SuDoKu Grabber with OpenCV

In the previous article (Detecting a SuDoKu puzzle in an Image, Part 3), we created the DigitRecognizer class.It had functions for training and classifying images as digits. We also put some pre-processing code to ensure the image is at the center. Now, we’ll combine it with our existing program. We’ll also use the merged lines we created earlier.

Training the classifier

We’ll start where we left the original SuDoKu program. Your last line was probably:

cv::warpPerspective(original, undistorted, cv::getPerspectiveTransform(src, dst), Size(maxLength, maxLength));

Unless you tried hacking a bit yourself!

First thing we’ll do is create a thresholded image. Till now, the text is in black. But we want it to be white. So, we’ll create a duplicate and use an inverted threshold:

Mat undistortedThreshed = undistorted.clone(); adaptiveThreshold(undistorted, undistortedThreshed, 255, CV_ADAPTIVE_THRESH_GAUSSIAN_C, CV_THRESH_BINARY_INV, 101, 1);

I’ve used the adaptive threshold (with a block size of 101 to calculate the dynamic threshold value) to keep the effects of lighting from spoiling the thresholding.

Had I used standard thresholding, it would look something like this:

With the adaptive thresholding, the thresholded image looks like this:

Next, we create an instance of the DigitRecognizer class and train it:

DigitRecognizer *dr = new DigitRecognizer(); bool b = dr-&gt;train("D:/Test/Character Recognition/train-images.idx3-ubyte", "D:/Test/Character Recognition/train-labels.idx1-ubyte");

If you want you can put a check on ‘b’. If true, the training was successful, otherwise its false.

Then, we create a small image that will hold one SuDoKu cell:

int dist = ceil((double)maxLength/9); Mat currentCell = Mat(dist, dist, CV_8UC1);

Now, we iterate through each cell:

for(int j=0;j&lt;9;j++) {     for(int i=0;i&lt;9;i++)     {

Each of these cells correspond to certain pixels. We copy these pixels into currentCell:

        for(int y=0;y&lt;dist &amp;&amp; j*dist+y&lt;undistortedThreshed.cols;y++)         {             uchar* ptr = currentCell.ptr(y);               for(int x=0;x&lt;dist &amp;&amp; i*dist+x&lt;undistortedThreshed.rows;x++)             {                 ptr[x] = undistortedThreshed.at&lt;uchar&gt;(j*dist+y, i*dist+x);             }         }

Once we’ve copied the pixels, we calculate the number of white pixels on it. If the number is less, it’s probably a blank cell. Otherwise, it’s probably a numbered cell.

        Moments m = cv::moments(currentCell, true);         int area = m.m00;         if(area &gt; currentCell.rows*currentCell.cols/5)         {             int number = dr-&gt;classify(currentCell);             printf("%d ", number);         }         else         {             printf("  ");         }

And finally, we end the loops:

    }     printf("\n"); }

Try running the program. You should see the SuDoKu printed on the console.

Accuracy

Well. the recognition isn’t very accurate. I can think of two main factors:

• The training data is for handwritten text
• The algorithm used isn’t great

Other possible things you can use for recognizing the text are neural networks, haar networks or some chain-codes thing (the text on the SuDoKu grid has proper and well defined boundaries).

Summary

Today, we loaded the DigitRecognizer class into the main program and made it recognize digits. And with that, the series comes to an end! Hope you’ve learned a lot over the course of this series.

Back to top

Summary

• We're working on something cool!
• To make this work, we're making a few of (reasonable) assumptions
• There are two key tasks - Extracting the grid and recognizing the digits

59 Comments

1. Posted August 25, 2010 at 6:17 pm | Permalink

What’s the motive behind it ? Just learning to perform basic tasks ?
What will be direct application ?

I mean to initiate discussion here basically… !!!

• Posted August 25, 2010 at 6:34 pm | Permalink

Hi! Motive? Well.. I’ve seen a few people ask for it on forums. So thought I’d put up my take on the problem.

Also, doing things that one can relate to are often much better for learning. Take line detection for example. Tell a person about line detection and he won’t be impressed. Relate it to a sudoku puzzle and they’ll remember it forever!

I think I’ll end this series with a SuDoKu solver. Take a snap of a puzzle in your newspaper and it shows your the solution

2. kay_hao
Posted August 25, 2010 at 7:16 pm | Permalink

Wonderful idea!
I will follow u!

