|
“Digital
Signal Processing” No. 1-2014 |
|
Generalization
of the Goertzel algorithm and the sliding parametric discrete Fourier
transform algorithm
Ponomarev V.A. Doctor of Technical Sciences, professor of the Kalashnikov
Izhevsk State Technical University, ponva@mail.ru
Ponomareva O.V. Ph.D. (technical science), associate professor of
Kalashnikov Izhevsk State Technical University
Ponomarev A.V. Ph.D. (economy), chief of staff of Central Election Commission
of Udmurt Republic
Ponomareva N.V. Director of software testing "NPO Computer" Ltd., ponva@mail.ru
Keywords:
parametric discrete Fourier transform, slide parametric discrete Fourier
transform, detection, harmonic component, Goertzel algorithm, comb filter.
Annotation
In this paper we introduce a new procedure for digital block processing,
which is called by the authors as procedure of digital block processing
with accumulation. On the basis of the procedure two generalizations of
Goertzel algorithm were proposed.
First generalized Goertzel algorithm, unlike standard Goertzel algorithm
allows simultaneous provide:
- high resolution digital spectral processing;
- stability of the filter.
Second generalized Goertzel algorithm, unlike standard Goertzel algorithm
allows simultaneous provide:
- high resolution digital spectral processing;
- stability of the filter;
- full control over the resonant frequency of Goertzel filter.
The paper considers the generalized algorithms of one-bin sliding DFT
(SDFT) and parametric one-bin sliding DFT (SDFT-P). The proposed algorithms
significantly reduce, in comparison with traditional SDFT and SDFT-P algorithms,
the number of operations required to force SDFT and SDFT-P algorithms
work in a sliding measurement mode.
Advantage, which is provided by the proposed generalized algorithms, is
approximately to 4 times reduce the number of real multiplications and
2 times reduce the number of real additions.
In addition, by applying the proposed generalized Goertzel algorithms
in SDFT and SDFT-P structures (first and second, respectively), possible
to achieve a high-resolution sliding spectral analysis while maintaining
the stability of the filters.
References
1. M. G. Serebrennikov and A. A. Pervozvanskii, Detection of Hidden
Periodicities (Nauka, Moscow, 1965) [in Russian].
2. E. Oppenheim, Digital Signal Processing (Mir, Moscow, 1980).
[Russian translation].
3. Nondestructive Control: a Reference Handbook, Ed. by V. V. Klyuyev
(Mashinostroenie, Moscow, 2007), Vol. 7.
4. G. Lyons, Understanding Digital Signal Processing (Binom-Press,
Moscow, 2007) [Russian translation].
5. V. A. Ponomarev and Î. V. Ponomareva, "Theory and Application of
the Parametric Discrete Fourier Transform," Tsifrovaya Obrabotka Signalov,
No. 1, 2-6 (2011).
6. Î. V. Ponomareva, "Fast Parametric Discrete Fourier Transform of
Real Sequences," Tsifrovaya Obrabotka Signalov, No. 2, 2-5 (2012).
7. V. A. Ponomarev and Î. V. Ponomareva, "Modification of the Discrete
Fourier Transform for Solving Problems of Interpolation and Convolution
of Functions," Radiotekhnika i Elektronika, 29 (8), 1561-1570 (1984).
8. V. A. Ponomarev, Î. V. Ponomareva, "A Generalization of the Discrete
Fourier Transform for Interpolation in the Time Domain," Izv. Vuzov,
Radioelektronika XXVI (9), 67-68 (1983).
9. Î. V. Ponomareva, A. V. Ponomarev, and N. V. Ponomareva, "Sliding
Parametric DFT in Detecting Tonal Components," Tsifrovaya Obrabotka
Signalov, No. 4, 2-7 (2012).
10. Î. V. Ponomareva, "Development of the Theory of Spectral Analysis
of Discrete Signals on Finite Intervals in the Basis of Parametric Exponential
Functions," Tsifrovaya Obrabotka Signalov, No. 2, 7-12 (2010).
11. Î. V. Ponomareva, "Probabilistic Properties of Spectral Estimates
Obtained by Parametric Discrete Fourier Transform," Intellektual'nye
Sistemy v Proizvodstve, No. 2(10), 36-42 (2010).
