% stanzaSeg.m
% takes in an image and returns an array of the Y-values between the stanza on the page
%
% arguments: imagename - string of filename
% returns: coorArray - array of Y-values seperating stanzas
function coorArray=stanzaSeg(imagename)

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%crop the original image so that only the page left


%read image to be segmented
pic = imread(imagename);

%get dimensions of image
[m,n,k] = size(pic);

%convert the image to grayscale
level = graythresh(pic);
pic1 = im2bw(pic,level+0.2);

%find the percentage of white space per column
rowMean = mean(pic1);

%find the coordinates for the top, bottom, left, and right 
%sides of the page by looking for the first row/column that has a lot of
%white space.

%if no new coordinates are found, use the edge of the image

%start at left side of image, find first instance where at least 10 rows are 50% white or more heading right
left = 0;
for i=1:10:700
    j=i+50;
    if(rowMean(i) > 0.5 && rowMean(j) > 0.5)
        left = i;
        break;
    end
end

%start at right side of image, find first instance where at least 10 rows are 50% white or more heading left
right = n;
for i=n:-10:n-700
    j=i-50;
    if(rowMean(i) > 0.5 && rowMean(j) > 0.5)
        right = i;
        break;
    end
end


%rotate the b/w image so that the above can be done for the top and bottom
%of the image
pic2 = imrotate(pic1,90);
colMean = mean(pic2);

top = 0;
for i=1:10:700
    j=i+50;
    if(colMean(i) > 0.2 && colMean(j) > 0.2)
        top = i;
        break;
    end
end

bottom = m;
for i=m:-10:m-500
    j=i-20;
    if(colMean(i) > 0.5 && colMean(j) > 0.5)
        bottom = i;
        break;
    end
end

%crop the image with the coordinates to get just the page
cropPic = imcrop(pic,[left top right bottom-top]);

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%create an array that will be used to hold the coordiantes for the stanzas
%on the page
coorArray = zeros(1);

%get size cooordinates, make b/w, rotate, and get get means from the
%cropped image, as well as the maxiumum value on the means
[m,n,k] = size(cropPic);
level = graythresh(cropPic);
pic1 = im2bw(cropPic,level+0.1);
pic2 = imrotate(pic1,90);
colMean = mean(pic2);
white = max(colMean);

%create a temporary array to hold the y-value coordinates of where the stanza breaks are
yValues = zeros(6);

%starting break will be where the top of the cropped page is
%"top" is always added to the coordinates because they are found from the cropped image, values need to work with the original image
yValues(1) = top;

%keep track of number of elements in the array
%this is incremented every time an element is added to the array
count = 1;

%keep track of the beginning and the end (fin) of a block of white space dividing two stanzas
begin = 0;
fin = 0;

%keep track of whether or not the program is already looking at white space
%works like a boolean
started = 1;

%for loop to find the stanza breaks
%start at top, increment by 20, stop and end of page
for i=1:20:m
	%if not already in white space, but in white space now: set "begin" to current and "started" to 1
    if(started == 0 && colMean(i) > white-0.03)
        begin = i;
        started = 1;
        
    %if not already in white space, and not in white space now: do nothing
	elseif(started == 0 && colMean(i) < white-0.03)
        
    %if already in white space, and in white space now: update "fin" to keep track of end of white space up to now
	elseif(started == 1 && colMean(i) > white-0.03)
        fin = i;
        
    %if in white space already, and out of white space now: check to see if y-value needs to be added to the array
	elseif(started == 1 && colMean(i) < white-0.03)
        %if white space was less than 20 pixels (fin never changed from 0): reset variables and do NOT add
		if(fin == 0)
            begin = 0;
            started = 0;
        %else: find the midpoint between "begin" and "fin" and add that value (plus top) to the array
		%reset variables
		else
            mid = (begin+fin)/2;
            yValues(count+1) = mid+top;
            count = count+1;
			started = 0;
            begin = 0;
            fin = 0;
        end
    end
	%if line is completely black (bottom of page): add to array and break the for loop
	%this is just in case there is black space at the bottom of the crop
    if(colMean(i) == 0)
        yValues(count+1) = i+top;
        count = count+1;
        break;
    
    end
end

%check distance from last value in the array to the bottom of the page
%if greater that 200 pixels, add bottom of page coordinate to the array
%used if bottom of page is mostly white and coordinates were not added after the final stanza
if(m+top - yValues(count) > 200)
    yValues(count+1) = m+top;
    count = count+1;
end

%copy contents of yValues to coorArray since coorArray will be returned
for i=1:count
    coorArray(i) = yValues(i);
end


end