Home > Information technology essays > Image denoising

Essay: Image denoising

Essay details and download:

  • Subject area(s): Information technology essays
  • Reading time: 8 minutes
  • Price: Free download
  • Published: 15 October 2015*
  • File format: Text
  • Words: 2,134 (approx)
  • Number of pages: 9 (approx)

Text preview of this essay:

This page of the essay has 2,134 words. Download the full version above.

Abstract: The presence of impulse noise is one of the most challenging problems in many digital image processing applications. Hence, the removal of such impulse noise by median based filter is the most widely used techniques. These Median filters cannot differentiate noise and impulse information at higher noise levels. Hence, extensive research is carried out in this area to remove noise in highly degraded images. In this paper provides an comprehensive literature survey on various filtering methods for removal of salt and pepper noise.
Keywords’ Image Denoising, Median Filter, Salt and Pepper noise, Decision Tree Based Median filter.
The most challenging problem in image pre-processing is Image Denoising. The two major sources of noise introduced in digital images are communication channel and imaging sensor. The noise can be segregated into various types such as Gaussian noise, impulse noise (Salt and Pepper noise and random valued noise), Shot noise, Quantization noise and anisotropic noise. Among these noises Salt and Pepper noise plays significant role in quality of images. The intensity values of salt noise are 0 and pepper noise are 255. However the major challenges in designing a filter for removing a salt and pepper noise lies in differentiating the image features as appear as a salt and pepper from the actual salt and pepper noise. In recent years,efficient techniques are developed for the above. However none of these techniques could completely remove salt and pepper noise by retaining the image features that appear as salt and pepper. Hence it is necessary to develop a filtering algorithm for removal of salt and pepper noise while preserving the desired information. However the major challenge lies in removing salt and pepper noise at high noise density levels. In this paper, the complete literature review is carried out on various filtering techniques for removing Salt and Pepper noise from image. The contents of the paper are as follows, section II discuss about literature survey of various filtering techniques, section III conclusion of the work.
M.Jayamanmadharao et al (2013) proposed a hybrid approach for removal of impulse noise from digital images. The pixels were classified as noisy and noise free- pixels based on the intensity values. Threshold was generated by statistical parameter values instead of setting or assigning a threshold for sliding windows. Initially the weighted mean and standard deviation values were calculated and the noisy pixel was determined (when the difference between center pixel and the weighted mean greater than the threshold value). The hybrid approach is achieved better Peak Signal-to-Noise Ratio (PSNR) for high Mean Square Error (MSE) value.
Karthik et al (2014) proposed a Modified Cascaded Filter for removal of salt and pepper noise from the gray and color images. Modified Cascaded Filter consists of Decision-Based Median Filtering (DMF) and Un-Symmetric Trimmed-Mean Filtering (UTMF).The corrupted pixels were replaced by the median value of the neighbourhood pixel in Standard median filter. The corrupted pixel was replaced by the median value of nearby pixel and the uncorrupted pixel values were retained, in DMF. The cascaded connection of DMF and UTMF were used to remove the salt and pepper noise with a noise density at high levels and it produced high peak signal to noise ratio. Thus the proposed Filter is achieved better performances at high noise levels.
Sajan P. Philip et al (2013) proposed a Decision Tree Based Algorithm for removal of salt and pepper noise and the random valued noise in images. The Decision tree was developed by the following levels: Isolation, Fringe and Similarity Module (SM). The noisy image was given to the decision-tree impulse detector. If the pixel was corrupted by the noise then it was given to the edge preserving filter otherwise directly given as a output image. The Decision Tree Based Algorithm provides a low computational complexity and less hardware cost.
Rutuja N. Kulkarni, P.C. Bhaskar (2014) proposed a Decision Based Median Filter for extract the impulse noise from the images. This method was constructed by noise detection and noise filtering. The current processing pixel was replaced by mean when all the elements were 0’s and 255’s. If the elements in the selected window other than 0’s and 255’s then the processing pixel is replaced by median value. The Decision Based Median Filter provides high PSNR than the existing filters.