12. V. A. Ponomarev and Î. V. Ponomareva, "Time Windows in Evaluating
Energy Spectra by Parametric Discrete Fourier Transform," Avtometriya,
No. 4, 39-45 (1983).
13. Î. V. Ponomareva, A. V. Ponomarev, and N. V. Ponomareva, "A Fast
Method for. Computing the Discrete Fourier Transform of Real Sequences,"
Tsifrovaya Obrabotka Signalov, No. 2,10-15 (2013).
14. V. A. Alekseev, V. A. Ponomarev, and Î. V. Ponomareva, "Methodology
for Determining Measurement Errors of Probability Characteristics of Random
Processes Implemented by Measuring Processing Means" Intellektul,nye
Sistemy v Proizvodstve, No.2(10), 91-99 (2010).
15. Î. V. Ponomareva, V. A. Alekseev, and V. A. Ponomarev, "Digital
Periodogram Analysis and Problems of its Practical Application," Vestn.
IzhGTU, No 2, 130-133 (2013)
16. Î. V. Ponomareva and N. V. Ponomareva, "Modification of a Filter
Based on Frequency Sampling for Solving Problems of Digital Processing
of Stochastic Processes with Hidden Periodicities," Intellektual'nye
Sistemy v Proizvodstve, No. 2(20), 122-129 (2012).
17. Î. V. Ponomareva, A. V. Ponomarev, and V. A. Ponomarev, "Generalization
of the Goertzel Algorithm for Detecting Hidden Periodicities," Intellektual'nye
Sistemy v Proizvodstve, No. 1(21), 41-46 (2013).
18. V. A. Ponomarev and Î. V. Ponomareva, "Vibroacoustic Diagnostics
of the Machine Gearboxes by Digital Methods," Stanki i Instrument,
No. 9, 18-21 (1983).
Cascade connection optimal linear phase FIR filters
N. O. Vzduleva, Post-graduate, Kalashnikov Izhevsk State Technical University,
sizovan@list.ru
V. B. Gitlin, DSc in Engineering, Professor, Kalashnikov Izhevsk
State Technical University, vbg_istu@mail.ru
Keywords: chromatograph, FIR filter, cascade connection, optimal, signal/noise ratio, software package MATLAB.
Abstract
Chromatographic analysis methods [1] are sufficiently accurate methods
of determining the presence of a substance component in the sample, and
calculating its concentration [1]. Chromatographic assay sensitivity is
determined by the relation of energy useful chromatograph signal to background
noise [1]. To increase the signal / noise ratio filtering signal having
a corresponding frequency response can be used. FIR filters with linear
phase [2] are preferable as chromatographic analysis is based on accurate
timing measurements of output chromatographic signal. Designing filter
is preferably carried using Chebyshev approximation ensures the minimum
order filter [2]. Experiments with the chromatographic signals has been
shows that the order of the designed FIR filter should have rather high
[3]. Trying to estimate the parameters of the filter Remez algorithm ,
by finding optimal solutions of equations by Gauss [4 ] has been show
that the solution diverges already at odds filter order N = 51. Using
an algorithm Pakrsa - McClellan [5] is also not allowed to increase the
order of the calculated filter. Limited accuracy of calculations , even
at higher bit number representation requires special methods for calculating
FIR filters using of a Chebyshev approximation [4]. To create the optimal
FIR filter with linear phase of high order , it was decided to construct
a FIR filter by cascading number of optimal FIR filters with the same
lower order. Such filters can be attributed to the quasi-optimal . Ones
preserve phase linearity, require significantly less computation for calculating
the parameters of the filter. Ones are not inferior about quality signal
filtration by the optimal FIR filters of the same order, calculated using
the software package MATLAB. Application of FIR filter with linear phase,
consisting of the cascade of two optimal FIR filters 51-th order , allowed
to increase the signal / noise ratio at the output of the Chromatographic
is almost eight times.
References
1. Sacodinsky R.I., Brajhnikov B.B. Volkov S.F. Analytical chromatography.
Moscow: Chemmistry, 1993.
