5. Jia estimated the fundamental matrix F. She then computed the epipolar lines, but found that they don't intersect at

[answerhappygod]

5 Jia Estimated The Fundamental Matrix F She Then Computed The Epipolar Lines But Found That They Don T Intersect At 1 (151.22 KiB) Viewed 30 times

5. Jia estimated the fundamental matrix F. She then computed the epipolar lines, but found that they don't intersect at a single point (see below). Why might that be? How can she correct this? 6. Suppose you know the fundamental matrix F. (a) Explain briefly how you can use F to remove some bad correspondences (including ones not shown in the figure). Assume that p and q are the homogenous coordinates of a pair of possibly corresponding points in the first and second image respectively. Write down the equation they must satisfy. (b) Can you remove all bad correspondences? Explain why / why not. 7. Suppose F is not known and must be estimated. (a) A single accurate pair of corresponding pixels (x,x) leads to a constraint on the entries of F. Is this constraint linear? What is the coefficient of F23 in this constraint? (b) Multiple accurate correspondences (X,x!:)–, lead to a set of constraints. How will you solve for F given these constraints? (c) What is the minimum number of correspondences needed to estimate F? (d) What is the rank of F? How will you ensure that F satisfies this rank constraint? (e) What technique will you use to make this estimation resilient to outlier correspondences?