General approach for extracting specific lines or line segments in an image












3















I have this sample cropped image:



enter image description here



I need to make black thick lines (horizontal and vertical) disappear or extracted while leave all other info intact. These specific lines are either 4 or 5 pixels thick. I tried:




  1. Simple filtering of rows having more zeros/ones if image is read as NumPy array but the filtering condition doesn't terminate till a single row is left with zero or one.

  2. Erosion with simple kernel (3,3) but it leaves some noise because some symbols are also thick black

  3. Dilation with line structuring element of the width of image width but there are overwhelming variations on line segments' sizes connecting different symbols that the basic info about each small line segment is lost.


Can someone give insights or directions about what kind of structuring elements, what type of morphological ops should be considered or may be any other clever heuristics? The output, if extraction of thick black lines is done, will then look like this grid of random line segments:
enter image description here










share|improve this question

























  • The "thick black lines" you speak of... They appear to have a height or width that's higher than the other lines, yes?

    – rayryeng
    Nov 14 '18 at 7:21








  • 3





    You should look into Hough transformation - after erosion, it is probably possible to detect the straight lines left

    – T A
    Nov 14 '18 at 7:47






  • 1





    what about erosion, followed by lineSegmentDetector? The noise should hopefully not build up line segments?

    – Micka
    Nov 14 '18 at 7:49






  • 4





    An ideal solution as picture would be helpful, otherwise e.g. "thick horizontal line" is ambiguous.

    – Vroomfondel
    Nov 14 '18 at 9:45






  • 1





    I think the expected output would be desirable... Or else we are all going to propose solutions that may not be what you want. Please update your question to provide what the expected output is.

    – rayryeng
    Nov 14 '18 at 20:21


















3















I have this sample cropped image:



enter image description here



I need to make black thick lines (horizontal and vertical) disappear or extracted while leave all other info intact. These specific lines are either 4 or 5 pixels thick. I tried:




  1. Simple filtering of rows having more zeros/ones if image is read as NumPy array but the filtering condition doesn't terminate till a single row is left with zero or one.

  2. Erosion with simple kernel (3,3) but it leaves some noise because some symbols are also thick black

  3. Dilation with line structuring element of the width of image width but there are overwhelming variations on line segments' sizes connecting different symbols that the basic info about each small line segment is lost.


Can someone give insights or directions about what kind of structuring elements, what type of morphological ops should be considered or may be any other clever heuristics? The output, if extraction of thick black lines is done, will then look like this grid of random line segments:
enter image description here










share|improve this question

























  • The "thick black lines" you speak of... They appear to have a height or width that's higher than the other lines, yes?

    – rayryeng
    Nov 14 '18 at 7:21








  • 3





    You should look into Hough transformation - after erosion, it is probably possible to detect the straight lines left

    – T A
    Nov 14 '18 at 7:47






  • 1





    what about erosion, followed by lineSegmentDetector? The noise should hopefully not build up line segments?

    – Micka
    Nov 14 '18 at 7:49






  • 4





    An ideal solution as picture would be helpful, otherwise e.g. "thick horizontal line" is ambiguous.

    – Vroomfondel
    Nov 14 '18 at 9:45






  • 1





    I think the expected output would be desirable... Or else we are all going to propose solutions that may not be what you want. Please update your question to provide what the expected output is.

    – rayryeng
    Nov 14 '18 at 20:21
















3












3








3


1






I have this sample cropped image:



enter image description here



I need to make black thick lines (horizontal and vertical) disappear or extracted while leave all other info intact. These specific lines are either 4 or 5 pixels thick. I tried:




  1. Simple filtering of rows having more zeros/ones if image is read as NumPy array but the filtering condition doesn't terminate till a single row is left with zero or one.

  2. Erosion with simple kernel (3,3) but it leaves some noise because some symbols are also thick black

  3. Dilation with line structuring element of the width of image width but there are overwhelming variations on line segments' sizes connecting different symbols that the basic info about each small line segment is lost.


Can someone give insights or directions about what kind of structuring elements, what type of morphological ops should be considered or may be any other clever heuristics? The output, if extraction of thick black lines is done, will then look like this grid of random line segments:
enter image description here










share|improve this question
















I have this sample cropped image:



enter image description here



I need to make black thick lines (horizontal and vertical) disappear or extracted while leave all other info intact. These specific lines are either 4 or 5 pixels thick. I tried:




  1. Simple filtering of rows having more zeros/ones if image is read as NumPy array but the filtering condition doesn't terminate till a single row is left with zero or one.