3. Ankit Malpani
Posted August 26, 2010 at 1:54 pm | Permalink

innovative idea man !

problems : what if there are 2 squares surrounding the actual puzzle, i think in sudoku’s which come in hindu, there are 2 enclosing squares. So, in step 2- it may detect the outer square and then dividing into 9 lines horizontally and vertically may not give you the actual grid right?

waise, once this is done , how bout adding some more code to solve the puzzle ? [atleast the easy ones for a start]

• Posted August 28, 2010 at 8:30 am | Permalink

Hmm.. Those might be a problem.. Will have to think of how to work with such pictures. Got any ideas? Yes, the app will solve a puzzle once it has successfully recognized it!

4. champie
Posted September 4, 2010 at 6:12 pm | Permalink

Hey you have a nice blog!
Can you give me an Ideas and what do i need in detecting available parking slots??? and object differencing( car, people and other objects) some functions/methods??.

Thank you so much , you’re so genius!!

5. Ymehdi
Posted September 15, 2010 at 5:41 pm | Permalink

Hello Boss,
It’s so useful blog, I thank you for that,
the problem of merging line poste above is so interesing. may I have the complete source code for meging lines closed by ?
your help is greatly appreciated
thanks alot
Ymehdi

• Posted September 20, 2010 at 12:24 am | Permalink

Hi! The code for merging close by lines is right here

6. Y.M
Posted October 12, 2010 at 3:46 am | Permalink

Great work boss

I’ll steal time to look at this valuable blog

bravo

7. Abhi
Posted October 12, 2010 at 8:28 am | Permalink

Hi Utkarsh,
I find your posts really innovative and informative.
Please keep up the good work. Your thought process and the clarity of your explanation is worth admiring!
Cheers!!

• Posted October 12, 2010 at 4:50 pm | Permalink

Glad you liked this stuff!

8. arash
Posted December 26, 2010 at 8:23 pm | Permalink

thank you so much for your very useful blog and tutorial
could you say more about training procedure?
in this line of your code i got error.
i should change the path “D:/Test/Character Recognition/train-images.idx3-ubyte”, “D:/Test/Character Recognition/train-labels.idx1-ubyte” to what?

9. Posted December 27, 2010 at 1:27 am | Permalink

Hi, Your tutorials are great. Thanks a lot. I am developing an ancient coins recognition system using opencv for my undergraduate final year project these days. I have a small question on this tutorial.
Under “Finding the biggest blob” sub title

uchar *row = outerBox.ptr(y);

in this code, how did you get the ptr?

Thanks

• Posted January 11, 2011 at 7:38 pm | Permalink

Hi! I’m not sure if I understood your question. What do you mean by get the ptr? Its a function in the new C++ interface.

• Nadeeshani
Posted January 12, 2011 at 10:11 pm | Permalink

Do I have to use any include directive to use ptr? for the variable outerBox, ptr is not listed. When building the project… I get this error “struct_iplimage has no member named ‘ptr’ ”
for

uchar *row = dilImg-&gt;ptr(y);
• Nadeeshani
Posted January 12, 2011 at 10:43 pm | Permalink

Still I couldnt find how to use new c++ interface.

That line

uchar *row = dilImg-&gt;ptr(y);

is equal to this right?

uchar *row = (uchar*)(dilImg-&gt;imageData + y*dilImg-&gt;widthStep);

One more problem is cvFloodFill doesnt return a int value. It’s a void function. But this is something you have written

int area = floodFill(outerBox, Point(x,y), CV_RGB(0,0,0));

Can we return an int if we use new c++ interface?

10. arash
Posted December 27, 2010 at 6:42 pm | Permalink

thank you again:)the result of the code you have written is something like this:
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
could you please explain?

• Posted January 11, 2011 at 8:19 pm | Permalink

Hmm… It seems like its not recognizing the digits properly.

11. Ashwin
Posted January 6, 2011 at 4:48 am | Permalink

Hi,

Good stuff, I was planning to do something similar for Android phone i.e. take a snapshot of puzzle and we would provide the solution for it.
Would definitely use your idea for my application.

cheers!!

• Posted January 7, 2011 at 2:01 pm | Permalink

Great! All the best!

12. Posted January 11, 2011 at 6:20 am | Permalink
• Posted January 11, 2011 at 9:37 am | Permalink

Yup! In fact, there are dozens of similar apps for android and iPhone out there.

