How to erode parts of an image

Does anyone know the best way to get rid of the black part on the left side of the image and the black border around the white. I need to keep the numbers and letters. I tried different types of imerode, but could never get rid of it. Any pointers would be great thank you!

[John D'Errico]

Just crop the image. Whats the problem?

[Adam Kelly]

How do I find the right dimensions to crop the part I want. Sorry if it is a dumb question, I am new to Matlab

Answer 1

[Image Analyst]

This works without changing the shape of the "good" characters, like erosion would do:

grayImage = imread('NeedtobeEroded.jpeg');
if ndims(grayImage) > 1
	% It's RGB - convert to gray scale.
	grayImage = grayImage(:,:,1);
end
subplot(2, 2, 1);
imshow(grayImage);
binaryImage = grayImage > 128; % Convert to binary.
% Get rid of largest blob.
mask = bwareafilt(~binaryImage, 1);
binaryImage(mask) = true;
subplot(2, 2, 2);
imshow(binaryImage);
% Get aspect ratios so we can find the tall, narrow vertical line at the right.
props = regionprops(~binaryImage, 'BoundingBox', 'PixelIdxList');
bb = vertcat(props.BoundingBox)
aspectRatios = bb(:, 4) ./ bb(:, 3) % Heights over widths
% Find aspect ratios more than, say, 50.
badRegions = aspectRatios > 50;
% Fill them in
for k = 1 : length(props)
	if badRegions(k)
		% It's bad.  Fill this one in.
		binaryImage(props(k).PixelIdxList) = true;
	end
end
subplot(2, 2, 3);
imshow(binaryImage);

%------------------------------------------------------------------------------
% Set up figure properties:
% Enlarge figure to full screen.
set(gcf, 'Units', 'Normalized', 'OuterPosition', [0, 0.04, 1, 0.96]);
% Get rid of tool bar and pulldown menus that are along top of figure.
% set(gcf, 'Toolbar', 'none', 'Menu', 'none');
% Give a name to the title bar.
set(gcf, 'Name', 'Demo by ImageAnalyst', 'NumberTitle', 'Off')

Another way to do it rather than getting rid of blobs with bad aspect ratios is, if you know the bad line is always there and always at the right side, you can simply fill in the rightmost blob.

grayImage = imread('NeedtobeEroded.jpeg');
if ndims(grayImage) > 1
	% It's RGB - convert to gray scale.
	grayImage = grayImage(:,:,1);
end
subplot(4, 1, 1);
imshow(grayImage);
binaryImage = grayImage > 128; % Convert to binary.
% Get rid of largest blob.
largestBlob = bwareafilt(~binaryImage, 1);
binaryImage(largestBlob) = true;
subplot(4, 1, 2);
imshow(binaryImage);

% Get rid of rightmost blob.
[labeledImage, numBlobs] = bwlabel(~binaryImage);
rightMostBlob = ismember(labeledImage, numBlobs);
subplot(4, 1, 3);
imshow(rightMostBlob);

binaryImage(rightMostBlob) = true;
subplot(4, 1, 4);
imshow(binaryImage);

%------------------------------------------------------------------------------
% Set up figure properties:
% Enlarge figure to full screen.
set(gcf, 'Units', 'Normalized', 'OuterPosition', [0, 0.04, 1, 0.96]);
% Get rid of tool bar and pulldown menus that are along top of figure.
% set(gcf, 'Toolbar', 'none', 'Menu', 'none');
% Give a name to the title bar.
set(gcf, 'Name', 'Demo by ImageAnalyst', 'NumberTitle', 'Off')

Answer 2

[Thiago Henrique Gomes Lobato]

If you just have one image, doing a matlab code to find the borders would be an overkill, so it would be better to just check the pixel values that you want to cut. If, however, you have many similar images it start to make sense to develop some algorithm to find the borders for you. I made an example that may give you some ideas if this is the case (the resulting image is only with the white part and numbers):

I = imread('NeedtobeEroded.jpeg');
Ib = im2bw(I,0.5); % Binarize the image, so only values of 0 and 1 remains in the matrix

% Get Vertical and Horizontal Lines
VerticalLine = Ib(:,round(end/2));
HorizontalLine = Ib(round(end/2),:);

% Take the derivative, i.e, when the white becomes black and vice-versa
Diffy = diff(VerticalLine);
Diffx = diff(HorizontalLine);

% Get the color change locations
IndexChangeX = find(Diffy~=0);
IndexChangeY = find(Diffx~=0);

% Use the color change locations as index references
SafetyOffset = 5; % This is just a safety offset
xi = IndexChangeX(2)+SafetyOffset;xf = IndexChangeX(end-1)-SafetyOffset;
yi = IndexChangeY(2)+SafetyOffset;yf = IndexChangeY(end-1)-SafetyOffset;

INew = I (xi:xf,yi:yf,:);
figure,imshow(INew)

This code doesn't take into account boundary index problems, but as I said, it is just to give you a roughly idea what could be done and, for your case, it works fine.