Karmaşık Yayınımla İmge Gürültü Azaltma

Bu çalışmada girdi imgesindeki Gauss gürültüsünün giderilmesi için iyileştirilmiş karmaşık yayınım yaklaşımları önerilmiştir. Karmaşık yayınımda gerçel kısım alçak geçiren süzgeç davranışı gösterirken, sanal kısım ise yüksek geçiren süzgeç davranışı göstermektedir. Böylece imgedeki gürültüyü gerçel kısımdaki süzgeç azaltırken imgenin yapı bilgisi sanal kısımdaki süzgeçle korunmaktadır. Önerilen yöntemlerde, doğrusal yayınımı dikkate alan ve yönden bağımsız olarak çalışan ısı denkleminde ve doğrusal olmayan yayınımı göz önüne alan Perona-Malik yaklaşımında düzenlileştirme terimine ek olarak uygunluk terimi de kullanılmıştır. Uygunluk terimi sonuç imgesinin yapı bilgisini daha iyi korumuştur. Diğer yandan gürültü standart sapması yarı otomatik olarak kestirilmiş,    uygunluk terimindeki Lagrange çarpanı da her iterasyonda optimize edilmiş ve böylece karmaşık yayınım yaklaşımlarıyla girdi imgesindeki toplamsal gürültü azaltma başarımı iyileştirilmiştir. Önerilen yöntemlerdeki bu başarım, gürültü standart sapmasının fazla yüksek olmadığı durumlarda,  hem nitel hem de nicel sonuçlarla desteklenmiştir.

Image Noise Reduction via Complex Diffusion

In this study, improved complex diffusion approaches are proposed for eliminating Gaussian noise in the input image. In the complex diffusion, meanwhile, the real component behaves as a low pass filter and the imaginary component behaves as a high pass filter. Thus, while the filter in the real component reduces the noise in the image, the filter in the imaginary component protects the structure information of the image. In the proposed methods, the fidelity term is used in the heat equation approach, where the linear diffusion is taken into account and operations is isotopically performed, and in the Perona-Malik approach, where the nonlinear diffusion is considered, addition to the regularization term. The fidelity term better preserves the structure information of the resulting image. On the other hand, the noise standard deviation is semi-automatically estimated, the Lagrange multiplier in the fidelity term is also optimized in each iteration and thus the additive noise reduction performance in the input image is improved via complex diffusion approaches. This performance in the proposed methods is supported by both qualitative and quantitative results when the noise standard deviation is not too high.

___

  • Aström, F. 2015. Variational tensor-based models for image diffusion in non-linear domains. Linköping University, Ph.D. Thesis, 170s, Sweden.
  • Gilboa, G., Sochen, N., and Zeevi, Y. Y. 2006. Variational Denoising of Partly Textured Images by Spatially Varying Constraints. IEEE Transactions on Image Processing, 15.8(2006), 2281-2289.
  • Immerkaer, J. 1996. Fast Noise Variance Estimation. Computer Vision and Image Understanding, 64.2(1996), 300-302.
  • Tikhonov, A. N. 1963. Solution of Incorrectly Formulated Problems and the Regularization. Soviet Math. Dokl., 4(1963), 1035-1038.
  • Perona, P., Malik, J. 1990. Scale-Space and Edge Detection Using Anisotropic Diffusion. IEEE Transactions on Pattern Analysis and Machine Intelligence, 12.7(1990), 629-639.
  • Weickert, J. 1998. Anisotropic Diffusion in Image Processing. Teubner-Verlag, Stuttgart, Germany, 170s.
  • Tschumperle, D., Deriche, R. 2005. Vector-Valued Image Regularization with PDE’s: A Common Framework for Different Applications. IEEE Transactions on Pattern Analysis and Machine Intelligence, 27.4(2005), 506-517.
  • Buades, A. Coll, B., Morel, J. -M. 2005. A Non-Local Algorithm for Image Denoising. Computer Vision and Pattern Recognition, 2(2005), 60-65.
  • Gilboa, G., Osher, S. 2007. Nonlocal Linear Image Regularization and Supervised Segmentation, Multiscale Modeling & Simulation, 6.2(2007), 595-630.
  • Gilboa, G., Sochen, N., Zeevi, Y. Y. 2004. Image Enhancement and Denoising by Complex Diffusion Processes. IEEE Transactions on Pattern Analysis and Machine Intelligence, 26.8(2004), 1020-1036.
  • Tschumperle, D. 2006. Fast Anisotropic Smoothing of Multi-Valued Images Using Curvature-Preserving PDE's. International Journal of Computer Vision 68.1 (2006), 65-82.
  • Vese, L. A., Guyader, C. L. 2015. Variational Methods in Image Processing. Chapman and Hall/CRC, 386s.
  • Araújo, A., Barbeiro, S., Cuesta, E., Durán, A. 2017. Cross-Diffusion Systems for Image Processing: I. The Linear Case. Journal of Mathematical Imaging and Vision, 58.3(2017), 447-467.
  • Araújo, A., Barbeiro, S., Cuesta, E., Durán, A. 2017. Cross-Diffusion Systems for Image Processing: II. The Nonlinear Case. Journal of Mathematical Imaging and Vision, 58.3(2017), 427-446.
  • Ulu, A., Dizdaroğlu, B. 2016. Variational Additive Noise Removal of Color Images. Signal Processing and Communication Application Conference (SIU), 16-19 May 2016, Zonguldak.
  • Wang, Z., Bovik, A.C., Sheikh, H. R., Simoncelli, E.P. 2004. Image Quality Assessment: From Error Measurement to Structural Similarity. IEEE Transactions on Image Processing, 13.4(2004), 600-612.
  • Tschumperle, D. 2004. The CImg Library. http://cimg.eu/. (Erişim Tarihi: 15.03.2018).
  • Sutour, C., Aujol, J. F., Deledalle, C. A. 2016. Automatic Estimation of the Noise Level Function for Adaptive Blind Denoising. Signal Processing Conference (EUSIPCO), 28 August - 2 September 2016, Budapest, Hungary, 76-80.