Scribd Logo
Upload

Welcome to Scribd!

  • 26 views

    Bannari Amman

    Uploaded by

    Jabeena Afthab
    This document summarizes a research paper that proposes a new adaptive soft switching median filter called the Noise Adaptive Soft Switching Median Filter (NASM) to effectively detect and remove salt and pepper noise while preserving image details. The NASM filter uses a soft switching noise detection scheme to classify each pixel as uncorrupted, isolated impulse noise, or non-isolated impulse noise/edge pixel. It then applies either no filtering, standard median filtering, or fuzzy weighted median filtering accordingly. Simulation results on the Lena image show the NASM filter superiorly removes high densities of impulse noise while better preserving fine details compared to other filters.

    Copyright:

    © All Rights Reserved
    Download as DOC, PDF, TXT or read online from Scribd

    Bannari Amman

    Uploaded by

    Jabeena Afthab
    26 views4 pages
    This document summarizes a research paper that proposes a new adaptive soft switching median filter called the Noise Adaptive Soft Switching Median Filter (NASM) to effectively detect and remove salt and pepper noise while preserving image details. The NASM filter uses a soft switching noise detection scheme to classify each pixel as uncorrupted, isolated impulse noise, or non-isolated impulse noise/edge pixel. It then applies either no filtering, standard median filtering, or fuzzy weighted median filtering accordingly. Simulation results on the Lena image show the NASM filter superiorly removes high densities of impulse noise while better preserving fine details compared to other filters.

    Copyright

    © © All Rights Reserved

    Available Formats

    DOC, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document summarizes a research paper that proposes a new adaptive soft switching median filter called the Noise Adaptive Soft Switching Median Filter (NASM) to effectively detect and remove salt and pepper noise while preserving image details. The NASM filter uses a soft switching noise detection scheme to classify each pixel as uncorrupted, isolated impulse noise, or non-isolated impulse noise/edge pixel. It then applies either no filtering, standard median filtering, or fuzzy weighted median filtering accordingly. Simulation results on the Lena image show the NASM filter superiorly removes high densities of impulse noise while better preserving fine details compared to other filters.

    Copyright:

    © All Rights Reserved
    Download as DOC, PDF, TXT or read online from Scribd
    Download as doc, pdf, or txt
    26 views4 pages

    Bannari Amman

    Uploaded by

    Jabeena Afthab
    This document summarizes a research paper that proposes a new adaptive soft switching median filter called the Noise Adaptive Soft Switching Median Filter (NASM) to effectively detect and remove salt and pepper noise while preserving image details. The NASM filter uses a soft switching noise detection scheme to classify each pixel as uncorrupted, isolated impulse noise, or non-isolated impulse noise/edge pixel. It then applies either no filtering, standard median filtering, or fuzzy weighted median filtering accordingly. Simulation results on the Lena image show the NASM filter superiorly removes high densities of impulse noise while better preserving fine details compared to other filters.

    Copyright:

    © All Rights Reserved
    Download as DOC, PDF, TXT or read online from Scribd
    Download as doc, pdf, or txt
    of 4

    Proceedings of National Conference on Signals, Systems & Security (NCSSS-2006), Feb 24 – 25, 2006

