how to normalized a hsv histogram?

i want to normalize hsv histogram. i have three different of each h,s,and v. i want to normalize them so that i can compare it with h,s and v plot. for this firstly i have to normalize the histogram i tried in following way but it did not work..

rgb = imread('wheatimage.jpg');hsv = rgb2hsv(rgb);

h = hsv(:, :, 1);

v = hsv(:, :, 3);

numberOfBins = 50;

s = hsv(:, :, 2); % or whatever.

[countsH valuesH] = hist(h, numberOfBins);

figure; subplot(2, 2, 1);

bar(valuesH, countsH, 'BarWidth', 1);

title('Histogram of the H Channel', 'FontSize', 15);

[countsS, valuesS] = hist(s, numberOfBins);

subplot(2, 2, 2);

bar(valuesS, countsS, 'BarWidth', 1);

title('Histogram of the S Channel', 'FontSize', 15);

[countsV, valuesV] = hist(v, numberOfBins);

bar(valuesV, countsV, 'BarWidth', 1);

title('Histogram of the V Channel', 'FontSize', 15);%# create histogram from a normal distribution.

g=1/sqrt(2*pi)*exp(-0.5*valuesH.^2);%# pdf of the normal distribution

% DIVIDE BY AREA figure(1)

bar(valuesH,countsH/trapz(valuesH,countsH));hold on

plot(valuesH,g,'r');hold off

[Jan]

What exactly does not work? Do you get an error message or do the results differ from your expectations?

[R]

the output is not a normalized histogram...

[R]

is their any other way to normalize hsv histogram other than this.

[Jan]

What exactly should be "normal" after the normalization? All we see is the code, which does not do what you want. But this does not allow to identify, what you want.

Answer 1

[Jan]

What exactly should be normalized? The area under each curve or all 3 curves?

Perhaps this helps: FEX: NormalizedHistogram

[R]

sir i had tried histnorm but its not working well. suggest me something valid

[R]

sir i want just that my histgram of h get normalize ....

[Jan]

I cannot guess what "not well" and "valid" means in your case. Let me ask another time, what you exactly want: "normalization" can mean, that the maximum value is set to 1.0, or that the area under the fitted Gaussian distribution is set to 1.0, or that the sum of the measured values is set to 1.0, etc. I think if you find out, what you exactly want, the implementation is very easy and most likely you can solve this by your own.

Of course I could try to guess, but this is not why I want the nature of this forum should be. I like to offer assistence for learning how users can find solutions by their own.

Did you search in the net already? Perhaps Answers: normalizing a histogram helps.

[Image Analyst]

And even beyond that, what if you did normalize your histogram, through whatever means and algorithm you choose? What then? Why do you want to normalize it? What do you think you can do that you can't do without it normalized? Saying you want to normalize h so that you can "compare it with h plot" doesn't really make sense. I don't know what's being compared with what.

[R]

yes sir i tried histnorm , i histogram should be normalize in such away that it can be compare with histogram of another image h value..the values at boundary should be clear . in case of normal histogram of hsv it is hard to identify boundary if normalize graph build on histogram its boundry get clear..

[R]

sir from last many days i had tried what have been possible function or formula to get clear normalize histogram..

[Jan]

@Ramandeep: I have the impression that the communication got more complicated then necessary. So let's try to be as clear as possible: "normalizing a histogram" is not uniquely defined. A mathematically unique definition is required, before this can be implemented in a function. I expect something like: "The sum of XYZ is 1.0".

I do not understand, what a "clear boundary" is.

[R]

i want to compare to image on the base of hsv and compare their histogram on bases of chi-square dist method. so for that i want to normalize my hsv histograms..

[R]

comparsion is latter issue firstly graph need to be normalize.