13. Raingo
Posted January 15, 2011 at 1:43 pm | Permalink

Thank you very much for so many tutorials in this website, and it’s really useful to read through every tiny project.

Just one advice for the “Finding the approximate bounding box” stage, I find one bug in the algorithms.

when traverse through the image from the center in four directions, I think it is not an appropriate way to sum whole row or whole col, or this process will have no effect on the cell whose edges is full of original edge lines.

Maybe the differences between your algorithm and my propose will illustrate this problem:

 int sumtemp=0; if(rowBottom==-1) { sumtemp=0; for(int j(img.cols-i);j&lt;i;++j) sumtemp+=img.at(i,j); if(sumtemp&lt;thresholdBottom || i==img.rows-1) rowBottom=i; }   if(rowTop==-1) { sumtemp=0; for(int j(img.cols-i);j&lt;i;++j) sumtemp+=img.at(img.rows-i,j); if(sumtemp&lt;thresholdTop || i==img.rows-1) rowTop=img.rows-i; }   if(colRight==-1) { sumtemp=0; for(int j(img.rows-i);j&lt;i;++j) sumtemp+=img.at(j,i); if(sumtemp &lt; thresholdRight|| i==img.cols-1) colRight = i; }   if(colLeft==-1) { sumtemp=0; for(int j(img.rows-i);j&lt;i;++j) sumtemp+=img.at(j,img.cols-i); if(sumtemp &lt; thresholdLeft|| i==img.cols-1) colLeft = img.cols-i; }

however, my propose leads to a more difficult problem, since it does not works for some certain digits like 7 or 5 for their specific shape!

Maybe the search algorithm is not proper in this case!

Thank you very much again!

14. cloutak
Posted January 17, 2011 at 4:27 pm | Permalink

Sir can you please help me,im gettiing a lot of errors when compiling the different parts from the sudoku grabber explanation of program.
can you please upload the complete version in a zip or a rar file and post it here ?

15. Posted February 15, 2011 at 11:59 pm | Permalink

I guess the new version of Google Goggles on Android detects and solves the Sudoku. I guess you can as well add that feature :p. I love the work that you are doing by sharing loads of stuff. I am working on computer vision problems, so I find your stuff very relevant to my area of research. Good job!

• Posted February 16, 2011 at 7:47 am | Permalink

Glad you liked the site! Though I think the Google people copied the sudoku thing from here

16. René
Posted March 4, 2011 at 9:20 am | Permalink

Thanks a lot, I’ve learned so much from your articles, I followed carefully every explanation, and thinks worked (was much better than I was expecting for).
http://www.cec.uchile.cl/~rene.tapia/images/runningProgram.png
Again, you did a very good work, congratulations.

17. Posted March 4, 2011 at 7:27 pm | Permalink

Excellent tutorial, thanks. Many of these techniques would also to apply to my problem domain, which is scanning barcode images. Although I can think of many more interesting problems to solve this way!

Thanks for the great tutorial.

18. arkiazm
Posted March 7, 2011 at 10:09 pm | Permalink

hi Utkarsh
urs is really a nice blog… with excellent tutorials… thanks a lot…
i got one doubt regarding finding biggest blob and floodfill…
i am using python and opencv… not c++..
it takes a lot of time for finding biggest blog…
any way to reduce it?
thanks in advance

• Posted March 11, 2011 at 4:10 pm | Permalink

Try this – find contours in the thresholded image. The contour with the largest bounding box is the biggest box. I’m guessing this will be a lot faster. Let me know if this works!

• Linus
Posted April 23, 2011 at 3:42 am | Permalink

Would you mind writing a short example using cvFindContours?

• Posted April 29, 2011 at 9:10 pm | Permalink

Already did. Search for “introduction to contours” on AI Shack

19. Richárd Szabó
Posted April 13, 2011 at 7:52 pm | Permalink

It would be a good idea to make your final source code available.

Anyway it is an excellent tutorial.

• Posted April 29, 2011 at 9:23 pm | Permalink

I have that on my todo list. I’ve just been procrastinating too much lately. Should be up in sometime though

20. Richárd Szabó
Posted April 14, 2011 at 5:21 pm | Permalink

“Then we convert it into IplImage to use cvSum. (For some reason, there is no C++ version for this function on my system)”

There is a sum function in C++ version: http://opencv.willowgarage.com/documentation/cpp/core_operations_on_arrays.html?highlight=sum#sum

• Posted April 29, 2011 at 9:27 pm | Permalink

Perfect – you can use this. While writing this tutorial, I couldn’t find this. Thanks for pointing this out!

