This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Massachusetts Institute of Technology Department of Computer Science and Electrical Engineering 6.801/6.866 Machine Vision QUIZ II Handed out: 2004 Dec. 2nd Due on: 2004 Dec. 9th Problem 1: Suppose we allow for a scale factor in the problem of absolute orientationin addition to translation and rotationso that the transformation from the left to the right coordinate systems becomes r r = sR( r l ) + r Then the absolute orientation problem becomes one of minimizing n r r,i sR( r l,i ) r 2 i = 1 (a) By differentiating w.r.t. to r , dividing by n , and setting the result equal to zero, find a formula for the best fit translation involving the centroids r l and r r of the two sets of measurements (b) By shifting the origin of each set of measurements to its centroid, show that the translation r drops out and the error to be minimized simplifies to n r r,i sR( r l,i ) 2 i = 1 where r l,i = r l,i r l and r r,i = r r,i r r . (c) Show that this can be written in the form S r 2sD rl + s 2 S l where S r and S l are sums of squares of the lengths r r,i and r l,i respectively, while D rl is the sum of dotproducts of r r,i and R( r l,i ) . (c) Conclude that the best fit scale is given by s = D rl /S l (d) Now suppose we instead find the best fit transformation r l = s R ( r r ) + r from right to left coordinate system. We might expect that this trans formation is the inverse of the earlier one, that is, s = 1/s , R = R 1 , and r = (1/s)R 1 ( r ) . Unfortunately this is not so. In particular show that s = D lr /S r where D lr is the sum...
View Full
Document
 Spring '04
 BertholdHorn

Click to edit the document details