    Bannari Amman Institute of Technology,Sathyamangalam


    EFFECTIVE DETECTION AND REMOVAL OF SALT AND PEPPER
    NOISE BY USING NON-LINEAR FILTER( ADAPTIVE SOFT
    SWITCHING MEDIAN FILTER)
    (A.Jabeena.M.E, N.Sardar Basha.M.E, K Murali Babu.M.E, G.Suresh.M.E.)
    Abstract
    In this paper, a new algorithm for noise adaptive soft switching median filter is proposed. Median filters are
    well known for removal of Impulse noise. Switching based median filters are commonly found to be non adaptive
    to noise density variations and suffer in terms of misclassification of pixel characteristics at noise density
    interference. This points to the critical need for a sophisticated switching scheme and an adaptive weighted
    median filter. This project proposes a new switching based median filter is called Noise Adaptive Soft Switching
    Median Filter. This achieves a much-improved filtering performance in terms of effectiveness in removing
    impulse while preserving signal details and noise density variation. A soft switching noise detection scheme is
    developed to classify each pixel as uncorrupted pixel, isolated impulse noise, Non-isolated impulse noise or edge
    pixel. Signals unfiltered or standard median filter or Fuzzy weighted median filter will then be employed
    according to the respective characteristics type identified.
    Key words
    Median filter, Weighted median filter, switching based median filter, impulse detector, fuzzy weighted
    median
    1. Introduction
    Digital image processing finds its application in many important fields such us satellite communication,
    businesses application etc... Processing an image is a complex task. Any signal processing system, even though it is
    digital, is not perfect due to the constraints of noise. Therefore, main problem in image Processing is to reduce the
    effect of noise and present it with perceptible details.
    In the transmissions of images over channels, images are often corrupted by impulse noise due to faulty
    communications or noisy channels. Such noises may be introduced due to faulty camera or the like. In a black and
    white digitized image, pixels corrupted by positive impulses appear as white dots, and those corrupted by negative
    impulses appear as black dots. The objective is to remove impulses so that the noise –free image is fully recovered
    with a minimum signal distortion.
    To reduce the effect of noise, filtering is needed. Linear filters have been the primary tool for signal
    processing. They are easy to design and offer excellent performance. This is particularly true for spectral separation
    where the desired signal spectrum is significantly different from the interference. The drawbacks of linear filters are
    poor performance in the presence of nonadditive noise, non-Gaussian noise, channel non-linearity, and blur the
    edges. The transmission of the digital image requires communication channels. Communication channels are often
    model by theory of nonlinearity. So nonlinear filters must filter noise in image due to communication channels.
    Standard median[SM] filter [1] [2] was initially introduced to eliminate impulse noise with
    reasonably good performance. Standard median filter exploits the rank order information [3] of the input data to
    effectively remove impulsive noise by substuting the consider pixel with the middle position element of the
    reordered input data. Since then, it has been intensively studied and extended to such as weighted median[WM]
    filter center weighted median filters [4] and switching based median filters. The previously developed algorithms
    make use of fixed noise detection threshold obtained at a pre assumed noise density level. These algorithms lack
    adaptively to noise density variation. The mismatch between the designed algorithms and the actual noise density
    will cause noticeable and even substantial degradation on filtering performance. When the noise density increases,
    more misclassifications of pixel characteristic are going to occur and subsequently result in more degraded filtering
    42

    Page 75
    Effective detection and removal of salt and pepper noise by using non-linear filter
    performance. Therefore, an intelligent noise-detection process will be highly desirable and instrumental in correctly
    detecting various types of pixel characteristic. An adaptive filtering scheme is essential to remove the corrupted
    pixelwhile preserving image details when misclassification of pixel characteristic happens. These indicate that both
    noisedetection and corresponding filtering operation are crucial to achieve good median filtering performance,
    especiallyat high noise density interference.
    The proposed new switching based median filter incorporating fuzzy concepts called noise adaptive soft
    switching median filter [NASM] [5]. This achieves much improved filtering performance in terms of effectiveness
    inremoving impulse noise while preserving signal details and noise density variation. A soft switching noise
    detectionscheme is developed to classify each pixel as uncorrupted pixels, isolated impulsive noise, nonisolated
    impulsive noise. The filtering scheme is applied to according to pixel characteristics. Unfiltering effect is applied to
    uncorrupted pixel. Standard median filter and fuzzy weighted median filter would be subsequently used to remove
    impulse noise or preserve image object’s edge pixel. This filter superior performing in removing a high density of
    impulsive noise while preserving fine details.
    2. To find pixel characteristics and filtering action
    Uncorrupted pixels are identified by using global statistics based on the pixel intensities of the entire image.
    Impulse noise corrupts the image pixel by altering its intensities to either relatively high or relatively low value. The
    uncorrupted pixel has much smaller differences as compared to corrupted pixels to identify the uncorrupted pixel, to
    obtain the estimated original image. An estimation of the original image could be obtained by passing the noise
    image through standard median filter with an adaptive window size. Improve the detection performance estimated
    the
    original image is decomposed into nonoverlaping homogeneous rectangular block based on quadtree technique [6].
    Then calculate the pixel wise difference between noisy image and the original image is computed. A typical
    histogram of the particular block computed. From the histogram find the corrupted pixel and uncorrupted pixel.
    Distribution around the center is mainly contributed by those uncorrupted pixels.Set this pixels are uncorrupted pixel
    according to [5].
    The isolated impulsive noise possesses intensity which is relatively higher or lower then that of the its neighboring
    samples.Using fuzzy set concept the membership value is calculated for the pixel with respect to the uncorrupted
    pixel. Here the membership value divided into two groups, high-level group represents closely correlated pixel and
    lower value group indicating noncorralated pixel. The uncorrupted pixels within the sliding window are considered
    on computing their membership value associated to the center pixel. Set the threshold value for this condition and
    find the isolated impulse noise.set these pixels are isolated impulsive noise.
    Nonisolated impulse noise or edge pixels are most difficult to be discerned from each other since both are high
    frequency signals in essence. . If we consider pixel is isolated impulsive it may be edge pixel. it is discriminated by
    consider more sourrending pixels. If the pixel being considered is an edge pixel, such extension will be including.
    The consider pixel is a nonisolated impulsive noise; the inclusion of more impulse noise will not be possible since
    only uncorrupted pixels are considered in the window size. For this purpose the early computed membership value is
    used. Set the threshold value for this condition and find the nonisolated impulse noise and edge pixel. Set these
    pixels
    are isolated impulsive noise and edge pixel.
    Noise adaptive soft switching median filter detecting each pixel as uncorrupted pixels, isolated impulsive
    noise, nonisolated impulsive noise and edge pixel. After identifying pixel characteristics according to that
    Unfiltering
    effect is applied to uncorrupted pixel. Standard median filter and fuzzy weighted median filter would be
    subsequently
    used to remove impulse noise and preserve image object’s edge pixel. This filter superior performing in removing a
    high density of impulsive noise while preserving fine details.
    39

