Related
I'm writing a function in matlab which mimics the built-in 'imwarp' function (applying geometric transformation) without using any kind of loops. i'm in the final step when i have to call my function for bi-linear interpolation for every index in final 2D image.
I have 3 arrays here , 'pts' have homogenized vectors (x,y,1) for which i interpolate and 'row' and 'cols' have x and y coordinates respectively for resultant image where interpolated intensity value would be placed.
finalImage (rows(1,:),cols(1,:))=bilinear(pts(:,:),im);
Kindly correct my syntax here to do it properly. thanks in advance.
The following is a simple implementation of applying an affine transformation to an image. Some of the matrices may be reversed because I did this from memory. I don't know exactly how you are formatting your pts array so I figure a working example is the best I can do. The interp2 function applies bilinear interpolation, the bilinear function performs the bilinear transform which describes analog filters as digital filters. This is not what you want.
P.S. You have to make sure to use the inverse transform when applying image warping (that is, define the point you want to sample in the input image for each point in the output image). If you perform the forward transform (i.e. define the point in the output image that each point in the input image maps to) then you will end up with some serious aliasing effects and potentially holes in the output image.
Hope this helps. Let me know if you have questions.
img = double(imread('rice.png'))/255;
theta = 30; % rotate 30 degrees
R = [cosd(theta) -sind(theta) 0; ...
sind(theta) cosd(theta) 0; ...
0 0 1];
sx = 15; % skew by 15 degrees in x
Skx = [1 tand(sx) 0; ...
0 1 0; ...
0 0 1];
% Translate by 1/2 size of image
tx = -size(img, 2)/2;
ty = -size(img, 1)/2;
T = [1 0 tx; ...
0 1 ty; ...
0 0 1];
% Scale image down by 1/2
sx = 0.5;
sy = 0.5;
S = [sx 0 0; ...
0 sy 0; ...
0 0 1];
% translate, scale, rotate, skew, then translate back
A = inv(T)*Skx*R*S*T;
% create meshgrid points
[x, y] = meshgrid(1:size(img,2), 1:size(img,1));
% reshape so we can apply matrix op
V = [reshape(x, 1, []); reshape(y, 1, []); ones(1, numel(x))];
Vq = inv(A)*V;
% probably not necessary for these transformations but project back to the z=1 plane
Vq(1,:) = Vq(1,:) ./ V(3,:);
Vq(2,:) = Vq(2,:) ./ V(3,:);
% reshape back into a meshgrid
xq = reshape(Vq(1,:), size(img));
yq = reshape(Vq(2,:), size(img));
% use interp2 to perform bilinear interpolation
imgnew = interp2(x, y, img, xq, yq);
% show the resulting image
imshow(imgnew);
I have an array of pixel data frames for use with VideoWriter. I want to overlay lineseries/contour objects into each frame. I don't want to make the movie by iteratively drawing each frame to a figure and capturing it with getframe, because that gives poor resolution and is slow. I tried using getframe on a plot of just the contour, but that returns images scaled to the wrong size with weird margins, especially when using 'axis equal,' which I need.
Updated to accommodate feedback from OP
Getting the contour data as pixel data is not trivial (if possible at all) since using getframe doesn't yield predictable results
What we can do is to extract the contour data and then overlay it on the pixel data frames, forcing them to be to the same scale and then do a getframe on the resultant merged image. This will at least ensure that they two data sets area aligned.
The following code shows the principle though you'd need to modify it for your own needs:
%% Generate some random contours to use
x = linspace(-2*pi,2*pi);
y = linspace(0,4*pi);
[X,Y] = meshgrid(x,y);
Z = sin(X)+cos(Y);
[~,h] = contour(X,Y,Z);
This yields the following contours
Now we get the handles of the children of this image. These will all be 'patch' type objects
patches = get(h,'Children');
Also get the axis limits for the contours
lims = axis;
Next, create a new figure and render the pixel frame data into it
In this example I'm just loading an image but you get the idea.
