Try the RANSAC option to compute homography. Change the RANSAC parameters, and explain your

results. Print and analyze the inlier numbers.

$[]$ # Your code to display book location after RANSAC here

# using RANSAC

H$,$ mask $=$ cv$2.$findHomography$($src$\_$pts$,$ dst$\_$pts$,$ cv$2.$RANSAC$)$

# draw frame

# draw inliers

inlier number $3.$ Estimate a homography transformation based on the matches, using

cv$2.$ findHomography $().$ Display the transformed outline of the first reference book cover

image on the query image, to see how well they match.

We provide a function draw$\_$outline$()$ to help with the display, but you may need to edit

it for your needs.

Try the 'least square method' option to compute homography, and visualize the inliers by

using cv$2.$ drawMatches$().$ Explain your results.

Again, you don't need to compare results numerically at this stage. Comment on what you

observe visually.

$"!$I$\text{'}$I This is just an example to show the steps involved. Inputs: query: query imageIIIIpts $=$ np$.$float$32([[0,0],[0,$h$-1],[$w$-1,$h$-1],[$w$-1,0]]).$reshape$(-1,1,2)$img $=$ query.copy$()$plt$.$imshow$($img$,$ 'gray'$),$ plt$.$show$()$

def draw$\_$inliers$($img$1,$ img$2,$ kp$1,$ kp$2,$ matches, matchesMask$)$: Draw inlier between images kp$1/$ kp$2$: their keypoints matchesMask: Inlier mask returned in cv$2.$findHomography$()$matchesMask $=$ matchesMask.ravel$().$tolist$()$ singlePointColor $=$ None, flags $=2)$plt$.$imshow$($img$3,$ 'gray'$),$plt$.$show$()$

The provided original draw$\_$outline and draw$\_$inliers in the second image