    Page 76
    Effective detection and removal of salt and pepper noise by using non-linear filter
    3.Algorithm
    Input:Number0frows X Number0fcolumns image
    Moving window W,|W|=N=2k+1
    Weight vector a=(a1,a2…..,aN)
    Output:Number0frows X Number0fcolumns image
    Let Halfsum=ΣN
    i=1 ai /2
    For I=1 to Numberofcolumns
    Place the window W at (i,j)
    Store the image values inside W and the corresponding weight in X=((X1,a1),(X2,a2)……(XN,aN))
    Let sum=0,m=1
    Repeat
    Let sum=sum+am
    Let m=m+1
    Until sum>halfsum
    Let output(I,j)=X(m-1)
    End
    end
    4. SIMULATION RESULTS
    The Lena image is used to simulate this algorithm. The experimental results shows this filter gives superior
    performing in removing a high density of impulsive noise while preserving fine details Fig1 (a) shows a corrupted
    Lena image with an impulsive noise of 25 %. Fig1 (b) shows the image filtered by SM Filtered with fixed window
    size. Results show that it has still impulses to be removed. The image filtered by the WM Filter is shown in Fig 1
    (c).
    Here impulse noise is removed compared to SM filter, but not completely removed. The estimated image obtain
    using adaptive SM Filter is shown in Fig1 (d). Here impulses were removed, but it slightly blured the edges. The
    output of the NASM and New NASM filters are shown in fig 1(e) and (f) respectively. The NASM filter removes
    the
    more impulse noise, compare to SM filter, WM filter. The impulsive noise has been removed more effectively in a
    new NASM filter when compared with other filters.
    The output of the NASM and New NASM filters are shown in fig 1(e) and (f) respectively. The NASM filter
    removes the more impulse noise, compare to SM filter, WM filter. The impulsive noise has been removed more
    effectively in a new NASM filter when compared with other filters.
    40
    Page 77
    Effective detection and removal of salt and pepper noise by using non-linear filter
    (a) (b)
    ( c) ( d)
    (e) (f)
    Fig1 (a) Corrupted image with impulsive noise 25% (b) SM Filter output (c) WM Filter output (d) Estimated image
    of NASM (e) NASM Filter (f) New NASM Filter
    41
    Page 78
    Effective detection and removal of salt and pepper noise by using non-linear filter
    4. CONCLUSION
    The proposed noise adaptive soft-switching median filter simultaneously addressed the following issues commonly
    encounter
    in certain switching based median filters are adaptiveness and sensitivity of decision-making threshold, accuracy of
    the noisedetection
    process, especially at high noise density, and suitability of the median filtering scheme exploited.
    The performance of the new noise adaptive soft switching has been extensively compared with that of standard
    median
    filter. Simulation results make known that the proposed noise adaptive soft switching filters significantly
    outperforms other
    techniques by having higher mean square with consistent and stable performance across a wide range of noise
    densities. To
    compare to other filters it gives good result. The same algorithm is also applied for colour images.
    Bibliography
    1. I.Pitas and A.Venetsanopou Nonlinear Digital Filter: Principal and Application. Norwell, MA: Kluwer.1990
    2.J.Astola and P. Kuosmanen Fundamentals of Nonlinear Digital Filtering, Boca Raton, fl: crc.1997.
    3. H.A David Order statistics. New York: wiley.1980
    4.S.J Ko J. and S.J Lee ‘Center weighted median filters and their applications to image enchancement’, IEEE
    Trans.Circuit
    System, Vol, 15,pp.984-993.1991
    5.H.L Eng and K.K Ma‘Noise Adaptive Soft Switching Median Filter’, IEEE Trans.Image Processing, Vol, 10,
    pp.245-
    251.2001
    6.J.S Lim Two-Dimensional Signal and Image Processing. Englewood Cliffs, NJ, PrenticeHall.1999
    7.T. Sun and Y.Neuvo ‘Detail preserving median based filters in image Processing’, Pattern Recognit Lett.1994
    8.R.Yang M. Gabbouj and Y. Neuvo ‘An efficient design method for optimal weighted medianfiltering’, IEEE
    circuit
    system.1994
    9.Raymond H.Chan, Chung-Wa Ho and Mila Nikolova ‘Salt and Pepper Noise Removal by Median-Type Noise
    Detectors and
    Detail-Preseving Regularization’, IEEE Trans.image processing,vol.14,2005.
    42

    Page 79

    You might also like