Сергей Клавдиевич Абрамов, Виктория Валерьевна Абрамова, Сергей Станиславович Кривенко, Владимир Васильевич Лукин


The article deals with the analysis of the efficiency and expedience of applying filtering based on the discrete cosine transform (DCT) for one-dimensional signals distorted by white Gaussian noise with a known or a priori estimated variance. It is shown that efficiency varies in wide limits depending upon the input ratio of signal-to-noise and degree of processed signal complexity. It is offered a method for predicting filtering efficiency according to the traditional quantitative criteria as the ratio of mean square error to the variance of additive noise and improvement of the signal-to-noise ratio. Forecasting is performed based on dependences obtained by regression analysis. These dependencies can be described by simple functions of several types parameters of which are determined as the result of least mean square fitting. It is shown that for sufficiently accurate prediction, only one statistical parameter calculated in the DCT domain can be preliminarily evaluated (before filtering), and this parameter can be calculated in a relatively small number of non-overlapping or partially overlapping blocks of standard size (for example, 32 samples). It is analyzed the variations of efficiency criteria variations for a set of realizations; it is studied factors that influence prediction accuracy. It is demonstrated that it is possible to carry out the forecasting of filtering efficiency for several possible values of the DCT-filter parameter used for threshold setting and, then, to recommend the best value for practical use. An example of using such an adaptation procedure for the filter parameter setting for processing the ECG signal that has not been used in the determination of regression dependences is given. As a result of adaptation, the efficiency of filtering can be essentially increased – benefit can reach 0.5-1 dB. An advantage of the proposed procedures of adaptation and prediction is their universality – they can be applied for different types of signals and different ratios of signal-to-noise.


denoising efficiency; forecasting; DCT filter; additive noise; optimization


Gold, B., Ellis, D., Morgan, N., Bourlard, H., Fosler-Lussier, E. Speech and Audio Signal Processing: Processing and Perception of Speech and Music, 2-nd edition. USA, Wiley-Interscience. 688 p.

Viunytskyi, O., Shulgin, V. Signal processing techniques for fetal electrocardiogram extraction and analysis. Proc. of 2017 IEEE 37th International Conference on Electronics and Nanotechnology (ELNANO), Kiev, Ukraine, 2017, pp. 325-328.

Epifanov, S. V., Kononyhin, E. A. Sintez i analiz perspektivnoj SAU GTD [Synthesis and analysis of promising self-propelled guns GTE]. Aviacijno-kosmicna tehnika i tehnologia - Aerospace technic and technology, no. 10 (107), 2013, pp. 83-87.

Haykin, S., Widrow, B. Least-Mean-Square Adaptive Filters. Hoboken, NJ: Wiley-Interscience, 2003. 497 p.

Astola, J., Kuosmanen, P. Fundamentals of nonlinear digital filtering. Boca Raton (USA), CRC Press LLC, 1997. 288 p.

Gotchev, A., Nikolaev, N., Egiazarian, K. Improving the transform domain ECG denoising performance by applying interbeat and intra-beat decorrelating transforms. Proc. of ISCAS 2001, Sydney, NSW, vol. 2, 2001, pp. 17-20.

Donoho, D. L., Johnstone, I. M. Adapting to unknown smoothness by wavelet shrinkage. J. of American Statistical Association, vol. 90, no. 11, 1995, pp. 1200-1224.

Abosekeen, A., Noureldin, A., Korenberg, M.J., Improving the RISS/GNSS Land-Vehicles Integrated Navigation System Using Magnetic Azimuth Updates. IEEE Transactions on Intelligent Transportation Systems, 2019, pp. 1-14.

Lukin, V. V., Zelensky, A. A., Tulyakova, N. O., Melnik, V. P., Peltonen, S., Kuosmanen, P. Locally Adaptive Processing of 1-D Signals Using Z-parameter and Filter Banks. Proc. of NORSIG2000, Kolmarden, Sweden, 2000, pp. 195-198.

Lukin, V. V., Fevralev, D. V., Abramov, S. K., Peltonen, S., Astola, J. Adaptive DCT-based 1-D filtering of Poisson and mixed Poisson and impulsive noise. CDROM Proceedings of LNLA, Switzerland, 2008. 8 p.

Astola, J., Katkovnik, V., Egiazarian, K. Local Approximation Techniques in Signal and Image Processing. SPIE Press Monograph, vol. PM157, 2006. 576 p.

Chatterjee, P., Milanfar, P. Is Denoising Dead? IEEE Transactions on Image Processing, vol. 19, no. 4, 2010, pp. 895-911.

Rubel, O., Lukin, V., Abramov, S., Vozel, B., Pogrebnyak, O., Egiazarian, K. Is Texture Denoising Efficiency Predictable? International Journal of Pattern Recognition and Artificial Intelligence, vol. 32, no. 1, 2018. 32 p.

Chatterjee, P., Milanfar, P. Practical Bounds on Image Denoising: From estimation to information. IEEE Transactions on Image Processing, vol. 20, no. 5, 2011, pp. 1221-1233.

Abramov, S., Krivenko, S., Roenko, A., Lukin, V., Djurovic, I., Chobanu, M. Prediction of filtering efficiency for DCT-based image denoising. 2nd Mediterranean Conference on Embedded Computing (MECO), Budva, Montenegro, 2013, pp. 97-100.

Rubel, O., Lukin, V. An Improved Prediction of DCT-Based Filters Using Regression Analysis. Information and Telecommunications Sciences, Kiev, Ukraine, vol. 5, no. 1, 2014, pp. 30-41.

Wang, Z., Bovik, A., Sheikh, H., Simoncelli, E. Image quality assessment: From error visibility to structural similarity. IEEE Transactions on Image Processing, vol. 13, no. 4, 2004, pp. 600–612.

Abramov, S., Abramova, V., Lukin, V., Egiazarian, K. Prediction of Signal Denoising Efficiency for DCT-based Filter. Telecommunication and radio Engineering, 2019. 14 p.

Oktem, R., Yarovslavsky, L., Egiazarian, K. Signal and image denoising in transform domain and wavelet shrinkage: A comparative study. In Proceedings of 9th European Signal Processing Conference, 1998, pp. 2269-2272.

Cameron, C., Windmeijer, A., Frank, A. G., Gramajo, H., Cane, D. E., Khosla, C. An R-squared measure of goodness of fit for some common nonlinear regression models. J. of Econometrics, no. 77(2), 1997. 16 p.



  • There are currently no refbacks.