%% Render frame data
figure
i = imread( some_image_file_png );
This image is actually 194 x 259 x 3. I can display it and rescale the X and Y axes using
%% Set image axes
image(flipdim(i,1),'XData',[lims(1) lims(2)],'YData',[lims(4) lims(3)]);
Note the use of flipdim() to vertically flip the image since the image Y-axis runs in the opposite sense to the contour Y axis. This gives me:
Now I can plot the contours (patches) form the contour plot over the top of the image in the same coordinate space
%% Plot patches
for p =1:length(patches)
xd = get( patches(p), 'XData' );
yd = get( patches(p), 'YData' );
% This causes all contours to be rendered in white. You may
% want to play with this a little
cd = zeros(size(xd));
patch( xd, yd, cd, 'EdgeColor', 'w');
end
This yields
You can now use getframe to extract the frame. If it's important to have coloured contours, you will need to extract colour data from the original contour map and use it to apply an appropriate colouring in the overlaid image.
As a short cut, it's also possible to compile all patch data into a single MxN matrix and render with a single call to patch but I wrote it this way to demonstrate the process.
Well, here's a Bresenham-esque solution based on the ContourMatrix. Not ideal cuz doesn't handle line width, antialiasing, or any more than a single color. But it's pretty efficient (not quite Bham efficient).
function renderContour
clc
close all
x = randn(100,70);
[c,h] = contour(x,[0 0],'LineColor','r');
axis equal
if ~isnumeric(h.LineColor)
error('not handled')
end
cs = nan(size(c,2),4);
k = 0;
ci = 1;
for i = 1:size(c,2)
if k <= 0
k = c(2,i);
else
if k > 1
cs(ci,:) = reshape(c(:,i+[0 1]),[1 4]);
ci = ci + 1;
end
k = k - 1;
end
end
pix = renderLines(cs(1:ci-1,:),[1 1;fliplr(size(x))],10,h.LineColor);
figure
image(pix)
axis equal
end
function out = renderLines(cs,rect,res,color)
% cs = [x1(:) y1(:) x2(:) y2(:)]
% rect = [x(1) y(1);x(2) y(2)]
% doesnt handle line width, antialiasing, etc
% could do those with imdilate, imfilter, etc.
test = false;
if test
if false
cs = [0 0 5 5; 0 5 2.5 2.5];
rect = [0 0; 10 10];
else
cs = 100 * randn(1000,4);
rect = 200 * randn(2);
end
res = 10;
color = [1 .5 0];
end
out = nan(abs(res * round(diff(fliplr(rect)))));
cs = cs - repmat(min(rect),[size(cs,1) 2]);
d = [cs(:,1) - cs(:,3) cs(:,2) - cs(:,4)];
lens = sqrt(sum(d.^2,2));
for i = 1:size(cs,1)
n = ceil(sqrt(2) * res * lens(i));
if false % equivalent but probably less efficient
pts = linspace(0,1,n);
pts = round(res * (repmat(cs(i,1:2),[length(pts) 1]) - pts' * d(i,:)));
else
pts = round(res * [linspace(cs(i,1),cs(i,3),n);linspace(cs(i,2),cs(i,4),n)]');
end
pts = pts(all(pts > 0 & pts <= repmat(fliplr(size(out)),[size(pts,1) 1]),2),:);
out(sub2ind(size(out),pts(:,2),pts(:,1))) = 1;
end
out = repmat(flipud(out),[1 1 3]) .* repmat(permute(color,[3 1 2]),size(out));
if test
image(out)
axis equal
end
end
I am trying to rotate an image with Matlab without using imrotate function. I actually made it by using transformation matrix.But it is not good enough.The problem is, the rotated image is "sliding".Let me tell you with pictures.