2. L.R. Rabiner, B. Gold. Theory and Application of Digital Signal Processing. Prentice-Hall, Inc., Englewood Cliffs, New Jersey, USA., 1975/
3. N.O. Vzduleva, V.B. Gitlin. Selection of noise chromatograph filter parameters / Information Technology of Science, Industry and Education. Publishing House ISTU, Izhevsk, 2014. - P. 40-42.
4. F.I. Solonina, D.A. Ulachovich, S.M. Arbuzov, E.B. Solovieva. Digital Signal Processing. Second edition. Publishing House SPB.: Bckh.V. - Petersburg, 2005. - 768 p.
5. A. Oppeheim, P. Shafer. Discrete - Time Signal Processing. Prentice-Hall Pearson Education Inc. 1999/ - 856 p.
Optimization of the multirate adaptive filter with least-mean-square
algorithm in equal-width subbands
Linovich A.Yu., tor@rsreu.ru
Keywords: multirate
adaptive filter, least-mean-square algorithm, optimization, computer simulation,
equal-width subbands.
Annotation
The main subject is solving of the optimization problem for the multirate
adaptive filter with least-mean-square algorithm in equal-width subbands.
Analytical estimations for the best values of the main parameters of the
multirate adaptive filter are derived. The results of the computer simulation
which confirm the theoretical study are supplied. The analysis of the
theory and experiments results in the next decisions. First, while the
optimal ratio between the decimation ratio and quantity of channels holds
true, large changes of the quantity of channels tends to insignificant
increase in the computational complexity. So this ratio may be considered
as the main condition of optimality. Second, for the little quantity of
channels the discreteness of the quantity of channels and decimation ratio
should be taken into account. That is, in the neighborhood of the theoretical
minimum point it is necessary to choose those couples of numbers which
are closer to the optimal ratio.
References
1. Vityazev V.V. Digital frequency-domain selection of the signals.
– Moscow: Radio i svyaz, 1993. – 240 pp.
2. Davidson T.N. Enriching the art of FIR filter design via convex optimization // IEEE signal processing magazine. 2010. – No.3. – pp. 89 – 101.
3. Wilbur M.R., Davidson T.N., Reilly J.P. Efficient design of oversampled NPR GDFT filter banks // IEEE transactions on signal processing. 2004. – No.7. – pp. 1947 – 1963.
4. Linovich A.Yu. Adaptive filters with digital frequency selection without additional delay and spectrum aliasing // Fundamental problems of radioelectronics / International Conference «INTERMATIC–2012», 2012 Dec. 3–7, Moscow – Part 5. – 190 pp. – P. 69 – 71.
5. Linovich A.Yu. Adaptive filters with digital frequency selection of the signals in wideband telecommunication systems // Fundamental problems of radioelectronics / International Conference «INTERMATIC–2012», 2012 Dec. 3–7, Moscow – Part 5. – 190 pp. P. 144 – 147.
6. Haykin S. Adaptive Filter Theory. – London: Pearson, 5th ed., 2013. – 912 pp.
7. Widrow B., etc. Complex form of the NLMS-algorithm // TIIER. 1975. – No.3. – P. 49 – 51.
8. Diniz P. Adaptive Filtering: Algorithms and Practical Implementation. – Lexington (KY, USA): Springer, 3rd ed., 2011. – 912 pp.
9. Widrow B., Stearns S. Adaptive digital processing / Moscow: Radio i svyaz, 1989. – 440 pp.
10. Linovich A.Yu. Application of the time-frequency decomposition methods in the inverse modeling problem // Ciphrovaya obrabotka signalov. 2005. – No.3. – P. 28 – 37.
11. Weiss S., Stewart R.W. On Adaptive Filtering in Oversampled Subbands. – Aachen (Germany): Shaker Verlag, 1998.
12. Rabiner L., Gould B. Theory and applications of the digital signal processing. – Moscow: Mir, 1978. – 848 pp.
13. Bellanger M.G. Traitement Numerique Du Signal. – Paris: Masson, 1980. – 375 pp.
14. Corn G., Corn T. Mathematics hand-book for scientists and engineers. – Moscow: Nauka. Central publishing of physical and mathematical literature, 1984. – 832 pp.
