example 1: Door

Example 2: Window

Example 3: Desk

Part 1.5: Tell us what you've learned

It was very interesting to learn how panorama are made!

Part 2.1: Interest Point Detection

By running the Harris corner detector, we detect corners almost all over the image.

Part 2.2: Adaptive Non-Maximal Suppression

By running ANMS, we significantly reduce the number of corners detected, and these corners are distributed evenly across the image.

Part 2.3: Feature Descriptor

Around each detected corner, find a 40x40 patch around it, apply a Gaussian filter, and downsample it to an 8x8 feature descriptor.

Part 2.4: Feature Matching

Using the ratio between the error of nearest neighbor and the error of second nearest neighbor to match features across two images. We choose 0.2 as our threshold.

Part 2.5: RANSAC

Apply RANSAC to compute a robust homography estimate.

Part 2.6: Examples

We show manual mosaic on the left and automatic mosaic on the right. As well can see, automatic mosaic is more accurate for the first and third scene, but it is a bit off for the second scene.

Part 2.7: What have you learned?

It's really cool to implement autostitching!