ABSTRACTS OF ARTICLES OF THE JOURNAL "INFORMATION TECHNOLOGIES".
No. 7. Vol. 25. 2019

DOI: 10.17587/it.25.415-425

N. N. Nagornov, PhD student, North-Caucasus Federal University, e-mail: sparta1392@mail.ru, P. A. Lyakhov, Candidate of Physics and Mathematics Sciences, Assistant Professor, e-mail: ljahov@mail.ru, N. I. Chervyakov, Doctor of Technical Sciences, Professor, Head of Departament of Applied Mathematics and Mathematical Modeling, e-mail: k-fmf-primath@stavsu.ru

The Noise Research in Discrete Wavelet Transform Filters for 3D Images Processing in Medicine

Many different methods are used to denoise and compress various medical images in practice. Discrete wavelet transform (DWT) underlies many of them. The quantization noise of DWT filters coefficients for 3D medical image processing analyzed in this paper. We proposed a new method for coefficients quantizing using rounding up and rounding down operations which reduce the error and require less resources for software and hardware implementation. The method for estimating maximum error of 3D images DWT with various bits per color (BPC) was developed. We defined the dependence of the minimum peak signal-to-noise ratio (PSNR) of grayscale and color images transform result on wavelet used, the effective bit-width of filters coefficients and BPC is. Approximate formulas for determining the minimum bit-width that provide a high (PSNR ?40 dB) and maximum (PSNR = ? dB) quality of 3D medical image processing using DWT filter bank was derived. Software simulation of 3D tomographic 8-bit and 12-bit grayscale images transform confirmed obtained theoretical results. Concluded that the quality of images DWT primarily depends on their BPC, on the required processing quality, on the number of wavelet filters coefficients and to a lesser extent on the type of wavelet. We only used fixed-point numbers in the proposed method for 3D medical imaging making possible its efficient software and hardware implementation on modern devices such as field-programmable gate arrays and application-specific integrated circuits.
Keywords: quantization noise, wavelet transform, medical imaging

P. 415–425

To the contents