  2. Erosion with simple kernel (3,3) but it leaves some noise because some symbols are also thick black

  3. Dilation with line structuring element of the width of image width but there are overwhelming variations on line segments' sizes connecting different symbols that the basic info about each small line segment is lost.


Can someone give insights or directions about what kind of structuring elements, what type of morphological ops should be considered or may be any other clever heuristics? The output, if extraction of thick black lines is done, will then look like this grid of random line segments:
enter image description here







matlab opencv image-processing image-morphology






share|improve this question















share|improve this question













share|improve this question




share|improve this question








edited Nov 14 '18 at 20:35







SKR

















asked Nov 14 '18 at 6:05









SKRSKR

768




768













  • The "thick black lines" you speak of... They appear to have a height or width that's higher than the other lines, yes?

    – rayryeng
    Nov 14 '18 at 7:21








  • 3





    You should look into Hough transformation - after erosion, it is probably possible to detect the straight lines left

    – T A
    Nov 14 '18 at 7:47






  • 1





    what about erosion, followed by lineSegmentDetector? The noise should hopefully not build up line segments?

    – Micka
    Nov 14 '18 at 7:49






  • 4





    An ideal solution as picture would be helpful, otherwise e.g. "thick horizontal line" is ambiguous.

    – Vroomfondel
    Nov 14 '18 at 9:45






  • 1





    I think the expected output would be desirable... Or else we are all going to propose solutions that may not be what you want. Please update your question to provide what the expected output is.

    – rayryeng
    Nov 14 '18 at 20:21





















  • The "thick black lines" you speak of... They appear to have a height or width that's higher than the other lines, yes?

    – rayryeng
    Nov 14 '18 at 7:21








  • 3





    You should look into Hough transformation - after erosion, it is probably possible to detect the straight lines left

    – T A
    Nov 14 '18 at 7:47






  • 1





    what about erosion, followed by lineSegmentDetector? The noise should hopefully not build up line segments?

    – Micka
    Nov 14 '18 at 7:49






  • 4





    An ideal solution as picture would be helpful, otherwise e.g. "thick horizontal line" is ambiguous.

    – Vroomfondel
    Nov 14 '18 at 9:45






  • 1





    I think the expected output would be desirable... Or else we are all going to propose solutions that may not be what you want. Please update your question to provide what the expected output is.

    – rayryeng
    Nov 14 '18 at 20:21



















The "thick black lines" you speak of... They appear to have a height or width that's higher than the other lines, yes?

– rayryeng
Nov 14 '18 at 7:21







The "thick black lines" you speak of... They appear to have a height or width that's higher than the other lines, yes?

– rayryeng
Nov 14 '18 at 7:21






3




3





You should look into Hough transformation - after erosion, it is probably possible to detect the straight lines left

– T A
Nov 14 '18 at 7:47





You should look into Hough transformation - after erosion, it is probably possible to detect the straight lines left

– T A
Nov 14 '18 at 7:47




1




1





what about erosion, followed by lineSegmentDetector? The noise should hopefully not build up line segments?

– Micka
Nov 14 '18 at 7:49





what about erosion, followed by lineSegmentDetector? The noise should hopefully not build up line segments?

– Micka
Nov 14 '18 at 7:49




4




4





An ideal solution as picture would be helpful, otherwise e.g. "thick horizontal line" is ambiguous.

– Vroomfondel
Nov 14 '18 at 9:45





An ideal solution as picture would be helpful, otherwise e.g. "thick horizontal line" is ambiguous.

– Vroomfondel
Nov 14 '18 at 9:45




1




1





I think the expected output would be desirable... Or else we are all going to propose solutions that may not be what you want. Please update your question to provide what the expected output is.

– rayryeng
Nov 14 '18 at 20:21







I think the expected output would be desirable... Or else we are all going to propose solutions that may not be what you want. Please update your question to provide what the expected output is.

– rayryeng
Nov 14 '18 at 20:21














1 Answer
1






active

oldest

votes


















2














This is how you erode the image and extract hough lines:



I=rgb2gray(imread('https://i.stack.imgur.com/cbHFL.jpg'));

Ibw=I>200;

imshow(Ibw)
SE=strel('disk',1)
Ier=imerode(~Ibw,SE);

[H,T,R] = hough(Ier);
P = houghpeaks(H,100,'threshold',ceil(0.1*max(H(:))));

lines = houghlines(Ier,T,R,P);

%% plot
imshow(I);hold on
max_len = 0;
for k = 1:length(lines)
xy = [lines(k).point1; lines(k).point2];
plot(xy(:,1),xy(:,2),'LineWidth',2,'Color','green');