21. Palmendieb
Posted April 18, 2011 at 1:49 am | Permalink

Hi Utkarsh

First off all, I just need to gratulate to this grade Blogs. I just tried yout tutorial for sudoku recognisation and solving. I have some trouble withe that. I use OpenCV 2.1.0 and it seems there are some problems with the header inclusions and the cvcore source? So could you please have a log for that?

Thanks for that…

regards

Palmendieb

• Posted April 29, 2011 at 9:30 pm | Permalink

Well, look at the OpenCV site. I think they have a setup procedure. Use that and the installation procedure on AI Shack. You should be able to get it to work.

22. Donny
Posted April 26, 2011 at 7:59 pm | Permalink

hi.thank you for your article. currently i’m working on pattern recognition as a newbie.this tutorial does help me in understand it.thanks.
great work you have there.

• Posted April 29, 2011 at 9:06 pm | Permalink

Glad it helped you!

23. mala
Posted May 30, 2011 at 3:53 pm | Permalink

really great tutorial!
..but would you please make the source code available here??

24. Hadi
Posted June 1, 2011 at 11:35 pm | Permalink

Hi
Nice tutorial !! i really enjoyed

25. Nguyen Xuan Hoc
Posted June 25, 2011 at 6:47 am | Permalink

Hi, the first, thanks a lot. I find very helpful from your website.
I’m a new meb in openCV . I have some question about this article.

1,What is the line contain?

2,how you can calculate m and c?
float m = -1/tan(line[1]);
loat c = line[0]/sin(line[1]);

3, what is the relation of line[0] and line[1]

Thanks in advance!

• Posted June 25, 2011 at 9:46 pm | Permalink

line[0] is $\rho$ and line[1] is $\theta$. $\rho$ and $\theta$ describe a line (that was detected). Does it make sense now?

• Nguyen Xuan Hoc
Posted June 27, 2011 at 10:48 am | Permalink

Thanks you for your support!

26. Sai
Posted July 3, 2011 at 12:25 pm | Permalink

How about setting the border of the Sudoku puzzle via ROI technique? What is the disadvantage compared to this method?

• Posted July 4, 2011 at 12:39 am | Permalink

What do you mean the ROI technique? From what I guess, you’re making a “mask” and using it again and again to “select” cells from the grid. That way, you’d have to do math because the cells aren’t in perfect shape.

27. Sai
Posted July 5, 2011 at 1:08 am | Permalink

Instead of finding a largest blob, and then finding the individual grid lines, why not set a region of interest and then find the individual grid lines?

• Posted July 13, 2011 at 9:09 am | Permalink

How would you set the ROI? How do you figure out the coordinates of the grid?

28. kasun
Posted July 5, 2011 at 8:29 am | Permalink

Hi,
I red many of u’r tutorials and you are doing a grate job. Thankz for helping us through this blog. Do u know how to measure the orientation of an object from opencv. As an example, assume that there is a black colour mark is on the robot top surface. so i need to measure the angle of the robot. Do u have a method or a sample code for this??
thank you

• Posted July 13, 2011 at 9:10 am | Permalink

Working on something like that. Stay tuned!

29. Franz Wong
Posted August 8, 2011 at 2:08 pm | Permalink

Nice article. As a beginner, Sudoku is a good example for me to learn computer vision.

I changed the logic a little bit. I used cvFindContour to get the points of largest blob. And then found out the corner points which were nearest the max/min of x and y of the points.

30. Gabor
Posted July 29, 2011 at 4:56 pm | Permalink

Hi,

I fixed it, just switched back from openCV 2.2 to openCV 2.0, and that was it.
Matrix declaration itself did not want to work, by “Mat example;”, i got an error in mat.cpp, which i couldnt debug.
But now it works!

One Trackback

• By How I’ll be wasting my time… on March 14, 2012 at 4:42 pm

[...] quick search reveals all manner of interesting projects and sample code hosted online ranging from visual Sudoku grid ‘grabbing’ to creative uses of augmented reality that make use of the device’s other forms of input such as [...]