A Denoising of Images in Frequency Domain Using Optimized Neuro Hybrid Fuzzy Filter
M. Sindhana Devi1, M. Soranamageswari2
1M. Sindhana Devi, Research Scholar, Department of Computer Science and Research Development, Government Arts College, Coimbatore (Tamil Nadu), India.
2Dr. M. Soranamageswari, Assistant Professor, Department of Computer Science and Research Development, Government Arts College, Coimbatore (Tamil Nadu), India.
Manuscript received on 24 April 2019 | Revised Manuscript received on 02 May 2019 | Manuscript Published on 08 May 2019 | PP: 445-452 | Volume-7 Issue-5S3 February 2019 | Retrieval Number: E11790275S19/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In image processing domain, the image is corrupted by several types of noises especially when the final product is used for edge detection, image segmentation and data compression. So image de-noising has become a very essential exercise all through diagnose. In the gray scale image the impulse noise can be removed by using the neuro fuzzy (NF) network based impulse noise filtering approach. The each NF filtering approach is a first order sugeno type fuzzy inference system. Since the Sugeno type is not intuitive technique and it also less accurate. In order to improve the accuracy of the NF filtering approach, utilized the hybrid technique of Mamdani and Sugeno based fuzzy interference system approach and an optimized intelligent water drop technique (IWD) in the spatial domain. However the hybridized Sugeno-Mamdani based fuzzy interference system implemented in the spatial domain this leads to reduce the accuracy of the removal of noise. Also the IWD has the issues in selection domination and in ability to handle indistinguishable fitness. In order to overcome these issues in this paper proposed a denoising of images in frequency domain using optimized neuro hybrid fuzzy filter. The optimised Fuzzy intelligence noise filters approach the pixels in the image are converted into frequency domain by using discrete Fourier transform. The noise present in the pixels is filtered by using fuzzy intelligence noise filter. The modified intelligent water drop algorithm applied for frequency domain. After that by using inverse discrete Fourier transform frequency domain pixels of images are converted to original image. In that optimised method noise present in the images are fully eliminated. The performance of the proposed approach evaluated in terms of Mean Squared Error (MSE) and Peak Signal–to–noise Ratio (PSNR), Structural Similarity (SSIM), Mean Absolute Error (MAE) and Maximum Difference value (MD).
Keywords: Fuzzy Inference System, Neuro Fuzzy System, Sugeno Type, Intelligent Water Drop Algorithm.
Scope of the Article: Frequency Selective Surface