This is my image which I want to rotate:
But when I rotate it ,for example 45 degrees, it becomes this:
I am asking why this is happening.Here is my code,is there any mathematical or programming mistakes about it?
image=torso;
%image padding
[Rows, Cols] = size(image);
Diagonal = sqrt(Rows^2 + Cols^2);
RowPad = ceil(Diagonal - Rows) + 2;
ColPad = ceil(Diagonal - Cols) + 2;
imagepad = zeros(Rows+RowPad, Cols+ColPad);
imagepad(ceil(RowPad/2):(ceil(RowPad/2)+Rows-1),ceil(ColPad/2):(ceil(ColPad/2)+Cols-1)) = image;
degree=45;
%midpoints
midx=ceil((size(imagepad,1)+1)/2);
midy=ceil((size(imagepad,2)+1)/2);
imagerot=zeros(size(imagepad));
%rotation
for i=1:size(imagepad,1)
for j=1:size(imagepad,2)
x=(i-midx)*cos(degree)-(j-midy)*sin(degree);
y=(i-midx)*sin(degree)+(j-midy)*cos(degree);
x=round(x)+midx;
y=round(y)+midy;
if (x>=1 && y>=1)
imagerot(x,y)=imagepad(i,j); % k degrees rotated image
end
end
end
figure,imagesc(imagerot);
colormap(gray(256));
The reason you have holes in your image is because you are computing the location in imagerot of each pixel in imagepad. You need to do the computation the other way around. That is, for each pixel in imagerot interpolate in imagepad. To do this, you just need to apply the inverse transform, which in the case of a rotation matrix is just the transpose of the matrix (just change the sign on each sin and translate the other way).
Loop over pixels in imagerot:
imagerot=zeros(size(imagepad)); % midx and midy same for both
for i=1:size(imagerot,1)
for j=1:size(imagerot,2)
x= (i-midx)*cos(rads)+(j-midy)*sin(rads);
y=-(i-midx)*sin(rads)+(j-midy)*cos(rads);
x=round(x)+midx;
y=round(y)+midy;
if (x>=1 && y>=1 && x<=size(imagepad,2) && y<=size(imagepad,1))
imagerot(i,j)=imagepad(x,y); % k degrees rotated image
end
end
end
Also note that your midx and midy need to be calculated with size(imagepad,2) and size(imagepad,1) respectively, since the first dimension refers to the number of rows (height) and the second to width.
NOTE: The same approach applies when you decide to adopt an interpolation scheme other than nearest neighbor, as in Rody's example with linear interpolation.
EDIT: I'm assuming you are using a loop for demonstrative purposes, but in practice there is no need for loops. Here's an example of nearest neighbor interpolation (what you are using), keeping the same size image, but you can modify this to produce a larger image that includes the whole source image:
imagepad = imread('peppers.png');
[nrows ncols nslices] = size(imagepad);
midx=ceil((ncols+1)/2);
midy=ceil((nrows+1)/2);
Mr = [cos(pi/4) sin(pi/4); -sin(pi/4) cos(pi/4)]; % e.g. 45 degree rotation
% rotate about center
[X Y] = meshgrid(1:ncols,1:nrows);
XYt = [X(:)-midx Y(:)-midy]*Mr;
XYt = bsxfun(#plus,XYt,[midx midy]);
xout = round(XYt(:,1)); yout = round(XYt(:,2)); % nearest neighbor!
outbound = yout<1 | yout>nrows | xout<1 | xout>ncols;
zout=repmat(cat(3,1,2,3),nrows,ncols,1); zout=zout(:);
xout(xout<1) = 1; xout(xout>ncols) = ncols;
yout(yout<1) = 1; yout(yout>nrows) = nrows;
xout = repmat(xout,[3 1]); yout = repmat(yout,[3 1]);
imagerot = imagepad(sub2ind(size(imagepad),yout,xout,zout(:))); % lookup
imagerot = reshape(imagerot,size(imagepad));
imagerot(repmat(outbound,[1 1 3])) = 0; % set background value to [0 0 0] (black)
To modify the above to linear interpolation, compute the 4 neighboring pixels to each coordinate in XYt and perform a weighted sum using the fractional components product as the weights. I'll leave that as an exercise, since it would only serve to bloat my answer further beyond the scope of your question. :)
The method you are using (rotate by sampling) is the fastest and simplest, but also the least accurate.