15. Piskunov N.S.
Differential and integral calculus for institutes of technology.
– Moscow: Physmathgiz, 1963. – 856 pp.
Peculiarities and characteristics of frequency regeneration with the
RF signals recording-reproducing digital functional devices in the multisignal
mode
Kozlov S.V., Doctor of Engineering, the deputy chief of chair of
radio engineering and antenna and feeding devices of Air Force Education
and Research Centre “The Zhukovsky and Gagarin Air Force Academy”
Mazilov S. L., Candidate of Technical Sciences, the senior teacher of
chair of radio engineering and antenna and feeding devices of Air Force
Education and Research Centre “The Zhukovsky and Gagarin Air Force Academy”
Uskov A.V., graduated in a military academy of chair of radio engineering
and antenna and feeding devices of Air Force Education and Research Centre
“The Zhukovsky and Gagarin Air Force Academy”, sc79@mail.ru
Keywords: digital device of record and reproduction of radio signals,
frequency restoration, range.
Annotation
In a number of practical appendices of signal's processing the problem
of restoration (reproduction) of frequency on the basis of supervision
of radio signals on a limited interval of time is solved. For the solution
of this task digital devices of record and reproduction of radio signals
(digital radiofrequency memory - DRFM) now are widely used. Standard are
a situation of functioning of DRFM in a multisignal (multifrequency) mode
in the presence of noise and harmonious hindrances.
Research of the main regularities and receiving quantitative estimates
of influence of a multifrequency mode and hindrances on quality of restoration
of bearing frequencies of radio signals with use of digital devices of
record and reproduction were carried out with use of methods of statistical
radio engineering and imitating modeling.
The case formation of harmonious fluctuation from the written-down short
digital sample by duration was considered 
,
where 
- the period of the digitization consisting from 
complex
counting 
with
estimates of phases of counting 
,
with method use " sewing together of phases". The essence of a method
consists in: allocation of a short site - an analysis interval from the
first 
,
duration
;
search among 
counting
of a copy of a radio signal at 
,
,
groups from 
the counting which estimates of phases have the minimum average square
of a deviation from estimates 
,
;
calculation of average shift of phases between counting during signal
reproduction 
;
formation of demanded number 
counting of restored harmonious fluctuation by cyclic rewriting
(reproduction) of the written-down site with the reproduction period
and shift on frequency on 
.
For a case of existence of the unique harmonious fluctuation and noise
for a mistake's square root 
restoration of frequency of a look 
is received. The expression connecting parameters of algorithm
of restoration of frequency (Tb, Ta),
spectral density of capacity N0 noise in a strip of frequencies
of DRFM and capacity Pñ harmonious signal.
The main regularities of a multifrequency operating mode are revealed
on an example of biharmonic fluctuation by methods of the spectral theory
and imitating modeling. It is shown that:
in a case when the period of reproduction is chosen equal to the period
of multifrequency fluctuation, exact restoration of frequency of each
harmonious component from structure of multifrequency fluctuation will
take place;
if the condition of equality of the period of reproduction and the period
of multifrequency fluctuation is not carried out, errors of restoration
of frequency of each harmonious component and enrichment of a range of
the restored fluctuation will take place;
at equality of amplitudes of harmonious components of biharmonic fluctuation,
an error of reproduction of their frequencies are equal on the module
and have a different sign; at increase in the relation of amplitudes of
harmonious components of biharmonic fluctuation the error of reproduction
of frequency of a bigger component on amplitude decreases approximately
in proportion to the relation of amplitudes, and the error of reproduction
of frequency of a smaller component on amplitude slightly increases;
with increase in time of record and increase in a rating of frequencies
of an error of reproduction of frequencies of harmonious components of
multifrequency fluctuation decrease.
It is shown that at restoration of frequencies with use of digital devices
of record and reproduction on short samples situations of a multialarm
mode are the most dangerous at small frequency off-tuning of signals.
The received results can be used at justification of characteristics and
research of efficiency of functioning of digital devices of record and
reproduction of radio signals.
