110 likes | 241 Views
Week 1 Report Ruben Villegas. Lucas- Kanade Optical Flow. Problems I had Getting used to Matlab Ax = fx(i -1: i +1, j -1: j +1) I f matrix Ax = [1 2 3; 4 5 6; 7 8 9], Ax(:) == [1 4 7 2 5 8 3 6 9]. Solution Ax = fx (i-1:i+1;j-1:j+1)’, where
E N D
Lucas-Kanade Optical Flow • Problems I had • Getting used to Matlab • Ax = fx(i-1:i+1,j-1:j+1) • If matrix Ax = [1 2 3; 4 5 6; 7 8 9], Ax(:) == [1 4 7 2 5 8 3 6 9]. • Solution Ax = fx(i-1:i+1;j-1:j+1)’, where Ax = [1 4 7, 2 5 8, 3 6 9], and A(:) == [1 2 3 4 5 6 7 8 9] • pinvvsinv for U (A’A )-1A’ft
Functions Implemented • opticalFlow • Takes 2 images and outputs an optical flow image. • opticalFlow2 • Failed attempt to implement pyramids. • Ft calculation problems. • opticalFlow3 • Takes a sequence and outputs a video of the optical flow of that video
SIFT Descriptor • Problems I had • Figure out how to use 3-Dimensional Matrixes • Split 16x16 matrixes into 4x4 • uint8 underflow? • m(y-1,x-1) = sqrt(double((im(y+1,x)-im(y-1,x))^2+(im(y,x+1)-im(y,x-1))^2)) • Solution, conver im to double beforehand.