G. V. Manoj Gowtham and R. Sathish Kumar (2014) proposed a median filter for removing impulse noise from the images. The insertion of 0’s in row and column of 0 and 256 by method of padding array. This algorithm was designed by noisy candidate detection and design process. The processing steps of design process were adding noise to the image, processing of image and read the output image. The proposed median filter is removed all the impulse noise in images .
Manju Chouhan and C.D Khare (2013) proposed a median filter for removal of salt and pepper noise from the images. This algorithm used 4X4 images out of the input image and processed the pixels as a 3X3 matrix. The four median values were calculated simultaneously and the time consumption was reduced. This algorithm was implemented in FPGA using VHDL. Based on experimental results the Median Filter with parallelism achieved better image enhancement factor.
Sukhwinder Singh and Neelam Rup Prakash (2014) proposed a Modified Adaptive Median Filter for removal of salt and pepper noise from the image. This Rank Order Absolute Difference (ROAD) defines how the intensity of processing pixel is differing from their most neighbouring pixel. The ROAD values were calculated for particular window. If the ROAD value of window is greater than the predefined threshold value then the window was considered as a noisy and a filter is applied for noisy window or otherwise the windows were retained. The Modified Adaptive Median Filter is achieved better PSNR value and image enhancement factor.
P. LATHA et al (2014) developed an Adaptive Threshold Algorithm for removal of impulse noises from the video. Adaptive Threshold Algorithm was compared with other non linear filters (Median Filter, Adaptive Median Filter, etc.). The Adaptive Threshold Algorithm has divided into two parts such as noise detection and adaptive filter. The steps involved in these methods were as follows: noisy video file was converted into noisy frames, noise counter was initialised, the local mean of processed and unprocessed pixels were computed and the threshold value was determined by the edge parameter value and noise corner value. If the processing pixel value was greater than the threshold then it was called a noise free pixel or otherwise noisy pixel. It was replaced by the mean of the processed pixels. The Adaptive Threshold Algorithm yields a better and fast filtering of noise at high densities than the standard and modified version of the Median filter.
Okuwobi Idowu Paul and Yonghua Lu (2014) developed wavelet thresholding techniques for image denoising. Wiener filter was removing all the noises in the smooth region but the performances of the edges were very poor. Wavelet with wiener filter had preserved more information than the thresholding with wiener filter. Thus the wiener filter with wavelet transform provides better performance than the wiener filter with thresholding and Fourier transform.
A.Anilet Bala et al (2014) introduced a new method called image denoising method by curvelet transform and wiener filter. The Curvelet transform has consists of four different transforms namely, 2D wavelet transform, 2D fast transform, Random transform and Ridgelet transform. At first apply a noisy image into a Wiener filter then next the filtered image was given to the curvelet transform and finally the denoised output was produced. This algorithm is better than the existing transform techniques and also it is better for YCbCr than the RGB color model.
Suman Shrestha (2014) proposed a new Adaptive Based Median Filter for image denoising. Adaptive Decision Based Median Filtering Algorithm was developed by modifying the decision based median filter and adaptive median filter. The Decision Based Median Filter produced the better visual clarity at low noise density levels than the conventional techniques. At high noise density levels, Adaptive Decision Based Median Filter are used it gives better image enhancement factor.
Gobu.C K and Priya.R (2014) proposed realization of VLSI architecture for Decision Tree Based Denoising Method in images. The impulse noise was classified into two types such as salt and pepper noise and random valued noise. The random valued noise was removed by the Decision Tree Based Algorithm. Odd and even line buffers were used instead of full images. So, computational complexity was very less. Thus the Decision Tree Based Algorithm is implemented in FPGA and proved better performance with less cost.