References
1. Perunov Yu.M., Fomitchyov K.I., Yudin L.M. Radio-electronic
suppression of information channels of control systems by the weapon.
- M: Radio engineering, 2003.
2. Dobykin V.D., Kupriyanov A.I., Ponomarev V. G., Shustov L.N. Radio-electronic
fight. Digital storing and frequency reproduction / Under the editorship
of Kupriyanov A.I. - M: High school book, 2009.
3. Levin B. R. Theoretical bases of statistical radio engineering.
- M: Radio and communication, 1989.
Estimation method for interaction of local extremum for matrices
of coefficients in continuous wavelet transforms of EEG signals
Turovskij Ya.A., Kurgalin S.D., Semjonov A.G. , Voronezh state university
Keywords: electroencephalogram, continues wavelet analysis, local
spectrum, chain of the local maximum.
Annotation
Application of wavelet transformations for analysis of biomedical signals
has helped to expand the volume of useful information, obtained from processing
of data recorded from humans or animals during clinical or physiological
studies. However, the currently applied methods, based on continuous wavelet
transformations, in most cases do not use an approach based on a dynamic
estimation of the spectral characteristics of the time series of the analyzed
data. Consequently, it is important to establish such methods, which take
into account that biomedical signals, including electroencephalograms
(EEG), are the result of complex nonlinear interaction of a large number
of oscillators, which generate electrical signals from the studied organs
and systems. The resulting data are also influenced by features of organs
and tissues, through which the signal passes before it reaches the transducer.
Therefore, we need an approach aimed at all-round analysis of biomedical
signals, including EEG data, based on the continuous wavelet transformation,
which will help to identify the features of processes taking place in
the studied organs and systems. These features, as it was already noted,
are reflected in the form of specific dynamics of behaviour in the chains
of local maxima (CLM) or minimum (ClMin) matrices of squares of coefficients
of a continuous wavelet transformation. The given paper is aimed at developing
a research method for studying the dynamics of local extrema for matrices
of second powers of coefficients for continuousEEG signals wavelet transformations
to identify areas in which there is a high concentration of local maxima
and minima chains. An analysis of these areas will allow to evaluate the
performance of the processes occurring in organs and systems that generate
these signals.
In this paper we propose a method of analysis of biomedical signals on
the example of the EEG based on evaluation of structures formed in the
result of interaction between chains of local maxima - CLM and chains
of local minima - CLMin in space (a,b) "the scale of the wavelet transformation
- time". We have developed the algorithms of detection for such structures,
which are called "areas of convergence of extremes" - ACE. A number of
computative experiments held with different types of model signals (monofriquency
harmonics, the sum of several harmonic signals, amplitude modulated harmonic,
"white" and "colored" noises) showed that the formation of the ACE is
connected with the phenomenon of CLM and CLMin drift in frequency space.
We have demonstrated a number of approaches for handling the results of
a research on the example of the PEL, and determined an approach to the
classification of the identified EEG phenomena.
The obtained method can be further applied for studying a wide range of
signals of medico-biological character. This will significantly expand
the range of detected phenomena while analysing biomedical signals, and,
as a result, to raise the general informativeness of clinical-physiological
studies.
References
1. Turovskij Ja.A. Programma PikWave 1.0. Zaregistrirovana v Rossijskom
agentstve po patentam i tovarnym znakam, registracionnyj ¹ 2006613500.
2. Turovskij Ja.A., Semjonov A.G., Kiseleva E.V., Horoshih N.V. Programma
Wavemax 1.0. Zaregistrirovana v Rossijskom agentstve po patentam i tovarnym
znakam, registracionnyj ¹ 2012614720.
3. Turovskij Ja.A., Kurgalin S.D., Vahtin A.A. Obrabotka signala jelektrojenncefalogrammy na osnove analiza chastotnyh zavisimostej i vejvlet-preobrazovanija // Biomedicinskaja radiojelektronika. - 2012. - ¹ 12. - S.39-45.
4. Turovskij Ja.A., - 2012. - Vestnik Voronezhskogo gosudarstvennogo universiteta. Ser. Sistemnyj analiz i informacionnye tehnologii. - 2013. - ¹2. - S. 69-73.
