CN100440733C - A/D converter with high effective quantization to signal - Google Patents
A/D converter with high effective quantization to signal Download PDFInfo
- Publication number
- CN100440733C CN100440733C CNB2005101353660A CN200510135366A CN100440733C CN 100440733 C CN100440733 C CN 100440733C CN B2005101353660 A CNB2005101353660 A CN B2005101353660A CN 200510135366 A CN200510135366 A CN 200510135366A CN 100440733 C CN100440733 C CN 100440733C
- Authority
- CN
- China
- Prior art keywords
- signal
- converter
- input
- noise
- individual
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Abstract
This invention relates to a AD converter to the signal high efficient quantization, and it uses parallel processing structure with B' channels, which including the noise generator, B' dual-channel adders, B' comparators, multi-channel adder, in which B'=2B-1, B' being parallel channel number, B being AD converter bit number. The said noise generator generates B' independent same distribution white noises, respectively inputs B' parallel channels and overlays with the quantified input signal s in the dual-channel adder, and the output of the dual-channel adder respectively as one input of the B' comparators, and another input of the comparators respectively import threshold signal etai, in which i=1, 2,..., B'; the output of the B' comparators connects with the input of the multi-channel adder. This invention has the advantage: using the invention, it can use few quantify bits to get high-quality digital signals. The invention can be used in the fields such as signal processing, high-speed data transmission, and image compression, and others.
Description
Technical field
The invention belongs to digital processing field such as radar, sonar, image, particularly a kind of analog to digital converter to the signal high effective quantization.
Background technology
Natural signal is normally continuous on time and amplitude, and digital information processing system requires input signal to disperse in time, and is expressed as the Serial No. formed by 0,1.Therefore, analog signal need be converted into digital form.Fig. 1 has represented the basic composition of digital information processing system.The analog signal that is obtained by signal source 101 at first is converted into discrete-time signal by a sampling and holding circuit 102.Signal by sampling and holding circuit 102 outputs is input to an analog/digital (A/D) transducer 103, and the analog signal of elapsed time sampling is converted into digitally coded signal there.Digital Signal Processing (DSP) system 104 just can use these digital signals to finish specific digital signal processing algorithm then.According to different application, the output of dsp system can directly use digital form to output to display system 105, perhaps returns analog signal by 106 conversion of a digital-to-analog (D/A) transducer.
The amplitude range that input signal is set is
Then the layering of B bit A/D transducer is differential is Q=A2
-(B-1)If input signal is s, the signal after then conventional A/D converter quantizes is
Wherein
Expression is no more than
Maximum integer.
Temporal sampling problem can well be solved by sampling thheorem, and still, the quantization effect on the signal amplitude is but because its non-linear characteristics are far from being understood fully.Some A/D converters commonly used, such as (.Digital signal processing handbook.CRC Press such as V.K.Madisetti such as FlashA/D, Successive Approximation A/D, Pipelined A/D, Cyclic A/D, Boca Raton.1999), need inhibit signal waveform as far as possible, promptly reduce the nonlinear transformation effect on signal amplitude as far as possible.Resolution and lower quantization error in order to improve quantification should improve quantizing bit number, but the transmittability to multi channel signals has constituted great challenge in a lot of application such as Array Signal Processing, sensor network.
Fig. 2 is the schematic diagram of FlashA/D transducer.Because used highly-parallel to handle structure, the FlashA/D transducer can reach very high conversion speed.The Flash A/D converter has comprised analog comparator 201 and digital encoder 202.Comparative voltage V0, the V1 of analog comparator 201 ..., V (2
B-1) in the input voltage amplitude range, evenly distributes from small to large.Behind encoder 202 codings, the Serial No. 203 of Flash A/D converter output B bit.
