jm's profile - activity

Yeah it seems good, sorry for the late reply.

Since your application is intended for android - and tesseract is built in c++, You can use Android NDK for using (the existing c++) tesseract in your android project. (check last link)

Check out the following resources. http://rmtheis.wordpress.com/2011/08/06/using-tesseract-tools-for-android-to-create-a-basic-ocr-app/

http://kurup87.blogspot.in/2012/03/android-ocr-tutorial-image-to-text.html

http://droidcomp.wordpress.com/2012/08/04/building-the-tesseract-ndk-library-for-android/

I just googled about "wordsnap ocr" -> it seems like a single word ocr android app. and GOCR -> seems to be a linux-friendly ocr tool.

since you asked for something light, simple and SPECIALIZED for number recog, go for a NN OCR, please refer to this -> http://www.nithinrajs.in/ocr-using-artificial-neural-network-opencv-part-1/

I've switched to Code::Blocks, works perfectly.

Although I wonder if the VS2012 problem could be solved. I hope this post - helps anyone else, if they face the same problem.

Another person has faced this problem, it seems visual studio 2012, is the culprit (i don't know why?)

Here -> http://answers.opencv.org/question/4283/program-crashes-during-findcontours-execution/

You said you do not want to use Tesseract "because it will detect only an id number", regardless, it is still a number, and recognition of a number would be possible by OCR,

You can either implement your own OCR, or use an existing OCR tool/api.

I suggest you go ahead with Tesseract , since it will be better than implementing your own (it is tested and used a lot)

Unless, you are using OCR for a specific purpose i.e. (for handwriting, etc) - you would need to use your own implementation - though i've heard it can be done with Tesseract as well.

Please read the following :

http://stackoverflow.com/questions/9404319/optical-character-recognition-android-with-opencv

http://kurup87.blogspot.in/2012/03/android-ocr-tutorial-image-to-text.html

-> i've just tried with openCV 2.4.8 & Visual Studio 2012, on a 64bit release environment; and the same error comes up.

tried with 32bit as well; same problem.

Did anyone face the same problem?

On my machine (PC) - 2.4.6; my development environment is - visual studio 2012 - release configuration - 64bit On my other machine (laptop) - 2.4.6 my development environment is - visual studio 2008 - release configuration - 32bit.

I think there is some issue with visual studio 2012 coupled with OpenCV 2.4.6, for drawContours(); I will update my openCV version and update the post.

This is my full code -> http://pastebin.com/dZ7a3ted Here is the program's execution output -> http://pastebin.com/HaftZ9gV (though, of course, you'd have to pass the an input image as run-time parameter.)

I do agree, it should be perfectly fine (execution) - I even tried it on another machine. And it worked fine.

Total number of contour hierarchy is 120. Yes.

Hello, I am trying to draw contours, but it throws an assertion error

"OpenCV Error: Assertion failed (0 <= contourIdx && contourIdx < (int)last) in unknown function, 
file ..\..\..\src\opencv\modules\imgproc\src\contours.cpp, line 1810"

My input image is :

Here's a code snippet of what I'm doing :

Mat input_image = imread(filename, 0);

Mat canny_op;
Canny(input_image, canny_op, 100, 200, 3);

vector<vector<Point> > contours;
vector<Vec4i> hierarchy;
findContours( canny_op, contours, hierarchy, CV_RETR_LIST, CV_CHAIN_APPROX_NONE, Point(0, 0) );

cout << "Size of contours [vector] is " << contours.size() << endl;
cout << "Size of hierarchy [vector] is " << hierarchy.size() << endl;

// Draw contours
Mat drawing = Mat::zeros( input_image.size(), CV_8UC3 );
for( int i = 0; i< contours.size(); i++ ) {
cout << i << endl;
    Scalar color(0,0,255);
    drawContours( drawing, contours, i, color, 2, 8, hierarchy, 0, Point() );
}

showImage("output of drawing",drawing);

Here is the output of Canny :

And here is the output of the program :

Size of contours [vector] is 160
Size of hierarchy [vector] is 120
0
1
2
...
120
OpenCV Error: Assertion failed (0 <= contourIdx && contourIdx < (int)last) in un
known function, file ..\..\..\src\opencv\modules\imgproc\src\contours.cpp, line
1810

I think you asked for the installation files.

quick start guide (windows) [Pre-built Libraries] -> http://docs.opencv.org/doc/tutorials/introduction/windows_install/windows_install.html#windows-install-prebuild

openCV 2.4.8 -> http://sourceforge.net/projects/opencvlibrary/files/latest/download?source=files

If you are using VS then this would be useful -> http://docs.opencv.org/doc/tutorials/introduction/windows_visual_studio_Opencv/windows_visual_studio_Opencv.html#windows-visual-studio-how-to

Using a NN (backpropagation),

You could try the these demo's www.nithinrajs.in/ocr-using-artificial-neural-network-opencv-part-1/ http://www.nithinrajs.in/ocr-using-artificial-neural-network-opencv-part-2-preprocessing/ http://www.nithinrajs.in/ocr-artificial-neural-network-opencv-part-3final-preprocessing/

I successfully got them working.

Additional reference material: http://arxiv.org/ftp/arxiv/papers/1211/1211.4385.pdf

Hi,

I am trying to grid a tabular image, my source image is -> bit.ly/1gCDxk7

What i intend to do is - linearly iterate through the grid (row, column wise), and capture ROI within the box's edges - like the green box in this image -> bit.ly/19RnNuu

I was looking at this solution -> http://stackoverflow.com/questions/10196198/how-to-remove-convexity-defects-in-a-sudoku-square/11366549#11366549

But I am unsure if that is the right approach for this. (I would be using a scanner, and not a camera)

Hello, I'm trying to identify filled circles from non-filled circles (example pic - http://bit.ly/17bvGcS )

circle-filled.jpg circle-empty.jpg

How do I go about doing this?0

I tried searching, and found this page on Image Moments, however, the code as shown in the example doesn't work for me.

I'm using OpenCV 2.6.4 in a Visual Studio 2012 C++ Environment.