% Plot beginnings and ends of lines
plot(xy(1,1),xy(1,2),'x','LineWidth',2,'Color','blue');
plot(xy(2,1),xy(2,2),'x','LineWidth',2,'Color','red');

% Determine the endpoints of the longest line segment
len = norm(lines(k).point1 - lines(k).point2);
if ( len > max_len)
max_len = len;
xy_long = xy;
end
end


From here, you can start thinking on what to delete. This is not straightforward unless you have a dictionary of symbols, e.g. how do you delete the line around the structures with >-< shape? do you delete all the middle pixels or do you keep the entire middle thin bar? You can only know this if you know how the symbol should be without the thick lines.



enter image description here






share|improve this answer


























  • I used Hough Lines transform but due to the thickness of black lines there were many small lines detected in one big line and it was difficult to control through the parameters. I have a few questions: 1) why erode with a disk element of size 1? 2) Why not erode with rectangular element? 3) The green line seems to pass over the symbols whereas in the input image black line never crosses any symbol, actually we have to consider them as line segments before and after any symbol. 4) are the blue cross beginning and red cross end of a single line? 5) one small segment in bottom was not detected.

    – SKR
    Nov 15 '18 at 5:21











  • @SKR 1)empirical test showed good result, didnt think of it more. 2) same as 1. 3) The last paragraph is about that. This is not a full solution to your huge project, just a starting point. Your question is too big to answer in SO entirely. People get paid to solve this type of problems. 4) yes, check the code 5) Yes

    – Ander Biguri
    Nov 15 '18 at 9:51











Your Answer






StackExchange.ifUsing("editor", function () {
StackExchange.using("externalEditor", function () {
StackExchange.using("snippets", function () {
StackExchange.snippets.init();
});
});
}, "code-snippets");

StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "1"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);

StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});

function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});


}
});














draft saved

draft discarded


















StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53294066%2fgeneral-approach-for-extracting-specific-lines-or-line-segments-in-an-image%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown

























1 Answer
1






active

oldest

votes








1 Answer
1






active

oldest

votes









active

oldest

votes






active

oldest

votes









2














This is how you erode the image and extract hough lines:



I=rgb2gray(imread('https://i.stack.imgur.com/cbHFL.jpg'));

Ibw=I>200;

imshow(Ibw)
SE=strel('disk',1)
Ier=imerode(~Ibw,SE);

[H,T,R] = hough(Ier);
P = houghpeaks(H,100,'threshold',ceil(0.1*max(H(:))));

lines = houghlines(Ier,T,R,P);

%% plot
imshow(I);hold on
max_len = 0;
for k = 1:length(lines)
xy = [lines(k).point1; lines(k).point2];
plot(xy(:,1),xy(:,2),'LineWidth',2,'Color','green');

% Plot beginnings and ends of lines
plot(xy(1,1),xy(1,2),'x','LineWidth',2,'Color','blue');
plot(xy(2,1),xy(2,2),'x','LineWidth',2,'Color','red');

% Determine the endpoints of the longest line segment
len = norm(lines(k).point1 - lines(k).point2);
if ( len > max_len)
max_len = len;
xy_long = xy;
end
end


From here, you can start thinking on what to delete. This is not straightforward unless you have a dictionary of symbols, e.g. how do you delete the line around the structures with >-< shape? do you delete all the middle pixels or do you keep the entire middle thin bar? You can only know this if you know how the symbol should be without the thick lines.



enter image description here






share|improve this answer


























  • I used Hough Lines transform but due to the thickness of black lines there were many small lines detected in one big line and it was difficult to control through the parameters. I have a few questions: 1) why erode with a disk element of size 1? 2) Why not erode with rectangular element? 3) The green line seems to pass over the symbols whereas in the input image black line never crosses any symbol, actually we have to consider them as line segments before and after any symbol. 4) are the blue cross beginning and red cross end of a single line? 5) one small segment in bottom was not detected.

