Help with using SURF to analyze an image?

So I am trying to analyze and image using "surf" to make a 3D plot whose z value is the gray scale, where darker image will have lower z (Black is 0, White is 255). How would I use surf to analyze an image like that?

Answer 1

[Rik]

What your goal is, is not really clear to me, but you can use the code below to view an image as a surf. I used an example image packaged with the IPT, so if you don't have that toolbox, you'll have to use your actual image.

figure(1),clf(1)
peppers = double(imread('peppers.png'));
IM=peppers(:,:,1)*.3+peppers(:,:,1)*.6+peppers(:,:,1)*.1;
[X,Y]= meshgrid(1:size(IM,2),size(IM,1):-1:1);
surf(X,Y,IM,'EdgeColor','none')
colormap('gray')
view(10,90)

[Taylor Gates]

your code has me on the right track. Can you please explain the 4th line as far as what each one of those things does/represents for the final product?

[Image Analyst]

meshgrid just gives you a 2-D list of all the x,y coordinates in your image, which is what surf() wants.

That image doesn't look very 3D-ish.

Again, you would not use surf() to "analyze" an image unless it's just for you to qualitatively visualize and understand it better. It doesn't do any image analysis at all.

[Taylor Gates]

So what would be the meaning of the -1,1? I guess I worded it poorly because all I needed to do was to quantify black and white values from 0-256 using surf

[Image Analyst]

That just flips it bottom to top because graphs (like surf) have the origin at the bottom left while images and matrices have the origin at the upper left.

I don't know what "quantify" means to you. The image is digital, so the values have already been quantized into 256 discrete levels. What do you mean?

[Taylor Gates]

for example. If I delete the last line of the code you provided, it should display (in mine at least) a 3D surf of the grayscale. That is what I was trying to get.

[Rik]

Just to clarify: you can still 3D rotate this. The line with view just sets the camera to a point where you can easily recognize the image.

Answer 2

[Image Analyst]

Perhaps you'd like meshcanopy: https://www.mathworks.com/matlabcentral/fileexchange/29485-meshcanopy

Answer 3

[Image Analyst]

You wouldn't. The surf() function in base MATLAB just for visualization of smooth images, as opposed to SURF from the Computer Vision System Toolbox. The surf() function is not very good for visualizing rough images, and no good at all for analysis. See my Image Segmentation Tutorial in my File Exchange for an intro to image analysis.