Rotation by area mapping, as given below (this is a good reference), is much better at preserving color.
But: note that this will only work on greyscale/RGB images, but NOT on colormapped images like the one you seem to be using.
image = imread('peppers.png');
figure(1), clf, hold on
subplot(1,2,1)
imshow(image);
degree = 45;
switch mod(degree, 360)
% Special cases
case 0
imagerot = image;
case 90
imagerot = rot90(image);
case 180
imagerot = image(end:-1:1, end:-1:1);
case 270
imagerot = rot90(image(end:-1:1, end:-1:1));
% General rotations
otherwise
% Convert to radians and create transformation matrix
a = degree*pi/180;
R = [+cos(a) +sin(a); -sin(a) +cos(a)];
% Figure out the size of the transformed image
[m,n,p] = size(image);
dest = round( [1 1; 1 n; m 1; m n]*R );
dest = bsxfun(#minus, dest, min(dest)) + 1;
imagerot = zeros([max(dest) p],class(image));
% Map all pixels of the transformed image to the original image
for ii = 1:size(imagerot,1)
for jj = 1:size(imagerot,2)
source = ([ii jj]-dest(1,:))*R.';
if all(source >= 1) && all(source <= [m n])
% Get all 4 surrounding pixels
C = ceil(source);
F = floor(source);
% Compute the relative areas
A = [...
((C(2)-source(2))*(C(1)-source(1))),...
((source(2)-F(2))*(source(1)-F(1)));
((C(2)-source(2))*(source(1)-F(1))),...
((source(2)-F(2))*(C(1)-source(1)))];
% Extract colors and re-scale them relative to area
cols = bsxfun(#times, A, double(image(F(1):C(1),F(2):C(2),:)));
% Assign
imagerot(ii,jj,:) = sum(sum(cols),2);
end
end
end
end
subplot(1,2,2)
imshow(imagerot);
Output:
Rotates colored image according to angle given by user without any cropping of image in matlab.
Output of this program is similar to output of inbuilt command "imrotate" .This program dynamically creates background according to angle input given by user.By using rotation matrix and origin shifting, we get relation between coordinates of initial and final image.Using relation between coordinates of initial and final image, we now map the intensity values for each pixel.
img=imread('img.jpg');
[rowsi,colsi,z]= size(img);
angle=45;
rads=2*pi*angle/360;
%calculating array dimesions such that rotated image gets fit in it exactly.
% we are using absolute so that we get positve value in any case ie.,any quadrant.
rowsf=ceil(rowsi*abs(cos(rads))+colsi*abs(sin(rads)));
colsf=ceil(rowsi*abs(sin(rads))+colsi*abs(cos(rads)));
% define an array withcalculated dimensionsand fill the array with zeros ie.,black
C=uint8(zeros([rowsf colsf 3 ]));
%calculating center of original and final image
xo=ceil(rowsi/2);
yo=ceil(colsi/2);
midx=ceil((size(C,1))/2);
midy=ceil((size(C,2))/2);
% in this loop we calculate corresponding coordinates of pixel of A
% for each pixel of C, and its intensity will be assigned after checking
% weather it lie in the bound of A (original image)
for i=1:size(C,1)
for j=1:size(C,2)
x= (i-midx)*cos(rads)+(j-midy)*sin(rads);
y= -(i-midx)*sin(rads)+(j-midy)*cos(rads);
x=round(x)+xo;
y=round(y)+yo;
if (x>=1 && y>=1 && x<=size(img,1) && y<=size(img,2) )
C(i,j,:)=img(x,y,:);
end
end
end
imshow(C);
Check this out.
this is fastest way that you can do.
img = imread('Koala.jpg');
theta = pi/10;
rmat = [
cos(theta) sin(theta) 0
-sin(theta) cos(theta) 0
0 0 1];
mx = size(img,2);
my = size(img,1);
corners = [
0 0 1
mx 0 1
0 my 1
mx my 1];
new_c = corners*rmat;
T = maketform('affine', rmat); %# represents translation
img2 = imtransform(img, T, ...
