I want to project the texture of 3D surface (CylCoors
300000x3) into a 2D plane (Image
380x360). For doing so I take every unique value in Z (UniqueZ=unique(CylCoors(:,3))
) and and Theta (UniqueTheta=unique(CylCoors(:,1))
) and project all the texture values (PointValues
300000x1) where both meet like this:
Image=zeros(max(UniqueH),max(UniqueTheta)); %380x360
tic
HMat=bsxfun(@eq,CylCoors(:,3),UniqueH'); % 300000x380
ThetaMat=bsxfun(@eq,CylCoors(:,1),UniqueTheta'); %300000x360
for ii=1:length(UniqueH) % Sloooow and not nice :(
for jj=1:length(UniqueTheta)
Image(ii,jj)=sum(PointValues.*...
HMat(:,ii).*ThetaMat(:,jj))/...
sum(HMat(:,ii).*ThetaMat(:,jj));
end
end
toc
Here's an example with trimmed variables:
CylCoorsSample = [263.0000 184.2586 10.0000
264.0000 183.0417 10.0000
264.0000 182.1572 10.0000
82.0000 157.4746 11.0000
80.0000 158.2348 11.0000
86.0000 157.3507 11.0000
84.0000 157.7633 11.0000]
PointValuesSample = [0.4745
0.5098
0.5020
0.4784
0.4510
0.4431
0.5804]
UniqueTheta = [80
82
84
86
263
264]
UniqueH =[10
11]
ThetaMat = HMat =
0 0 0 0 1 0 1 0
0 0 0 0 0 1 1 0
0 0 0 0 0 1 1 0
0 1 0 0 0 0 0 1
1 0 0 0 0 0 0 1
0 0 0 1 0 0 0 1
0 0 1 0 0 0 0 1
Image = % size: length(UniqueH)xlength(UniqueTheta)
NaN NaN NaN NaN 0.4745 0.5059
0.4510 0.4784 0.5804 0.4431 NaN NaN
Is there a way to vectorize the for
loops using accumarray
? Something tells me this is the case, but I find the multidimensional sub indexing for this function quite confusing.
Any other vectorization solutions (using smart combinations of reshape
,bsxfun
, and sum
I assume) are also welcome, but I would really like to understand this use of accumarray
a bit better.
Given:
You can use
accumarray
as follows (using the third output ofunique
to generate the requiredsubs
input):