How to assign character to foreground and background in image?

Currently, I have an image with single letter in the middle it is white and background is black ,in grayscale. I managed to store them in different variable. But now, I would like to display them by cover the whole foreground and background with different character. For example, 'b' for the pixels of the image for background,while 'o' for the pixels of foreground. I try to search and try on my own but not successful.Therefore,I am here to ask for guide.

%create blank image
w = 150;
h = 150;
blankImage= 255*ones(w,h,3,'uint8');
%position of the letter in the empty cell
position_x = (w+1)/2;
position_y = (h+1)/2;
% varying the font size, start from 10 to 16
font_start = 58;
font_end = 64;
num_fontsA = font_start:2:font_end;
% get the number of fonts
numImagesA = length(num_fontsA)
% create a cell array with number of fonts to fill in the image in next step
A = cell(1, numImagesA);

% for loop to create letter 'A' 
% grayscale
% store into the cell array

for i=1:numImagesA
    for font_size = num_fontsA(i)
        img= insertText(blankImage,[position_x position_y],'A','Font','Times New Roman','FontSize',font_size,'TextColor','black','BoxColor','w','BoxOpacity',0,'AnchorPoint','Center');
        grayImage= rgb2gray(img);
        BWImage =  ~grayImage;
        background = BWImage == 0;
        foreground = ~background;
        %testing ,but it failed
        Newforegnd = foreground;
        L = repmat(foreground==1,1,1,size(Newforegnd,3));
        Newforegnd(L)='x';
        
        figure('Name','Background and Object','NumberTitle','off');
        montage({Newforegnd, foreground});
%         figure;
%         A{i} = BWImage;
%         subplot(2,2,i);
%         imshow(A{i});
        
        
% %         imshow(background);
%         A{i}= imnoise(img, 'gaussian',0, 0.01);
%         subplot(2,2,i);
%         imshow(A{i});
        axis on;
        title("A with font size " + font_size);
%         title("A with font size " + font_size +" with noise 0.01 variance");
    end
end

This is what I get from the background and foreground variable previously from help in Matlab forum.

[KALYAN ACHARJYA]

It seems those are binary images, have you looked at how array values ​​are perceived/displaying? The colors displayed are just a projection of the array data. It is either 0 or1 in a binary image. If the pixel is 1, it projects white and if 0 it projects black or vice versa. What do you mean by foreground or background here?

[Tuck Wai Yip]

That is not my main point, I need to make it in this manner.It is for OCR application. This is the example from senior, that is the only thing I get from my professor,I asked him he didn't tell anything to solve this. I need to produce similar type of image but on the letter this time.

It is a grayscale image, but now it only has two color, which is black on the letter 'A' and white on the background as you can see blankImage I prepare the image in grayscale.Therefore, It is not binary. I try to store the foreground which is the letter 'A' and background of the image in different variables.If there is way to work around to get that exact output,please guide.Thank you.

[Image Analyst]

"This is the example from senior" <== Who or what is "senior".

If your professor did not want you to have any additional information from him, then presumably he doesn't want us to give you any additional information either. So I guess you're on your own but I'm sure you can still use the help documentation. If your professor says it's OK to get help from the Answers forum, then let us know he says it's okay.

[Tuck Wai Yip]

It was the report from senior from university that graduated 3 years ago.

Sorry to offense,but it seems to me not that he didn't want me to have additional info from him, but it is probably that he didn't even know how to use Matlab. I tried to search for documentation and forum but it is not helpful as I cannot imagine how could I reach that step.Yeah,it is okay, since he is not going to help,and he didn't said anything regarding that as I told him about getting help in Matlab forum previously.

Answer 1

[DGM]

What you describe sounds like a simple logical masking operation, but the image appears to be a mosaic instead. If the latter is indeed the intended result, consider that the size of your output image will be the product of the sizes of the pattern image and the tile image(s). Considering that the pattern image has two values, then a logical approach should be evident. It may be prudent to reduce the size of the tile images, or at least ensure they have common geometry before trying to assemble the mosaic.

Actually if I'm going to offer an example, I might as well move it to an answer...

% generate some letter images.  textim() is from MIMT on the FEX
a = textim('A','ibm-vga-8x8');
b = textim('B','ibm-vga-8x8');
c = textim('C','ibm-vga-8x8');

% assemble mosaic using kronecker product
d = kron(a,b) + kron(1-a,c); 
% note 1-a denotes inversion.  a is of class double.  
% if a were uint8(), you'd need to use a different value.  
% you can also use imcomplement()

These images are obviously resized for viewing purposes. As I said, the output size will be the product of the pattern and tile sizes.

[Tuck Wai Yip]

Hi,the tile image that consists of other letters is not suitable since I need only one for each image and the image needs to be in symbol filled per pixels.Is it even possible to display in the form like the image above for my image of single letter,which each pixel is formed by symbol?Because it seems to be that the example from my senior is just a bunch of symbol printed per pixels maybe.Most probably he "print" the output in command window.

[DGM]

I'm not sure what you mean. Do you mean that you don't want the two combined?

da = kron(a,b);
dc = kron(1-a,c);

These images are one tile per pixel (the source images are only 8x8 pixels. Yours are obviously higher resolution, but I chose to just use these for sake of simplicity.

If you want to take your image and use it to generate character output, that's a bit different.

a = textim('A','ibm-vga-8x8')>0.5;
out = repmat('c',size(a));
out(a) = 'b';
disp(out)

gives

ccbbcccc
cbbbbccc
bbccbbcc
bbccbbcc
bbbbbbcc
bbccbbcc
bbccbbcc
cccccccc

[Tuck Wai Yip]

Yeah, this form is what I want to have(I mean the character output).I am sorry that I am not good at explaining thing.But this is exactly to what I think he wants the output to be.