    – SKR
    Nov 15 '18 at 5:21











  • @SKR 1)empirical test showed good result, didnt think of it more. 2) same as 1. 3) The last paragraph is about that. This is not a full solution to your huge project, just a starting point. Your question is too big to answer in SO entirely. People get paid to solve this type of problems. 4) yes, check the code 5) Yes

    – Ander Biguri
    Nov 15 '18 at 9:51
















2














This is how you erode the image and extract hough lines:



I=rgb2gray(imread('https://i.stack.imgur.com/cbHFL.jpg'));

Ibw=I>200;

imshow(Ibw)
SE=strel('disk',1)
Ier=imerode(~Ibw,SE);

[H,T,R] = hough(Ier);
P = houghpeaks(H,100,'threshold',ceil(0.1*max(H(:))));

lines = houghlines(Ier,T,R,P);

%% plot
imshow(I);hold on
max_len = 0;
for k = 1:length(lines)
xy = [lines(k).point1; lines(k).point2];
plot(xy(:,1),xy(:,2),'LineWidth',2,'Color','green');

% Plot beginnings and ends of lines
plot(xy(1,1),xy(1,2),'x','LineWidth',2,'Color','blue');
plot(xy(2,1),xy(2,2),'x','LineWidth',2,'Color','red');

% Determine the endpoints of the longest line segment
len = norm(lines(k).point1 - lines(k).point2);
if ( len > max_len)
max_len = len;
xy_long = xy;
end
end


From here, you can start thinking on what to delete. This is not straightforward unless you have a dictionary of symbols, e.g. how do you delete the line around the structures with >-< shape? do you delete all the middle pixels or do you keep the entire middle thin bar? You can only know this if you know how the symbol should be without the thick lines.



enter image description here






share|improve this answer


























  • I used Hough Lines transform but due to the thickness of black lines there were many small lines detected in one big line and it was difficult to control through the parameters. I have a few questions: 1) why erode with a disk element of size 1? 2) Why not erode with rectangular element? 3) The green line seems to pass over the symbols whereas in the input image black line never crosses any symbol, actually we have to consider them as line segments before and after any symbol. 4) are the blue cross beginning and red cross end of a single line? 5) one small segment in bottom was not detected.

    – SKR
    Nov 15 '18 at 5:21











  • @SKR 1)empirical test showed good result, didnt think of it more. 2) same as 1. 3) The last paragraph is about that. This is not a full solution to your huge project, just a starting point. Your question is too big to answer in SO entirely. People get paid to solve this type of problems. 4) yes, check the code 5) Yes

    – Ander Biguri
    Nov 15 '18 at 9:51














2












2








2







This is how you erode the image and extract hough lines:



I=rgb2gray(imread('https://i.stack.imgur.com/cbHFL.jpg'));

Ibw=I>200;

imshow(Ibw)
SE=strel('disk',1)
Ier=imerode(~Ibw,SE);

[H,T,R] = hough(Ier);
P = houghpeaks(H,100,'threshold',ceil(0.1*max(H(:))));

lines = houghlines(Ier,T,R,P);

%% plot
imshow(I);hold on
max_len = 0;
for k = 1:length(lines)
xy = [lines(k).point1; lines(k).point2];
plot(xy(:,1),xy(:,2),'LineWidth',2,'Color','green');

% Plot beginnings and ends of lines
plot(xy(1,1),xy(1,2),'x','LineWidth',2,'Color','blue');
plot(xy(2,1),xy(2,2),'x','LineWidth',2,'Color','red');

% Determine the endpoints of the longest line segment
len = norm(lines(k).point1 - lines(k).point2);
if ( len > max_len)
max_len = len;
xy_long = xy;
end
end


From here, you can start thinking on what to delete. This is not straightforward unless you have a dictionary of symbols, e.g. how do you delete the line around the structures with >-< shape? do you delete all the middle pixels or do you keep the entire middle thin bar? You can only know this if you know how the symbol should be without the thick lines.



enter image description here






share|improve this answer















This is how you erode the image and extract hough lines:



I=rgb2gray(imread('https://i.stack.imgur.com/cbHFL.jpg'));

Ibw=I>200;

imshow(Ibw)
SE=strel('disk',1)
Ier=imerode(~Ibw,SE);

[H,T,R] = hough(Ier);
P = houghpeaks(H,100,'threshold',ceil(0.1*max(H(:))));

lines = houghlines(Ier,T,R,P);

%% plot
imshow(I);hold on
max_len = 0;
for k = 1:length(lines)
xy = [lines(k).point1; lines(k).point2];
plot(xy(:,1),xy(:,2),'LineWidth',2,'Color','green');

% Plot beginnings and ends of lines
plot(xy(1,1),xy(1,2),'x','LineWidth',2,'Color','blue');
plot(xy(2,1),xy(2,2),'x','LineWidth',2,'Color','red');

% Determine the endpoints of the longest line segment
len = norm(lines(k).point1 - lines(k).point2);
if ( len > max_len)
max_len = len;
xy_long = xy;
end
end