5. Turovskij Ja.A., Kurgalin S.D., Maksimov A.V. Modelirovanie processa vydelenija chastotnyh lokal'nyh minimumov v signalah jelektrojencefalogramm // Vestnik Tambovskogo gosudarstvennogo tehnicheskogo universiteta. - 2012. - T. 18, ¹ 4. - S. 827-834.
6. Turovskij Ja.A., Kurgalin S.D., Semjonov A.G. Analiz jencefalogramm na osnove issledovanija cepochek lokal'nyh maksimumov skejlogramm // Cifrovaja obrabotka signalov. - 2013. - ¹2.- S.20-23.
7. Turovskij Ja.A., Kurgalin S.D., Semjonov A.G. Dinamika jenergeticheskih pokazatelej cepochek lokal'nyh maksimumov vejvlet-kojefficientov biomedicinskih signalov // Cifrovaja obrabotka signalov. - 2013. - ¹2. - S.24-29.
8. Astaf'eva N.M. Vejvlet-analiz: osnovy teorii i primery primenenija // Uspehi fizicheskih nauk. - 1996. - T. 166, ¹ 11. - S. 1145-1170.
9. Chubukova I. A.
Data Mining: uchebnoe posobie. - M.: Internet-universitet informacionnyh
tehnologij: BINOM: Laboratorija znanij, 2006. - 382 s.
Digital halftone image filtration based on markov chain with several
states
Petrov E.P., Kharina N.L., Rzhanikova E.D., Vyatka State University,
Kirov, eppetrov@mail.ru
Keywords: digital halftone image, Markov chain, non-linear image
filtration.
Abstract
The requirements for volume, quality and speed of transmitted data are
increasing at the moment. The most capacious data carrier is a color digital
image composed of a combination of monochromatic digital halftone images
described as a binary g-digit numbers set. The number of the brightness
gradations in monochromatic halftone images is equal 
.
The digital halftone image representation by binary numbers with more
than eight digits allows to improve the quality of the images transmitted
over communication channels. However, the higher the digit capacity of
a digital halftone image is, the more time is required for its transfer.
Therefore the problem of the time transfer reduction is actual and can
be solved by a transition from a multigraded to a low-graded image representation
of the transferring side of a communication channel and back to a multigraded
image of the receiving side. For this purpose a g-digit halftone image
is presented by a set of g digit binary images. Each of them is a superposition
of two one-dimensional Markov chains on an orthogonal image lattice. Each
of the two Markov chains is characterized by two equiprobable states and
one-step transition probability matrixes for both horizontal and vertical
directions. Note that digital binary images included in a digital halftone
image are correlated among themselves, and the higher the digital binary
images order is, the higher its correlation is. Constructing the neighboring
digital binary image clusters that should be uniform in size, we get an
image with a smaller number of brightness gradation. For example, clustering
the digital binary images forming the 8-digit halftone image pairwise,
we get low-graded images with four brightness gradations. Suppose that
the clusters of neighboring binary images, as well as digital binary image,
are two-dimensional Markov chains with several states. The clusters of
neighboring binary images can be transmitted through a communication channel
by multiposition phase-shift keyed signals. Using this kind of signals
reduces the noise stability of image transfer as compared with binary
signals. However, this reduction is compensated by means of non-linear
filtration algorithms synthesized in this work, efficiently realizing
the statistical redundance caused by the interdigit correlation of an
image.
The white Gaussian noise model was used, the filtration simulation of
a halftone image transmitted by means of binary and low-graded images
was carried out. For the evaluation filtration test, a mean square error
was calculated at the input and at the output of the nonlinear filter.
As a result of the filtration, the mean square error decreased four times
in both cases.
Hence, a transition from a g-digit halftone image to a low-graded image
allows to use the statistical redundance of an image as much as possible
and to reduce the image transfer time, i.e. "to compress" the digital
image by a number of times equal to the number of digital binary images
included into the low-graded image.
References
1. Petrov E.P., Trubin I.S., Chastikov I.A. Non-linear filtration of
video sequences of digital halftone images of Markov type. // Uspekhi
sovremennoi radioelektroniki (Achievements of Modern Radioelectronics)
- 2007, no.11, pp. 54-87.
