Method of isolating texture areas in images

详细信息    查看全文

• 作者：E. E. Kurbatova (1)
  E. V. Medvedeva (1)
  A. A. Okulova (1)
  
  1. Vyatka State University ; ul. Moskovskaya 36 ; Kirov ; 610000 ; Russia
  
• 关键词：texture segmentation method
• 刊名：Pattern Recognition and Image Analysis
• 出版年：2015
• 出版时间：January 2015
• 年：2015
• 卷：25
• 期：1
• 页码：47-52
• 全文大小：570 KB
• 参考文献：1. Kovtun, I V (2003) Image texture segmentation according to Markov random fields. Upravl. Sist. Mashiny, No. 4. pp. 46-55
  2. Kolodnikova, N V (2004) Review of texture characteristics for pattern recognition. Proceedings of Tomsk State Univ. of Control Systems and Radioelectronics. Automated Systems for Information^Processing, Control and Design. pp. 113-124
  3. Petrov, E P, Medvedeva, E V, Metelev, A P (2011) The way to synthesize mathematical models of videoimages on the base of multidimencial chains. Nelinein. Mir, No. 4. pp. 213-231
  4. Petrov, E P, Medvedeva, E V, Kharina, N L (2011) Mathematical model of digital half-tone Earth images seen from the space. Proc. 2nd All-Russian Sci.-Techn. Conf. 鈥淭opical Problems of Rocket-Space Engineering鈥? Samara Sci. Center RAS, Samara, pp. 179-185
  5. Medvedeva, E V, Petrov, E P (2008) The way to calculate informational characteristics of digital half-tone images. Inf. Komp鈥檡ut. Tekhn. 6: pp. 104-109
  6. Medvedeva, E V, Kurbatova, E E (2011) A two-stage image preprocessing algorithm. Pattern Recogn. Image Anal. 21: pp. 297-301 CrossRef
  7. Khabibi, A (1972) 2D Bayesian image estimation. Trudy Inst. Inzh. Elektrotekhn. Radioelektron. 60: pp. 153-159
  8. Shapiro, L, Stockman, G (2001) Computer Vision.
  
• 刊物类别：Computer Science
• 刊物主题：Pattern Recognition
  Image Processing and Computer Vision
  Russian Library of Science
  
• 出版者：MAIK Nauka/Interperiodica distributed exclusively by Springer Science+Business Media LLC.
• ISSN：1555-6212

文摘

A method of texture segmentation of images is proposed based on Markov random fields. An estimate of the probability of a transition between image elements is used as the texture feature. The method efficiently isolates texture areas with different statistical characteristics and makes it possible to reduce computational expenditures.