I am developing billingual OCR script detection using gabor filter. I am using Accord.net framework , I used gabor filter from Accord and it transformed image into gabor image, but I need feature extraction using gabor filter which can be used to train nural network to identify language of word. Please tell how I can extract feature of image using gabor filter. If any other option available (like aforge or emgu cv ) please tell. Language only C# . Just like this http://www.primaresearch.org/ICDAR2003/Papers/0178_682_ma_huanfeng.pdf Thanks Anil
Asked
Active
Viewed 4,065 times
0
-
here is python & c++ implementation: http://www.eml.ele.cst.nihon-u.ac.jp/~momma/wiki/wiki.cgi/OpenCV/Gabor%20Filter.html; here are c# implementations: http://stackoverflow.com/questions/12878754/how-to-convolve-an-image-with-different-gabor-filters-adjusted-according-to-the http://accord.googlecode.com/svn/trunk/Sources/Accord.Imaging/Filters/GaborFilter.cs – baci Feb 10 '14 at 15:16
1 Answers
0
In order to identify the language, you should extract texture features such as co-occurrence matrices, Tamura features or Gabor Texture Features.
"The Gabor texture features include the mean and the standard deviation of the magnitude of the Gabor wavelet transform coefficients". The idea is to apply different Gabor filters (with different scales and orientations) and then to extract features from the obtained images.
You can find more details here : Gabor feature extraction