Under the situation of little input signal, because serious distortion has taken place in nonlinear effect output waveform, can adopt way (the B.And ò .Stochastic Resonance:Theory andApplications.Kluwer Academic Publishers of " shake " on the engineering, Boston.2000), at signal input part adding one noise (equaling maximum quantization error) in a small amount, approach linear response.This method only be on the basis of conventional A/D converter with transfer function " linearisation ", do not change the essence and the structure of conventional A/D converter.In order to obtain the high digital signal of quality, still need the high quantization bit number.
In some non linear systems, the existence of noise can improve the transmittability of signal, and this phenomenon is called accidental resonance (.Stochastic resonance.Reviews of Modern Physics.1998:70 (1) .pp.223-287 such as Gammaitoni).Simple quantization system, such as two-value quantize or three value quantization systems in also can occur Stochastic Resonance Phenomenon (.Design of detectors based on stochastic resonance.SignalProcessing such as Saha, 2003:83.pp.1193-1212).In these systems, output signal-to-noise ratio often is higher than input signal-to-noise ratio, and therefore, this provides theoretical foundation for adopting less quantizing bit number to obtain high-quality digital signal.
Summary of the invention
The objective of the invention is to adopt the high quantization bit number to obtain high quality digital signal in order to solve conventional A/D converter, and the defective of the serious distortion of the output waveform of little input signal situation behind A/D converter, a kind of A/D converter to the signal high effective quantization is provided.
For achieving the above object, analog to digital converter to the signal high effective quantization provided by the invention (A/D converter) is realized by parallel processing structure shown in Figure 3, comprise noise generator 301, the individual binary channels adder 302 of B ', the individual comparator 303 of B ', multichannel adder 304, wherein B '=2
B-1, B ' is the parallel channel number, and B is the bit number of analog to digital converter.Need the signal s that quantizes to be input to the individual parallel channel of B ', noise generator 301 produces the individual independent same distribution white noise of B ' ξ
1, ξ
2..., ξ
B 'Import the individual parallel channel of B ' respectively and in binary channels adder 302 with input signal s to be quantified stack, as an input of the individual comparator 303 of B ', another input of comparator 303 is input threshold signal η respectively respectively for the output of binary channels adder 302
i, i=1 wherein, 2 ..., B '; The output of the individual comparator 303 of described B ' is connected with the input of multichannel adder 304.
In the technique scheme, in the described comparator on i passage, if the mixed signal s+ ξ of its input
iThreshold signal η greater than i passage
i, the output s of this comparator then
QiEqual 1, otherwise s
QiEqual 0.
In the technique scheme, described input signal s is an analog signal, or through the digital signal (the quantified precision height approaches analog signal) after the conventional higher bit analog to digital converter conversion.
In the technique scheme, the individual independent same distribution white noise of described B ' comprises even partition noise, Gaussian Profile noise, and more the Generalized Gaussian noise and the mixed Gaussian noise of broad sense, and the noise of other type.
In the technique scheme, the individual independent same distribution white noise of described B ', its variance equates.
In the technique scheme, the threshold signal of the individual comparator of described B ', its numerical value equates.
With the FlashA/D transducer of Fig. 2 relatively, the present invention has also adopted the parallel processing structure of height, but has increased noise generator 301 and two kind of adder 302 and 304, the coded system of numeral output is also different: because by the output s after the multichannel adder 304
qBe integer value, so do not need to encode again in addition by binary representation.In addition, though Flash A/D converter and accidental resonance A/D converter of the present invention all comprise comparator (being respectively 201 and 303), the thresholding η of comparator 303 among the present invention
i(i=1,2 ..., B ') and numerical value equates, as a rule, η
i(i=1,2 ..., B ') and can all be set to 0.
Compare with common A/D converter, the invention has the advantages that: can improve signal to noise ratio (equivalence substantially under the Gaussian noise) in output of the present invention under the non-Gaussian noise situation; Output of the present invention is not subjected to the influence of little input signal; In the quantification of image, the result after the present invention quantizes when hanging down quantizing bit number has obviously better visual effect.Therefore adopt the present invention can use quantizing bit number seldom just to obtain high-quality digital signal.The present invention can be applied in the fields such as signal processing, high speed data transfer and image compression.