'XData',[min(new_c(:,1)) max(new_c(:,1))],...
'YData',[min(new_c(:,2)) max(new_c(:,2))]);
subplot(121), imshow(img);
subplot(122), imshow(img2);
I am trying to rotate an image with Matlab without using imrotate function. I actually made it by using transformation matrix.But it is not good enough.The problem is, the rotated image is "sliding".Let me tell you with pictures.
This is my image which I want to rotate:
But when I rotate it ,for example 45 degrees, it becomes this:
I am asking why this is happening.Here is my code,is there any mathematical or programming mistakes about it?
image=torso;
%image padding
[Rows, Cols] = size(image);
Diagonal = sqrt(Rows^2 + Cols^2);
RowPad = ceil(Diagonal - Rows) + 2;
ColPad = ceil(Diagonal - Cols) + 2;
imagepad = zeros(Rows+RowPad, Cols+ColPad);
imagepad(ceil(RowPad/2):(ceil(RowPad/2)+Rows-1),ceil(ColPad/2):(ceil(ColPad/2)+Cols-1)) = image;
degree=45;
%midpoints
midx=ceil((size(imagepad,1)+1)/2);
midy=ceil((size(imagepad,2)+1)/2);
imagerot=zeros(size(imagepad));
%rotation
for i=1:size(imagepad,1)
for j=1:size(imagepad,2)
x=(i-midx)*cos(degree)-(j-midy)*sin(degree);
y=(i-midx)*sin(degree)+(j-midy)*cos(degree);
x=round(x)+midx;
y=round(y)+midy;
if (x>=1 && y>=1)
imagerot(x,y)=imagepad(i,j); % k degrees rotated image
end
end
end
figure,imagesc(imagerot);
colormap(gray(256));
The reason you have holes in your image is because you are computing the location in imagerot of each pixel in imagepad. You need to do the computation the other way around. That is, for each pixel in imagerot interpolate in imagepad. To do this, you just need to apply the inverse transform, which in the case of a rotation matrix is just the transpose of the matrix (just change the sign on each sin and translate the other way).
Loop over pixels in imagerot:
imagerot=zeros(size(imagepad)); % midx and midy same for both
for i=1:size(imagerot,1)
for j=1:size(imagerot,2)
x= (i-midx)*cos(rads)+(j-midy)*sin(rads);
y=-(i-midx)*sin(rads)+(j-midy)*cos(rads);
x=round(x)+midx;
y=round(y)+midy;
if (x>=1 && y>=1 && x<=size(imagepad,2) && y<=size(imagepad,1))
imagerot(i,j)=imagepad(x,y); % k degrees rotated image
end
end
end
Also note that your midx and midy need to be calculated with size(imagepad,2) and size(imagepad,1) respectively, since the first dimension refers to the number of rows (height) and the second to width.
NOTE: The same approach applies when you decide to adopt an interpolation scheme other than nearest neighbor, as in Rody's example with linear interpolation.
EDIT: I'm assuming you are using a loop for demonstrative purposes, but in practice there is no need for loops. Here's an example of nearest neighbor interpolation (what you are using), keeping the same size image, but you can modify this to produce a larger image that includes the whole source image:
imagepad = imread('peppers.png');
[nrows ncols nslices] = size(imagepad);
midx=ceil((ncols+1)/2);
midy=ceil((nrows+1)/2);
Mr = [cos(pi/4) sin(pi/4); -sin(pi/4) cos(pi/4)]; % e.g. 45 degree rotation
% rotate about center
[X Y] = meshgrid(1:ncols,1:nrows);
XYt = [X(:)-midx Y(:)-midy]*Mr;
XYt = bsxfun(#plus,XYt,[midx midy]);
xout = round(XYt(:,1)); yout = round(XYt(:,2)); % nearest neighbor!