Priya Kapoor and Samandeep Singh (2014) proposed an Improved Modified Decision Based Filter to remove high density impulse noise. The Decision Based Filter was replaced the noisy pixel and retained the noise free pixel. The Improved Decision Based Switching Median Filter is used to find global mean for highly corrupted images. The proposed filter is produced a good visual quality than the existing filters (standard median filter, switched median filter).
S.Surendhar et al (2014) developed a denoising architecture for removing impulse noise in image. Here, cloud noise filtering algorithm was promoted to remove the impulse noise. From their experimental results demonstrate that cloud based filtering method can obtained better performances in terms of both subjective and objective evaluations than the other techniques and it is provide low computational complexity and low hardware cost.
S. Athi Narayanan et al (2013) developed an Iterative Trimmed Median Filter and an Adaptive Window Trimmed Median Filter for removal of salt and pepper noise from the images. The Iterative Median Filter works as follows: If the noise pixels neighbourhood window was totally affected by noise, such pixel values were retained in the current iteration and it was processed on the next iteration. The Adaptive Window Trimmed Median Filter works as follows: If the noise pixels neighbourhood window was totally affected by noise then the neighbourhood window size was adaptively increased until an image pixel was found the neighbourhood value. Thus the Iterative Trimmed Median Filter and an Adaptive Window Trimmed Median Filters are effectively restored the images by high density salt and pepper noise.
Vivek Chandra et al (2013) proposed an Adaptive and Unsymmetric Trimmed Median Filter for removal of high density salt and pepper noise from the images. This algorithm was used to restore the gray-scale as well as color images which were highly corrupted by salt and pepper noise. The proposed filter was used to replace the noisy pixel by a value of mean or a median of all other uncorrupted pixels in the selected window. The results show that the proposed algorithm exhibits better performance than the other existing algorithms in terms of MSE, PSNR and IEF. This filter is removed the salt and pepper effectively in high noise density levels.
Abhishek R and Srinivas N (2013) developed A new Weighted Median Filter (WMF) was best for removal of high density salt and pepper noise from the images. The existing non linear filters (Robust Estimation Algorithm (REA), Standard Median Filter (SMF), and Adaptive Median Filter (AMF)) were produced better results in low noise level only. So, overcome this disadvantage by Weighted Median Filter. From the experimental results, WMF algorithm is proposed to remove noise in gray and color images which exhibits better performance than the other existing algorithms in terms of MSE and PSNR.
V. Pranava Jyothy and K. Padmavathi (2013) introduced a Modified Decision Based Unsymmetrical Trimmed Median Filter (MDBUTM) algorithm for removal of salt and pepper noise from the gray scale, and color video’s. MDBUTMF is proposed to remove noise in gray and color videos which gives better than the existing median filter, adaptive median filter, and other existing noise removal algorithms in terms of PSNR and IEF. The proposed algorithm is effective for removal of salt and pepper noise in Videos at high noise densities.
This paper provides an extensive literature survey on the various median filters used for removing salt and pepper noise. The following are the key points of conclusion from the literature survey.
1. Even though the special filters produced the best performance they resulted in degradation of image resolution.
2. Gaussian filter does not preserve the fine details.
3. The arithmetic mean filter smoothens the local variation in images. This filter cannot remove all the noises from the images and also the image is too blurred.
4. The Weiner filter removes noise better than the Gaussian filter but the image is still blurry.
5. Median filter is used to preserve the edges and to remove the salt and pepper noise from the images. But it synthesised by Xilinx software it is computationally more complex.
6. The major drawback of median filter is their computational cost. Calculating Median value for every processing window requires 8 clock cycles.
7. The power consumption is more because of the clock cycles (frequency) P=CV^2F

...(download the rest of the essay above)

About this essay:

If you use part of this page in your own work, you need to provide a citation, as follows:

Essay Sauce, Image denoising. Available from:<https://www.essaysauce.com/information-technology-essays/essay-image-denoising/> [Accessed 02-10-23].

These Information technology essays have been submitted to us by students in order to help you with your studies.

* This essay may have been previously published on Essay.uk.com at an earlier date.