convert 2d image to 3d

my requirement is I have to read two images and should compare the two images in terms of deformation.let me elaborate my question so that I can get a brief answer.For example there are two images say car I want to find the difference in terms of deformation.

 I tried    
  a=imread();
  surf(double(a));
  shading flat;
  but i am getting Warning: Matrix dimensions must agree, not rendering mesh

[Adam]

Is this supposed to be a comment on another question? If it is a question in its own right it needs a bit more information since you clearly seem to be responding to an answer by 'Chad' to some previous question.

[sivaramakrishna]

yes Adam it is a comment to another question...But it is elaborated so I thought of keeping it as a new question

Answer 1

[Image Analyst]

You can't use surf() with a color image. Extract just one color channel or use rgb2gray() on it to convert (the badly-named) "a" to gray scale.

Answer 2

[Mike Garrity]

I'm just guessing, but you would get that error message if your image was truecolor. In that case, a would be a MxNx3 variable with red, green, and blue values for each pixel. The surf command wants to display a single value as the height. It doesn't know what to do with 3 values.

If that's the case, then you need to convert the RGB triplets into a scalar value. For example:

surf(rgb2gray(a,3))

Once you've done that, you can use the RGB values as CData on the surface, just not as ZData. So you could do something like this:

a = imread('street1.jpg');
surf(rgb2gray(a,3),a,'EdgeColor','none')

Which would let you see the colors of your image on the surface, like this:

[sivaramakrishna]

Hi Mike,This answer is very useful for me.Could you please elaborate so that I can use according to my usage "compare two 3d images in terms of deformation". It is throwing an error rgb2gray sholud have only one argument

[Mike Garrity]

Well, I don't have a lot of info about what you're doing, but my guess was that you were trying to create a surface for each of two different images so that you could see the differences in where the bright & dark areas are. If so, it would look something like this:

img1 = imread('street1.jpg');
img2 = imread('street2.jpg');
surf(rgb2gray(img1),'EdgeColor','none','FaceColor','red');
hold on
surf(rgb2gray(img2),'EdgeColor','none','FaceColor','blue');
camlight

The idea is that each of the two calls to rgb2gray turns one of the RGB images into a scalar. Then the two calls to surf each turn one of these 2D scalar images into a surface.

I don't know whether I think it's a very effective visualization though. It might be in some use cases. I don't think that it's very helpful for the images I used here.