The novel noise classification techniques found on quadruple threshold statistical detection filter under fix intensity impulse outlier environment

Vorapoj Patanavijit, Kornkamol Thakulsukanant

Abstract


Because of the enormous necessity of contemporary noise suppressing algorithms, this article proposes the novel noise classification technique found on QTSD filter improved from the TTSD filter. The four thresholds for each auxiliary situations are incorporated into the proposed QTSD framework for dealing with the limitation of the earlier noise classification technique. The mathematical pattern is modeled by each photograph elements and is investigated in contradiction to the 1st threshold for analyzing whether it is non-noise or noise photograph elements. Subsequently, the calculated photograph element is analyzed with the contradiction between the 2nd threshold, which is modeled by using the normal distribution (mean and variance), and is analyzed with the contradiction between the 3rd threshold, which is modeled by using the quartile distribution (median). Finally, the calculated photograph element is investigated in contradiction to the 4th threshold, which is modeled from maximum or minimum value for analyzing whether it is non-noise or noise photograph elements FIIN. For performance evaluation, extensive noisy photographs are made up of nine photographs under FIIN environment distribution, which are synthesized for investigating the proposed noise classification techniques found on QTSD filter in the objective indicators (noise classification, non-noise classification and overall classification correctness). From these results, the proposed noise classification technique can outstandingly produce the higher correctness than the earlier noise classification techniques.

Keywords


Digital image denoising; Digital image processing; Quadruple threshold statistical detection; Standard median filter; Triple threshold statistical detection

Full Text:

PDF


DOI: https://doi.org/10.11591/eei.v10i5.3105

Refbacks

  • There are currently no refbacks.


Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Bulletin of EEI Stats