Is that possible to take my original image that formed with insertText to produce this or textim

can receive font type and customize font size?

ccbbcccc
cbbbbccc
bbccbbcc
bbccbbcc
bbbbbbcc
bbccbbcc
bbccbbcc
cccccccc

[DGM]

If you want it to look like your picture (the font, etc) then you'd probably need to use insertText() or something similar. Textim() only has a selection of relatively compact legacy bitmap system/bios fonts -- nothing modern. This shows all the fonts:

http://mimtdocs.rf.gd/manual/html/textim.html

I'm not familiar with what insertText offers since I don't have CVT. If it's not to your liking, I'm pretty sure there are some other text-image converters on the File Exchange.

[Tuck Wai Yip]

Thank you,I will try the code soon and see how it goes.I will report my progress later in this thread.

Answer 2

[Image Analyst]

For what it's worth, here's perhaps a start:

% Demo by Image Analyst
clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format long g;
format compact;
fontSize = 20;

%--------------------------------------------------------------------------------------------------------
%    READ IN IMAGE
folder = pwd;
baseFileName = 'ocr.png';
grayImage = imread(baseFileName);

% Get the dimensions of the image.
% numberOfColorChannels should be = 1 for a gray scale image, and 3 for an RGB color image.
[rows, columns, numberOfColorChannels] = size(grayImage)
if numberOfColorChannels > 1
	% It's not really gray scale like we expected - it's color.
	% Extract the red channel (so the magenta lines will be white).
	grayImage = grayImage(:, :, 1);
end

% Get rid of junk in the lower left by erasing the leftmost 7 columns
grayImage(:, 1:7) = 255;

%--------------------------------------------------------------------------------------------------------
% Display the image.
subplot(2, 2, 1);
imshow(grayImage, []);
axis('on', 'image');
title('Original Gray Scale Image', 'FontSize', fontSize, 'Interpreter', 'None');
impixelinfo;
hFig = gcf;
hFig.WindowState = 'maximized'; % May not work in earlier versions of MATLAB.
drawnow;

% Take histogram
subplot(2, 2, 2);
edges = 0 : 256;
histObject = histogram(grayImage, edges)
grid on;
title('Histogram', 'FontSize', fontSize);
xlabel('Gray Level', 'FontSize', fontSize);
ylabel('Pixel Count', 'FontSize', fontSize);

% Binarize
mask = grayImage < 230;
subplot(2, 2, 3);
imshow(mask, []);
axis('on', 'image');
title('Binary Image', 'FontSize', fontSize, 'Interpreter', 'None');
impixelinfo;

% Get horizontal and vertical profiles.
horizontalProfile = mean(grayImage, 1);
verticalProfile = mean(grayImage, 2);
subplot(2, 2, 4);
plot(horizontalProfile, 'b-', 'LineWidth', 1);
hold on;
grid on;
plot(verticalProfile, 'r-', 'LineWidth', 1);
legend('Horizontal Profile', 'Vertical Profile');
drawnow;

% Find the centroids of each "lane" of characters
% First find vertical lanes by finding horizontal dividing points.
propsh = regionprops(horizontalProfile > 230, 'Centroid');
xy = vertcat(propsh.Centroid)
xh = unique(xy(:, 1)')
subplot(2, 2, 3);
hold on;
for k = 1 : length(xh)
	xline(xh(k), 'Color', 'r');
end

% Next find horizontal lanes by finding vertical dividing points.
propsv = regionprops(verticalProfile > 230, 'Centroid');
xy = vertcat(propsv.Centroid)
yv = unique(xy(:, 2)')
subplot(2, 2, 3);
hold on;
for k = 1 : length(yv)
	yline(yv(k), 'Color', 'y');
end

% Round centroids to the nearest line or column
xh = round(xh);
yv = round(yv);
figure;
numRows = length(yv);
numColumns = length(xh);
% Extract each subimage
plotIndex = 1;
for col = 1 : length(xh)-1
	col1 = xh(col);
	col2 = xh(col+1) - 1;
	fprintf('Extracting columns %d to %d.\n', col1, col2);
	for row = 1 : length(yv) - 1
		row1 = yv(row);
		row2 = yv(row+1)-1;
		subImage = grayImage(row1:row2, col1:col2);
		subplot(numRows, numColumns, plotIndex);
		imshow(subImage, []);
		plotIndex = plotIndex + 1;
		drawnow;
	end
end
g = gcf;
g.WindowState = 'maximized'; % May not work in earlier versions of MATLAB.

[Tuck Wai Yip]

Hi, may I ask what is you trying to achieve here? Because the input should be some letter image that is in the code at the top comment, but not the hand image from sign language.As what I said, he just give me a reference with university senior's example.What I want to achieve now is to produce output in that form using symbol but this time it is focused on what I trying to do, which is the single letter image.Is that possible to display that in command window,it seems to be as if the image is produced from command window through some method.Just my small guess,maybe it is wrong.Hope that you could take note what I am trying to do.If there is any guide,I would definitely appreciated.Thank you.

[Image Analyst]

If you have an ASCII art image, you first need to determine what each character is. Then you can map each character to a gray level, like A = 1, O = 2, T = 3 or whatever. Once you have that you can label the image with bwlabel() to find connected regions of a single letter. Then once you have all the regions, you can try to classify it as some particular shape.

[Tuck Wai Yip]

Oh,thank you for your guide.Now I understand what you try to achieve, maybe I can try that on my own.But somehow,thank you for your detailed code and guidance.Now I have more clearer route to how to get the character recognized.

[Image Analyst]

You're welcome. If you have more specific questions, post them and attach your image. Otherwise, please "Accept" the best answer (you can only accept one but you can vote for more than one).