2. Petrov E.P., Medvedeva E.V. Non-linear filtration of statistically related video sequences based on hidden Markov chains // Radiotekhnika i elektronika (Journal of Communications Technologies and Electronics), 2010, Ò.55, ¹ 3, pp. 330-339.
3. Medvedeva E.V.
Adaptive non-linear filtration of color video images // Informatsionnye
tekhnologii (Information Technologies), ¹11, 2009, pp. 61 - 64.
Image compression based on block decomposition in the wavelet-packets
domain
Prof. Dr. Sergei V. Umnyashkin, National Research University of
Electronic Technology, e-mail: vrinf@miee.ru
PhD student Ruslan R. Gizyatullin, National Research University of Electronic
Technology, e-mail: ruslan.gizyatullin@gmail.com
Keywords: adaptive basis selection, block decomposition, discrete
wavelet transform, wavelet packets, image compression.
Annotation
An algorithm of still image compression is proposed, which is based on
wavelet packet transform. The compression algorithm uses a limited number
of wavelet packet bases which are adjusted to the local spatial properties
of the image processed.
First, 4-level wavelet decomposition is applied to an image to be compressed.
Then the 6 subbands of the 2 highest frequency levels of the wavelet spectrum
are split into adjacent square blocks, three 16×16 blocks and three
32×32 blocks of wavelet coefficients from 6 different subbands. All
sets of those 6 blocks in the wavelet domain correspond to 64×64
pixel blocks mapping the original image.
The sets of 6 wavelet decomposition blocks are processed independently.
We selected some 27 different templates from all possible wavelet-packet
bases which our algorithm tests then as an alternative for each set of
6 wavelet blocks, the same filter banks are applied. Entropy of the bases
is used as a decision rule for the basis selection.
When the wavelet packet bases are selected for all block sets then scalar
quantization is applied to the decomposition coefficients. The quantized
coefficients and the keys defining the basis structure are encoded by
context arithmetic coder at the final step of the proposed image compression
algorithm.
Experiments prove that the basis adjustment increases the peak signal-to-noise
ratio (PSNR) up to 1 dB. The final performance of the proposed codec never
gives the rate-distortion characteristics less than JPEG2000 standard
shows, an average improvement of 0,3-0,5 dB is observed.
Methods of selection of personnel sync pulse input in uncompressed
video stream from a unidirectional single-bit digital lines and their
implementation on FPGA
Aminev D., ZAO "MNITI" research associate, aminev.d.a@ya.ru
Fokin A., ZAO "MNITI" engineer, fw@bk.ru
Keywords: selection, data transmission, implementation on FPGA,
video streams.
Abstract
Òhe article investigates the problem of personnel selection in the
transmission of video streams data inter-module connections is studied.
When modules are located from a few meters and more often used unidirectional
single-digit digital line that transmits an electrical signal to LVDS
or LVTTL standards that use personnel selection. Derived formulas for
calculating the ratios of receiving and transmitting frequencies and other
selector settings. A generalized structure of the video stream and the
selector circuit, built on the FPGA. The methods of selection of personnel
sync options and their implementation on FPGA. Discussed in detail the
four methods of personnel selection and describe their advantages and
disadvantages:
In method 1, the frequency corresponds to the frequency selection CLKd
Fin bitstream. When this bit shift register is M×2n.
In method 2, the frequency selection in CLKd 2n (n = 2) times larger than the flow rate and compared with a standard is carried out at each cycle. When this bit shift register is M×2n.
In method 3 CLKd selection frequency so as to 2n n (n = 2) times the frequency of the stream, but the comparison of each bit stream is carried on a reference corresponding to the second and third bits of the shift register word length M×2n, the first and fourth bits are ignored.
In method 4 the frequency of selection in the same CLKd 2n (n = 2) times larger than the flow rate, however, the comparison with the benchmark performed based transient personnel signal (bits 7, 8, 15, 16, 23, 24 and 31).
Selected method 4 eliminating flaws other methods, which tracks transients in sync and provides the highest probability of selection. For the proposed methods presented are timing charts personnel selection processes.
