How can I rescale an histogram ?

Hello,

I am using a code which allows to plot an histogram of H (Hue parameter) values of each pixel from a picture. I would like to compare histograms from two different pictures. I know how to overlay them but one is smmaller than the other one so it's difficult to read the whole figure. I don't know how to normalize or rescale (I am not sure about the right word) them. I want to set the value of the higher bin to 1 for each histogram.

Please find below the code I am using :

    % ** 00 ** ***Define path to folder and pick movie file***
    [file,fold] = uigetfile('*.tif','Select the tif movie file');

    % ** 1 ** ***Extract all tif data 
    info=imfinfo([fold,file]); %Yields error but give length of movie
    name=file(1:length(file)-4);
    s=struct;
    for i=1:length(info)
    im=imread([fold,file],i);
    s(i).ima=im; %put images in an array structure
    end
    clear im

    % ** 2 ** ***Pick several images and plot all points in colorspace***
    list=[1];
    sl=struct;
    n=0;
    for i=list
    n=n+1;
    sl(n).ima=s(i).ima;
    end

    for i=1:length(list)
    im=sl(i).ima; %image in rgb

    %Plot RGB
    figure('Color',[1 1 1])
    imagesc(im)
    axis equal tight
    set(gca,'XTick',[],'YTick',[])
    title(['Frame ' num2str(list(i))],'Fontsize',16)
    saveas(gcf,[fold,'Frame_' num2str(list(i),'%.3i') '.png'],'png')

    %Convert RBG to HSV and keep only H
    imH=rgb2hsv(im);
    imH=imH(:,:,1);

    %Convert RGB to XYZ
    imX=rgb2xyz(im);

    %Convert to xyz
    x=imX(:,:,1)./(imX(:,:,1)+imX(:,:,2)+imX(:,:,3));
    y=imX(:,:,2)./(imX(:,:,1)+imX(:,:,2)+imX(:,:,3));

    %Reshape to vector
    xx=reshape(x,[1 size(x,1)*size(x,2)]);
    yy=reshape(y,[1 size(y,1)*size(y,2)]);

   %Remove the background based on criterion on RBG sum
    Sc=10000; %ENTER VALUE
    S=im(:,:,1)+im(:,:,2)+im(:,:,3);
    S(S<Sc)=0;
    S=double(reshape(S,[1 size(x,1)*size(x,2)]));
    H=reshape(imH,[1 size(x,1)*size(x,2)]);
    xx(S==0)=[];
    yy(S==0)=[];
    H(S==0)=[];

    figure('Color',[1 1 1],'units','normalized','outerposition',[0 0 1 1])
    subplot(2,2,1)
    imagesc(im)
    axis equal tight
    set(gca,'XTick',[],'YTick',[])
    title(['Frame ' num2str(list(i))],'Fontsize',16)
    subplot(2,2,[2,4])
    plotChromaticity
    hold all
    hd=histogram2(xx,yy,300,'DisplayStyle','tile');
    colormap gray
    title('CIE plot','Fontsize',16)
    subplot(2,2,3)
    histogram(H,300)
    title('Histo H layer','Fontsize',16)
    saveas(gcf,[fold,'Frame_' num2str(list(i),'%.3i') '_CIE.png'],'png') 
    end

Answer 1

[Guillaume]

Your code would be so much easier to understand if you'd used meaningful variable names instead of useless xx, yy, S, etc. As it is, I've not really tried to understand what you're plotting.

Rather than setting the highest bin to 1, which I don't think is a good way of comparing histograms, I'd recommend that you normalise your histograms so that the sum of the bar is 1. This is easily achieved by setting the 'Normalization' option of histogram to 'probability' (or 'pdf' if your bins are not equal size)

histogram(H, 300, 'Normalization', 'probability')

[Benjamin Poggi]

Thank you very much for your advice.

[Tanguy RALLON]

I have this code and I just want my Yaxis sum is 1. In other words I want to divide my Y scale by 1000000.

How I can do that easily ?

%roll is like random values between 1 and 6

nbins = [1 2 3 4 5 6];

hist(roll,nbins)