Description of drawings
Fig. 1 is the digital information processing system structural representation that A/D converter is used in expression;
Fig. 2 is an expression Flash A/D converter structural representation;
Fig. 3 is the structural representation of expression accidental resonance A/D converter of the present invention;
Fig. 4 is the noise generator schematic diagram that expression adopts microprogram control unit to realize;
Fig. 5 is the block diagram of the basic unit of explanation industry control type computer;
Fig. 6 is common A/D converter and the quantizing bit number of accidental resonance A/D converter in Gaussian noise and the variation relation figure of detection probability;
Fig. 7 is the digital information processing system structural representation that the digital signal after conventional higher bit A/D converter is changed in expression is imported as the accidental resonance A/D converter;
Fig. 8 a-Fig. 8 c is common A/D converter of expression and the performance comparison diagram of accidental resonance A/D converter in image quantization;
Fig. 9 is the application of accidental resonance A/D converter in the figure image intensifying.
Embodiment
The present invention is a kind of analog to digital converter that utilizes accidental resonance to the signal high effective quantization, below in conjunction with accompanying drawing accidental resonance A/D converter of the present invention is elaborated.As Fig. 1, A/D converter 103 is the accidental resonance A/D converters that adopt circuit structure to realize, and the analog signal of the signal source 101 that receives changed through accidental resonance A/D converter 103, to reach the purpose of using less bit number just can obtain high quality digital signal.This example adopts detection mode to weigh the quality height of output digital signal.As Fig. 3, the binary channels adder 302 in the accidental resonance A/D converter adopts the analog adder device, and comparator 303 adopts adjustable reference analog comparator device, and multichannel adder 304 adopts long number adder device.Noise generator 301 can be realized by the microprogram control unit programming.As shown in Figure 4, various types of noise can use algorithm and program ripe in the document to calculate, and procedure stores is in microprogram control unit 401, and the noisy digit sequence that obtains is passed through the analogue noise that D/A converter 402 obtains B ' road again.
In the present embodiment, simulation signal generator 101 produces the mixed signal of noise (white Gaussian noise of voltage variance 1) and sinusoidal signal (voltage amplitude 1), is input to accidental resonance A/D converter 103 through sampling and holding circuit 102.The ξ that noise generator 301 produces in the accidental resonance A/D converter 103
iBe that standard variance is 0.5 even partition noise, the thresholding η in the comparator 303
iAll be transferred to 0 value.
Digital information processing system 104 can be realized by industrial computer 501.As Fig. 5, after the digital signal of accidental resonance A/D converter 103 output transmits by first network interface card 502 at first by be transformed into positive and negative value signal (deducting B '/2) on the occasion of signal, detect sinusoidal signal by matched filter detector then, its program is to use algorithm ripe in the document to realize, is stored in the industry control type computer 501.Second network interface card 503 by industry control type computer 501 sends result to display 105.
Real point line among Fig. 6, imaginary point line are respectively fixedly after the false alarm probability 0.1, the detection probability result after the accidental resonance A/D converter that obtains under the different quantizing bit number B and common A/D converter quantize.The square frame line is direct result to analog signal detection.See from Fig. 6, can not make full use of at input signal under the situation of A/D conversion dynamic range, low quantizing bit number, the result after common A/D converter quantizes during such as 1 to 3 bit far is worse than the accidental resonance A/D converter.Signal is used the accidental resonance A/D converter,, also can obtain very high-quality digital signal even use quantizing bit number seldom.