The article presents fragments selector description language Verilog, where the selection of the frame at a frequency lower frequency 2n flow with flexible reference to the standard deviation has the highest frequency. The principle of selection frame frequency flow is simple to implement, but despite the high has a greater probability of failure at mismatch, both the frequency and phase of the signal. Such personnel selector copes with problems of the frame and line selections in the transmission of uncompressed video format 512×512.
References
1. Aminev D.A, Uvaisov S.U, analysis protocols for high-speed data
transmission in the inter-module network connections / / In.: Information
and communication technologies in education, science and industry. Proceedings
of the VI International Scientific and Practical Conference. In two parts.
Charles II. Protvino, 2012. pp. 198-201.
2. Altera Serial Digital Interface (SDI) MegaCore Function http://www.altera.com/literature/ug/ug_sdi.pdf
3. Specifications of the Camera Link Interface Standard for Digital Cameras and Frame Grabbers, Version 1.1 Automated Imaging Association, Jan 2004 http://www.fastvideo.ru/info/cameralink/CameraLinkOfficial.pdf
4. Aminev D.A, Soroka E.Z. Universal input device and unipolar differential digital signals / / RF Patent number 2440666, 20.01.2012g., Priority of invention from 05.07.2010
5. Aminev D.A. Uvaisov S.U, Kondrashov A.V. Analysis of technical achievements in addressing registration information flows. / / Proceedings of the International Conference "Problems of protection of intellectual property in a variety of industries, science, education and medicine in terms of Russia's accession to the WTO." Togliatti. 2012.
6. Aminev D.A. Kondrashov A.V. Analysis and classification methods for converting digital data streams for high-speed processing and recording systems / / Systems and communication and broadcasting. ¹ 1, 2. 2012. p. 37-41 .
7. Aminev D.A. Kondrashov A.V. Harmonisation Russian standards for the transmission of high-speed information flows / / Digital Signal Processing. 2013. ¹ 2. Pp. 64-65.
8. Aminev D.A. Implementing
embedding additional information when encoding MPEG-2 video stream using
FPGA / / Technique means. Series: Tech TV. 2011. ¹ 1. p. 98-103.
Calculation of
uncertainty function for passive radar by means of FPGA and GPU
Pankratov V.G., v_pankratov@mail.ru
Karih A.A., karih@bk.ru
Panfilov V.N.,
Gyrov A.D., adg-elint@rambler.ru
Keywords: ambiguity function, correlation function, Doppler filtering,
correlator, assess hardware costs, Fourier transformation.
Annotation
Algorithm for real-time calculating uncertainty function of the received
signal on a delay-frequency plane is discussed. The algorithm consists
of three main stages: calculation of correlation function and forming
correlation matrix, forming Doppler frequencies matrix, multiplication
of the obtained matrixes and obtaining uncertainty function matrix. Thus
the scalar product of the vector row of the correlation matrix on a vector
is a column matrix of twiddle factor gives the matrix element function
uncertainty. To reduce the number of operations the calculation of the
correlation function is made using the standard FFT kernel via the spectral
representation of the input signals. The scope of the algorithm applicability
is defined, limitation in Doppler frequencies and delays ranges are marked.
Comparative evaluation of computation expenses when calculating uncertainty function by different methods has show that the proposed algorithm requires 500 times less operations than the direct method. The choice of element base for algorithm realization is carried out. Best of all FPGA and GPU is suitable for these tasks. The algorithm was implemented as a devices for obtaining uncertainty function for several types of signals average performance FPGA and GPU. The experiment has confirmed the gain in signal-to-noise ratio as compared with the simple correlator.
References
1. Skolnik M. Radar Handbook. V.1, McGraw-Hill, BookCompany,1970,
456 p.
2. Gonorovsky I.S. Radio circuits and signals. M., Radio and communication, 1986, 512 p.
3. Shirman A.D., Manzshos A.D. Theory and technique of processing of radar information on the background noise. M., Radio and communication, 1981, 416 p.
4. Lyons R.G. Understanding digital signal processing. New Jersey., Prentice Hall, 2001, 517 p.
If you have any question please write: info@dspa.ru