Finding minimum and maximum value in a histogram

I have an image of a face with all its non-skin part removed (hair, shirt, eyebrow, etc). From that image, I want to find its minimum and maximum value in its histogram of X and Y channel (CIE XYZ). I've tried sorting value from left side to middle and right side to middle, but the minimum value gets messed up. Anyone have an idea I can try?

Here's the code I've tried:

clc
clear all
close all

img_rgb=imread('crop02.jpg');

img_xyz=rgb2xyz(img_rgb);
img_x=img_xyz(:,:,1);
img_y=img_xyz(:,:,2);
hist_x=imhist(img_x);
hist_y=imhist(img_y);
min_x=0;
max_x=0;
min_y=0;
max_y=0;

for i=1:1:length(hist_x)
    if min_x==0
        if hist_x(i)~=0
            min_x=i;
        end
    end
end

for i=length(hist_x):-1:1
    if max_x==0
        if hist_x(i)~=0
            max_x=i;
        end
    end
end

for i=1:1:length(hist_y)
    if min_y==0
        if hist_y(i)~=0
            min_y=i;
        end
    end
end

for i=length(hist_y):-1:1
    if max_y==0
        if hist_y(i)~=0
            max_y=i;
        end
    end
end

min_x=(min_x-1)/(length(hist_x)-1)
max_x=(max_x-1)/(length(hist_x)-1)
min_y=(min_y-1)/(length(hist_y)-1)
max_y=(max_y-1)/(length(hist_y)-1)

+EDIT+ I'm trying to find minimum and maximum value of X-axis, not the Y-axis, in the area I marked. With min() and max() function, the minimum value is 0, and maximum value is about 0.7~0.9 with all three of them. As I see it, those values do not represent the area I marked. The second histogram should have min value of ~0.07 and max value of ~0.32. How should I do it?

Answer 1

[Image Analyst]

If you want the max and min in the image (rather than of the bin center or edge values), you can use max():

min_x = min(img_x(:));
min_y = min(img_y(:));
max_x = max(img_x(:));
max_y = max(img_y(:));

[Alvin Radeka]

Same results, but much shorter, thank you. But this result is not the one I was hoping for. I will add screenshot of the histogram and the value I am trying to extract ASAP. Hope it will give clearer picture of what I was trying to do.

[Image Analyst]

If you want the values from the histogram instead of the image itself, they will be quantized of course. But use the new histogram() function to return a histogram object. That object has tons of information in it

histObj = histogram(yourData);

[Alvin Radeka]

Yes, this new function will be easier to process I guess. And I've edited my question to add few pictures and short explanation. If you don't mind, please take a look at it.

[Image Analyst]

If you have a single pixel, it will affect the min or max. You may want to use cumsum() to take the CDF of the histogram and find the 1% and 99%, or 5% and 95% points for the x and y channels.

Answer 2

[Walter Roberson]

Use the two-output versions of max() and min() . Those return indices in the second output.

   [largescount, index_of_largest] = max(YourCountArray);

Now you can use index_of_largest to connect to whatever that slot in the count represents.

[Alvin Radeka]

I see. But still the returned value was different than what I had in mind. I will edit my question to add few screenshots. If you don't mind, please take a look at it.