outbound = yout<1 | yout>nrows | xout<1 | xout>ncols;
zout=repmat(cat(3,1,2,3),nrows,ncols,1); zout=zout(:);
xout(xout<1) = 1; xout(xout>ncols) = ncols;
yout(yout<1) = 1; yout(yout>nrows) = nrows;
xout = repmat(xout,[3 1]); yout = repmat(yout,[3 1]);
imagerot = imagepad(sub2ind(size(imagepad),yout,xout,zout(:))); % lookup
imagerot = reshape(imagerot,size(imagepad));
imagerot(repmat(outbound,[1 1 3])) = 0; % set background value to [0 0 0] (black)
To modify the above to linear interpolation, compute the 4 neighboring pixels to each coordinate in XYt and perform a weighted sum using the fractional components product as the weights. I'll leave that as an exercise, since it would only serve to bloat my answer further beyond the scope of your question. :)
The method you are using (rotate by sampling) is the fastest and simplest, but also the least accurate.
Rotation by area mapping, as given below (this is a good reference), is much better at preserving color.
But: note that this will only work on greyscale/RGB images, but NOT on colormapped images like the one you seem to be using.
image = imread('peppers.png');
figure(1), clf, hold on
subplot(1,2,1)
imshow(image);
degree = 45;
switch mod(degree, 360)
% Special cases
case 0
imagerot = image;
case 90
imagerot = rot90(image);
case 180
imagerot = image(end:-1:1, end:-1:1);
case 270
imagerot = rot90(image(end:-1:1, end:-1:1));
% General rotations
otherwise
% Convert to radians and create transformation matrix
a = degree*pi/180;
R = [+cos(a) +sin(a); -sin(a) +cos(a)];
% Figure out the size of the transformed image
[m,n,p] = size(image);
dest = round( [1 1; 1 n; m 1; m n]*R );
dest = bsxfun(#minus, dest, min(dest)) + 1;
imagerot = zeros([max(dest) p],class(image));
% Map all pixels of the transformed image to the original image
for ii = 1:size(imagerot,1)
for jj = 1:size(imagerot,2)
source = ([ii jj]-dest(1,:))*R.';
if all(source >= 1) && all(source <= [m n])
% Get all 4 surrounding pixels
C = ceil(source);
F = floor(source);
% Compute the relative areas
A = [...
((C(2)-source(2))*(C(1)-source(1))),...
((source(2)-F(2))*(source(1)-F(1)));
((C(2)-source(2))*(source(1)-F(1))),...
((source(2)-F(2))*(C(1)-source(1)))];
% Extract colors and re-scale them relative to area
cols = bsxfun(#times, A, double(image(F(1):C(1),F(2):C(2),:)));
% Assign
imagerot(ii,jj,:) = sum(sum(cols),2);
end
end
end
end
subplot(1,2,2)
imshow(imagerot);
Output:
Rotates colored image according to angle given by user without any cropping of image in matlab.
Output of this program is similar to output of inbuilt command "imrotate" .This program dynamically creates background according to angle input given by user.By using rotation matrix and origin shifting, we get relation between coordinates of initial and final image.Using relation between coordinates of initial and final image, we now map the intensity values for each pixel.
img=imread('img.jpg');
[rowsi,colsi,z]= size(img);
angle=45;
rads=2*pi*angle/360;
%calculating array dimesions such that rotated image gets fit in it exactly.
