Rotation of the Anisotropic Gaussian Kernel for the Improvment of the Text Skew Identification

Izdanje: Naučna konferencija Uniteh 2010

Oblast: Computer Technologies

Stranice: 461-466

Apstrakt:
In this paper, an approach to algorithm with the implementation of the anisotropic Gaussian kernel is presented. As a result of algorithm, the growing region around text is created. To improve skew identification in text, anisotropic Gaussian kernel is rotated. Furthermore, algorithm is evaluated under different text samples. Results are given and comparative analysis is made. From the obtained results, the optimal value of the parameters defining rotated anisotropic Gaussian kernel dimension is proposed.
Ključne reči: OCR, Gaussian kernel, region growing algorithm, skew rate
Priložene datoteke:
Kategorije objave: Radovi na konferenciji Unitech 2010, Gabrovo, Bugarska

Preuzimanje citata:

BibTeX format 
@article{article,
  author  = {D. Brodić, Z. Milivojević and D. Milivojević}, 
  title   = {Rotation of the Anisotropic Gaussian Kernel for the Improvment of the Text Skew Identification},
  journal = {Naučna konferencija Uniteh 2010},
  year    = 2010,
  pages   = {461-466}}
RefWorks Tagged format 
RT Conference Proceedings
A1 Darko Brodić
A1 Zoran Milivojević
A1 Dragan Milivojević
T1 Rotation of the Anisotropic Gaussian Kernel for the Improvment of the Text Skew Identification
AD Naučna konferencija Unitech, Gabrovo, Bugarska
YR 2010
Unapred formatirani prikaz citata
D. Brodić, Z. Milivojević and D. Milivojević, Rotation of the Anisotropic Gaussian Kernel for the Improvment of the Text Skew Identification, Naučna konferencija Unitech, 2010