From here, you can start thinking on what to delete. This is not straightforward unless you have a dictionary of symbols, e.g. how do you delete the line around the structures with >-< shape? do you delete all the middle pixels or do you keep the entire middle thin bar? You can only know this if you know how the symbol should be without the thick lines.



enter image description here







share|improve this answer














share|improve this answer



share|improve this answer








edited Nov 14 '18 at 11:39

























answered Nov 14 '18 at 11:33









Ander BiguriAnder Biguri

26.3k105491




26.3k105491













  • I used Hough Lines transform but due to the thickness of black lines there were many small lines detected in one big line and it was difficult to control through the parameters. I have a few questions: 1) why erode with a disk element of size 1? 2) Why not erode with rectangular element? 3) The green line seems to pass over the symbols whereas in the input image black line never crosses any symbol, actually we have to consider them as line segments before and after any symbol. 4) are the blue cross beginning and red cross end of a single line? 5) one small segment in bottom was not detected.

    – SKR
    Nov 15 '18 at 5:21











  • @SKR 1)empirical test showed good result, didnt think of it more. 2) same as 1. 3) The last paragraph is about that. This is not a full solution to your huge project, just a starting point. Your question is too big to answer in SO entirely. People get paid to solve this type of problems. 4) yes, check the code 5) Yes

    – Ander Biguri
    Nov 15 '18 at 9:51



















  • I used Hough Lines transform but due to the thickness of black lines there were many small lines detected in one big line and it was difficult to control through the parameters. I have a few questions: 1) why erode with a disk element of size 1? 2) Why not erode with rectangular element? 3) The green line seems to pass over the symbols whereas in the input image black line never crosses any symbol, actually we have to consider them as line segments before and after any symbol. 4) are the blue cross beginning and red cross end of a single line? 5) one small segment in bottom was not detected.

    – SKR
    Nov 15 '18 at 5:21











  • @SKR 1)empirical test showed good result, didnt think of it more. 2) same as 1. 3) The last paragraph is about that. This is not a full solution to your huge project, just a starting point. Your question is too big to answer in SO entirely. People get paid to solve this type of problems. 4) yes, check the code 5) Yes

    – Ander Biguri
    Nov 15 '18 at 9:51

















I used Hough Lines transform but due to the thickness of black lines there were many small lines detected in one big line and it was difficult to control through the parameters. I have a few questions: 1) why erode with a disk element of size 1? 2) Why not erode with rectangular element? 3) The green line seems to pass over the symbols whereas in the input image black line never crosses any symbol, actually we have to consider them as line segments before and after any symbol. 4) are the blue cross beginning and red cross end of a single line? 5) one small segment in bottom was not detected.

– SKR
Nov 15 '18 at 5:21





I used Hough Lines transform but due to the thickness of black lines there were many small lines detected in one big line and it was difficult to control through the parameters. I have a few questions: 1) why erode with a disk element of size 1? 2) Why not erode with rectangular element? 3) The green line seems to pass over the symbols whereas in the input image black line never crosses any symbol, actually we have to consider them as line segments before and after any symbol. 4) are the blue cross beginning and red cross end of a single line? 5) one small segment in bottom was not detected.

– SKR
Nov 15 '18 at 5:21













@SKR 1)empirical test showed good result, didnt think of it more. 2) same as 1. 3) The last paragraph is about that. This is not a full solution to your huge project, just a starting point. Your question is too big to answer in SO entirely. People get paid to solve this type of problems. 4) yes, check the code 5) Yes

– Ander Biguri
Nov 15 '18 at 9:51





@SKR 1)empirical test showed good result, didnt think of it more. 2) same as 1. 3) The last paragraph is about that. This is not a full solution to your huge project, just a starting point. Your question is too big to answer in SO entirely. People get paid to solve this type of problems. 4) yes, check the code 5) Yes

– Ander Biguri
Nov 15 '18 at 9:51




















draft saved

draft discarded




















































Thanks for contributing an answer to Stack Overflow!


  • Please be sure to answer the question. Provide details and share your research!

But avoid



  • Asking for help, clarification, or responding to other answers.

  • Making statements based on opinion; back them up with references or personal experience.


To learn more, see our tips on writing great answers.




draft saved


draft discarded














StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53294066%2fgeneral-approach-for-extracting-specific-lines-or-line-segments-in-an-image%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown





















































Required, but never shown














Required, but never shown












Required, but never shown







Required, but never shown

































Required, but never shown














Required, but never shown












Required, but never shown







Required, but never shown







Popular posts from this blog

Florida Star v. B. J. F.

Danny Elfman

Retrieve a Users Dashboard in Tumblr with R and TumblR. Oauth Issues