% we are using absolute so that we get positve value in any case ie.,any quadrant.
rowsf=ceil(rowsi*abs(cos(rads))+colsi*abs(sin(rads)));
colsf=ceil(rowsi*abs(sin(rads))+colsi*abs(cos(rads)));
% define an array withcalculated dimensionsand fill the array with zeros ie.,black
C=uint8(zeros([rowsf colsf 3 ]));
%calculating center of original and final image
xo=ceil(rowsi/2);
yo=ceil(colsi/2);
midx=ceil((size(C,1))/2);
midy=ceil((size(C,2))/2);
% in this loop we calculate corresponding coordinates of pixel of A
% for each pixel of C, and its intensity will be assigned after checking
% weather it lie in the bound of A (original image)
for i=1:size(C,1)
for j=1:size(C,2)
x= (i-midx)*cos(rads)+(j-midy)*sin(rads);
y= -(i-midx)*sin(rads)+(j-midy)*cos(rads);
x=round(x)+xo;
y=round(y)+yo;
if (x>=1 && y>=1 && x<=size(img,1) && y<=size(img,2) )
C(i,j,:)=img(x,y,:);
end
end
end
imshow(C);
Check this out.
this is fastest way that you can do.
img = imread('Koala.jpg');
theta = pi/10;
rmat = [
cos(theta) sin(theta) 0
-sin(theta) cos(theta) 0
0 0 1];
mx = size(img,2);
my = size(img,1);
corners = [
0 0 1
mx 0 1
0 my 1
mx my 1];
new_c = corners*rmat;
T = maketform('affine', rmat); %# represents translation
img2 = imtransform(img, T, ...
'XData',[min(new_c(:,1)) max(new_c(:,1))],...
'YData',[min(new_c(:,2)) max(new_c(:,2))]);
subplot(121), imshow(img);
subplot(122), imshow(img2);
I am trying to transform an image using a 3D transformation matrix and assuming my camera is orthonormal.
I am defining my homography using the plane-induced homography formula H=R-t*n'/d (with d=Inf so H=R) as given in Hartley and Zisserman Chapter 13.
What I am confused about is when I use a rather modest rotation, the image seems to be distorting much more than I expect (I'm sure I'm not confounding radians and degrees).
What could be going wrong here?
I've attached my code and example output.
n = [0;0;-1];
d = Inf;
im = imread('cameraman.tif');
rotations = [0 0.01 0.1 1 10];
for ind = 1:length(rotations)
theta = rotations(ind)*pi/180;
R = [ 1 0 0 ;
0 cos(theta) -sin(theta);
0 sin(theta) cos(theta)];
t = [0;0;0];
H = R-t*n'/d;
tform = maketform('projective',H');
imT = imtransform(im,tform);
subplot(1,5,ind) ;
imshow(imT)
title(['Rot=' num2str(rotations(ind)) 'deg']);
axis square
end
The formula H = R-t*n'/d has one assumption which is not met in your case:
This formula implies that you are using pinhole camera model with focal length=1
But in your case, for your camera to be more real and for your code to work, you should set the focal length to some positive number much greater than 1. (focal length is the distance from your camera center to the image plane)
To do this you can define a calibration matrix K which handles the focal length. You just need to change your formula to
H=K R inv(K) - 1/d K t n' inv(K)
in which K is a 3-by-3 identity matrix whose two first elements along the diagonal are set to the focal length (e.g. f=300). The formula can be easily derived if you assume a projective camera.
Below is the corrected version of your code, in which the angles make sense.
n = [0;0;-1];
d = Inf;
im = imread('cameraman.tif');
rotations = [0 0.01 0.1 30 60];
for ind = 1:length(rotations)
theta = rotations(ind)*pi/180;
R = [ 1 0 0 ;
0 cos(theta) -sin(theta);
0 sin(theta) cos(theta)];
t = [0;0;0];
K=[300 0 0;
0 300 0;
0 0 1];
H=K*R/K-1/d*K*t*n'/K;
tform = maketform('projective',H');
imT = imtransform(im,tform);
subplot(1,5,ind) ;
imshow(imT)
title(['Rot=' num2str(rotations(ind)) 'deg']);
axis square
end
You can see the result in the image below:
You can also rotate the image around its center. For it to happen you should set the image plane origin to the center of the image which I think is not possible with that method of matlab (maketform).
You can use the method below instead.
imT=imagehomog(im,H','c');
Note that if you use this method, you'll have to change some settings in n, d, t and R to get the appropriate result.