As shown in Figure 7, present embodiment is that the digital signal after conventional higher bit A/D converter 701 conversions is changed through accidental resonance A/D converter 703 of the present invention again, to reach the purpose of using less bit number just can obtain high quality digital signal.This example adopts detection mode to weigh the quality height of output digital signal.As shown in Figure 7, the input signal 702 of accidental resonance A/D converter 703 is the results after sinusoidal signal (voltage amplitude 1) in the noise (white Gaussian noise of voltage variance 1) the conventional A/D converter 701 that uses 16 bit quantizations (maximum input voltage be set at ± 8) quantizes.Signal 702 is sent to accidental resonance A/D converter 703, and this accidental resonance A/D converter 703 adopts the industry control type computer of buying on the market 501; And noise generator 301, binary channels adder 302, comparator 303, multichannel adder 304 supervisors are stored in the industry control type computer 501, as shown in Figure 3, the ξ that produces of noise generator 301 wherein
iBe that standard variance is 0.5 even partition noise, the thresholding η in the comparator 303
iAll be made as 0.Evenly partition noise can use algorithm ripe in the document and program to realize.After signal 702 transmits by first network interface card 502, after handling, accidental resonance A/D converter 703 is input to digital information processing system 704, digital information processing system 704 at first is transformed into positive and negative value signal (deducting B '/2) with what accidental resonance A/D converter 703 was exported on the occasion of signal in this example, detect sinusoidal signal by matched filter detector then, its program also is to use algorithm ripe in the document to realize, is stored in the industry control type computer 501.Second network interface card 503 by industry control type computer 501 sends result to display 705.
Real point line among Fig. 6, imaginary point line are respectively fixedly after the false alarm probability 0.1, the detection probability result after the accidental resonance A/D converter that obtains under the different quantizing bit number B and common A/D converter quantize.The square frame line is direct result to analog signal detection.See from Fig. 6, can not make full use of at input signal under the situation of A/D conversion dynamic range, low quantizing bit number, the result after common A/D converter quantizes during such as 1 to 3 bit far is worse than the accidental resonance A/D converter.Signal is used the accidental resonance A/D converter,, also can obtain very high-quality digital signal even use quantizing bit number seldom.
Below in conjunction with accompanying drawing accidental resonance A/D converter of the present invention is elaborated.As Fig. 7, present embodiment is that the data image signal after conventional higher bit A/D converter 701 conversions is changed through accidental resonance A/D converter 703 of the present invention again, to reach the purpose of using less bit number just can obtain the quality digital image.As shown in Figure 7, the input signal 702 of accidental resonance A/D converter 703 is to use the Lena figure after the conventional A/D converter 701 of 8 bit quantizations quantizes.To be sent to accidental resonance A/D converter 703 after gradation of image signal 702 normalization (between [0.5,0.5]), this accidental resonance A/D converter 703 adopts the industry control type computer of buying on the market 501; And gradation of image normalization calculating, noise generator 301, binary channels adder 302, comparator 303, multichannel adder 304 supervisors are stored in the industry control type computer 501, as shown in Figure 3, the thresholding η in the comparator 303 wherein
iAll be made as 0, the ξ that noise generator 301 produces
iBe even partition noise, intensity illustrates below.Evenly partition noise can use algorithm ripe in the document and program to realize.After picture signal 702 transmitted by first network interface card 502, the output image after accidental resonance A/D converter 703 is handled sent display 705 to by second network interface card 503 of industry control type computer 501.
Fig. 8 a-Fig. 8 c is respectively to importing the figure as a result after Lena figure takes 1,2,3 bit quantizations, the corresponding common A/D converter in the left side, the corresponding accidental resonance A/D converter in the right.Even partition noise ξ in the accidental resonance A/D converter
iStandard variance be respectively 0.05,0.07,0.10.Can see that both image display effects are different fully when hanging down quantizing bit number, the figure as a result after the accidental resonance A/D converter quantizes has obviously better visual effect.Therefore, accidental resonance A/D transformation energy is applied in Signal Compression and the field of signal transmissions.
Fig. 9 is submerged in (standard variance is 0.3 mixed Gaussian noise) in the very noisy with input Lena figure, adopts the result behind common A/D converter (left side) and accidental resonance A/D converter (the right) 8 bit quantizations.Even partition noise ξ in the accidental resonance A/D converter
iStandard variance be 0.10.As seen accidental resonance A/D conversion can improve signal noise ratio (snr) of image greatly.
