Description
DESCRIPTION
In this assignment, you will write software which stitches images together to form panoramas. Your
software will detect useful features in the images, find the best matching features in your other images,
align the photographs, then warp and blend the photos to create a seamless panorama.
This assignment can be thought of as two major components:
1. Feature Detection and Matching [Assignment #2]
2. Panorama Mosaic Stitching [this assignment]
FEATURE DETECTION AND MATCHING [Steps 1 and 2] – assignment 2
Step 1: Compute the Harris corner detector using the following steps
A. Compute the x and y derivatives on the image
B. Compute the covariance matrix H of the image derivatives. Typically, when you compute
the covariance matrix you compute the sum of Ix
2
, Iy
2
and IxIy over a window or small area of
the image. To obtain a smooth result for better detection of corners, use a Gaussian
weighted window.
C. Compute the Harris response using determinant(H)/trace(H).
D. Find peaks in the response that are above the threshold, and store the interest point
locations.
Required: Open “Boxes.png” and compute the Harris corners. Save an image “1a.png” showing the Harris
response on “Boxes.png” (you’ll need to scale the response of the Harris detector to lie between 0 and
255. ) When you’re debugging your code, I would recommend using “Boxes.png” to see if you’re getting the
right result.
C.POULLIS COMP425/6341 COMPUTER VISION WINTER 2020
Compute the Harris corner detector for images “Rainier1.png” and “Rainier2.png”. Save the images
“1b.png” and “1c.png” of the detected corners (use the drawing function cv::circle(…) or
cv::drawKeypoints(…) to draw the interest points overlaid on the images)
Step 2: Matching the interest points between two images.
A. Compute the descriptors for each interest point.
B. For each interest point in image 1, find its best match in image 2. The best match is defined
as the interest point with the closest descriptor (SSD or ratio test).
C. Add the pair of matching points to the list of matches.
D. Display the matches using cv::drawMatches(…). You should see many correct and incorrect
matches.
Required: Compute the Harris corner detector and find matches for images “Rainier1.png” and
“Rainier2.png”. Save the image “2.png” showing the image with the found matches [use
cv::drawMatches(…) to draw the matches on the two image].
PANORAMA MOSAIC STITCHING [Steps 3 and 4] – this assignment
Step 3: Compute the homography between the images using RANSAC (Szeliski, Section 6.1.4). Following
these steps:
A. Implement a function project(x1, y1, H, x2, y2). This should project point (x1, y1) using the
homography “H”. Return the projected point (x2, y2). Hint: See the slides for details on how
to project using homogeneous coordinates. You can verify your result by comparing it to
the result of the function cv::perspectiveTransform(…). [Only use this function for
verification and then comment it out.].
B. Implement the function computeInlierCount(H, matches, numMatches, inlierThreshold).
computeInlierCount is a helper function for RANSAC that computes the number of inlying
points given a homography “H”. That is, project the first point in each match using the
function “project”. If the projected point is less than the distance “inlierThreshold” from the
second point, it is an inlier. Return the total number of inliers.
C. Implement the function RANSAC (matches , numMatches, numIterations, inlierThreshold,
hom, homInv, image1Display, image2Display). This function takes a list of potentially
matching points between two images and returns the homography transformation that
relates them. To do this follow these steps:
C.POULLIS COMP425/6341 COMPUTER VISION WINTER 2020
a. For “numIterations” iterations do the following:
i. Randomly select 4 pairs of potentially matching points from “matches”.
ii. Compute the homography relating the four selected matches with the
function cv::findHomography(…)***. Using the computed homography,
compute the number of inliers using “computeInlierCount”.
iii. If this homography produces the highest number of inliers, store it as the
best homography.
b. Given the highest scoring homography, once again find all the inliers. Compute a
new refined homography using all of the inliers (not just using four points as you
did previously. ) Compute an inverse homography as well, and return their values in
“hom” and “homInv”.
c. Display the inlier matches using cv::drawMatches(…).
Required: Compute the Harris corner detector, find matches and run RANSAC for images “Rainier1.png”
and “Rainier2.png”. Save the image “3.png” of the found matches (use cv::drawMatches(…) to show the
inlier matches). You should only see correct matches, i.e. , all the incorrect matches from the previous
step should be removed. If you see all or some incorrect matches try running RANSAC with a larger
number of iterations. You may try tuning the other parameters as well.
***NOTE: The function cv::findHomography(…) can optionally use RANSAC internally, if that is enabled.
You should not have RANSAC enabled in this function i.e. third input parameter should be 0.
Step 4: Stitch the images together using the computed homography. Following these steps:
A. Implement the function stitch(image1, image2, hom, homInv, stitchedImage). Follow these
steps:
a. Compute the size of “stitchedImage. ” To do this project the four corners of “image2”
onto “image1” using project(…) and “homInv”. Allocate the image.
b. Copy “image1” onto the “stitchedImage” at the right location.
c. For each pixel in “stitchedImage”, project the point onto “image2”. If it lies within
image2’s boundaries, add or blend the pixel’s value to “stitchedImage. ” When
finding the value of image2’s pixel use bilinear interpolation [cv::getRectSubPix(…)].
C.POULLIS COMP425/6341 COMPUTER VISION WINTER 2020
Required: Compute the Harris corner detector, find matches, run RANSAC and stitch the images
“Rainier1.png” and “Rainier2.png”. Save the stitched image as “4.png”. It should look like the image
“Stitched.png”.
EXTRA CREDIT (1-5 for undergraduates) (1, 2 mandatory for graduates, 3-5 extra credit)
1. Create a panorama that stitches together the six Mt. Rainier photographs, i.e. , Rainier1.png, …
Painier6.png. The final result should look similar to “AllStitched.png”. (15pts)
2. Create your own panorama using three or more images. You must capture the images yourself,
and not find them on the web. I would recommend downsampling them before stitching, i.e. , make
them approximately the same size as the images in the homework assignment. (5pts)
3. Implement a new image descriptor or detector that can stitch the images “Hanging1.png” and
“Hanging2.png”. Save the Stitched image and the “match” images. In case you’re wondering – No, you
cannot rotate “Hanging2.png” by hand before processing, that’s cheating. (10pts)
4. Do a better job of blending the two images. That is, make the seams between the two images
invisible. One possible method to do this is to use center-weighting. See Szeliski, Sections 9.3.2 – 9.3.4.
(10pts)
5. Implement a new image descriptor and/or detector that can stitch the images ND1.png and
ND2.png. Save the Stitched image and the “match” images. (10pts)
Submission (electronic submission through EAS only)
Please create a zip file containing your C/C++ code, the required images, and a readme text file (.txt).
In the readme file document the features and functionality you have implemented, and anything else you
want the grader to know i.e. control keys, keyboard/mouse shortcuts, etc.
Additional Information
● The assignment’s images can be found here
Credits
Based on the two assignments developed by Ali Farhadi and Steve Seitz, respectively.