That method can be found at: https://github.com/covarep/covarep/blob/master/external/voicebox/imagehomog.m
The result of the program with imagehomog and some changes in n, d, t , and R is shown below which seems more real.
New settings are:
n = [0 0 1]';
d = 2;
t = [1 0 0]';
R = [cos(theta), 0, sin(theta);
0, 1, 0;
-sin(theta), 0, cos(theta)];
Hmm... I'm not 100% percent on this stuff, but it was an interesting question and relevant to my work, so I thought I'd play around and give it a shot.
EDIT: I tried this once using no built-ins. That was my original answer. Then I realized that you could do it your way pretty easily:
The easy answer to your question is to use the correct rotation matrix about the z-axis:
R = [cos(theta) -sin(theta) 0;
sin(theta) cos(theta) 0;
0 0 1];
Here's another way to do it (my original answer):
I'm going to share what I did; hopefully this is useful to you. I only did it in 2D (though that should be easy to expand to 3D). Note that if you want to rotate the image in plane, you will need to use a different rotation matrix that you have currently coded. You need to rotate about the Z-axis.
I did not use those matlab built-ins.
I referred to http://en.wikipedia.org/wiki/Rotation_matrix for some info.
im = double(imread('cameraman.tif')); % must be double for interpn
[x y] = ndgrid(1:size(im,1), 1:size(im,2));
rotation = 10;
theta = rotation*pi/180;
% calculate rotation matrix
R = [ cos(theta) -sin(theta);
sin(theta) cos(theta)]; % just 2D case
% calculate new positions of image indicies
tmp = R*[x(:)' ; y(:)']; % 2 by numel(im)
xi = reshape(tmp(1,:),size(x)); % new x-indicies
yi = reshape(tmp(2,:),size(y)); % new y-indicies
imrot = interpn(x,y,im,xi,yi); % interpolate from old->new indicies
imagesc(imrot);
My own question now is: "How do you change the origin about which you are rotating the image? Clearly, I'm rotating about (0,0), the top left corner.
EDIT 2 In response to the asker's comment, I've tried again.
This time I fixed a couple of things. Now I'm using the same transformation matrix (about x) as in the original question.
I rotated about the center of the image by redoing the way i do the ndgrids (put 0,0,0) in the center of the image. I also decided to show 3 planes of the image. This was not in the original question. The middle plane is the plane of interest. To get just the middle plane, you can leave out the zero-padding and redefine the 3rd ndgrid option to be just 1 instead of -1:1.
im = double(imread('cameraman.tif')); % must be double for interpn
im = padarray(im, [0 0 1],'both');
[x y z] = ndgrid(-floor(size(im,1)/2):floor(size(im,1)/2)-1, ...
-floor(size(im,2)/2):floor(size(im,2)/2)-1,...
-1:1);
rotation = 1;
theta = rotation*pi/180;
% calculate rotation matrix
R = [ 1 0 0 ;
0 cos(theta) -sin(theta);
0 sin(theta) cos(theta)];
% calculate new positions of image indicies
tmp = R*[x(:)'; y(:)'; z(:)']; % 2 by numel(im)
xi = reshape(tmp(1,:),size(x)); % new x-indicies
yi = reshape(tmp(2,:),size(y)); % new y-indicies
zi = reshape(tmp(3,:),size(z));
imrot = interpn(x,y,z,im,xi,yi,zi); % interpolate from old->new indicies
figure;
subplot(3,1,1);imagesc(imrot(:,:,1)); axis image; axis off;
subplot(3,1,2);imagesc(imrot(:,:,2)); axis image; axis off;
subplot(3,1,3);imagesc(imrot(:,:,3)); axis image; axis off;
You are performing rotations around the x-axis: in your matrix, the 1st component (x) is left unchanged by the rotation matrix. This is confirmed by the perspective deformations from your examples.
The actual amount of deformation will then depend on the distance between the camera and the image plane (or more accurately on its value relative to the focal length of the camera). It can be important when the cameraman image plane is located near the camera.