Claims (6)
1, a kind of analog to digital converter to the signal high effective quantization adopts the parallel processing structure with the individual passage of B ', comprises the individual comparator of B ', it is characterized in that,
Also comprise noise generator, the individual binary channels adder of B ', multichannel adder, wherein B '=2
B-1, B ' is the parallel channel number, and B is the bit number of analog to digital converter;
Described noise generator produces the individual independent same distribution white noise of B ' ξ
1, ξ
2..., ξ
B ', import the individual parallel channel of B ' respectively and in the binary channels adder with input signal s to be quantified stack, as an input of the individual comparator of B ', another input of comparator is input threshold signal η respectively respectively for the output of binary channels adder
i, i=1 wherein, 2 ..., B '; The output of the individual comparator of described B ' is connected with the input of multichannel adder.
2, by the described analog to digital converter of claim 1, it is characterized in that, in the described comparator on i passage, if the mixed signal s+ ξ of its input to the signal high effective quantization
iThreshold signal η greater than i passage
i, the output s of this comparator then
QiEqual 1, otherwise output s
QiEqual 0; I=1 wherein, 2 ..., B '.
3, by the described analog to digital converter of claim 1, it is characterized in that described input signal s is an analog signal to the signal high effective quantization, or the digital signal of process high accuracy quantification.
4, by the described analog to digital converter of claim 1 to the signal high effective quantization, it is characterized in that the individual independent same distribution white noise of described B ' comprises even partition noise, Gaussian Profile noise, and more the Generalized Gaussian noise and the mixed Gaussian noise of broad sense.
5, by the described analog to digital converter of claim 1, it is characterized in that the signal high effective quantization, the individual independent same distribution white noise of described B ', its variance equates.
By the described analog to digital converter of claim 1, it is characterized in that 6, the numerical value of the threshold signal of the individual comparator of described B ' equates to the signal high effective quantization.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005101353660A CN100440733C (en) | 2005-12-31 | 2005-12-31 | A/D converter with high effective quantization to signal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005101353660A CN100440733C (en) | 2005-12-31 | 2005-12-31 | A/D converter with high effective quantization to signal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1992531A CN1992531A (en) | 2007-07-04 |
| CN100440733C true CN100440733C (en) | 2008-12-03 |
Family
ID=38214521
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005101353660A Expired - Fee Related CN100440733C (en) | 2005-12-31 | 2005-12-31 | A/D converter with high effective quantization to signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100440733C (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103929177B (en) * | 2014-04-29 | 2017-02-08 | 武汉大学 | A/D quantization bit conversion system and method in GNSS receiver |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0308982B1 (en) * | 1987-09-25 | 1995-09-06 | Nec Corporation | Analog-to-digital converter having an excellent signal-to-noise ratio for small signals |
| US6307492B1 (en) * | 1999-05-28 | 2001-10-23 | Thomson-Csf | Device to compensate for the non-linearity of an analog/digital converter |
| CN1595180A (en) * | 2004-07-09 | 2005-03-16 | 天津大学 | Frequency self-tracing dimension-variation stochastic resonance fault diagnosis system |
-
2005
- 2005-12-31 CN CNB2005101353660A patent/CN100440733C/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0308982B1 (en) * | 1987-09-25 | 1995-09-06 | Nec Corporation | Analog-to-digital converter having an excellent signal-to-noise ratio for small signals |
| US6307492B1 (en) * | 1999-05-28 | 2001-10-23 | Thomson-Csf | Device to compensate for the non-linearity of an analog/digital converter |
| CN1595180A (en) * | 2004-07-09 | 2005-03-16 | 天津大学 | Frequency self-tracing dimension-variation stochastic resonance fault diagnosis system |
Non-Patent Citations (1)
| Title |
|---|
| 利用随机共振技术的微弱信号方位估计. 叶青华，黄海宁，何心怡，张春华.声学学报(中文版)，第29卷第4期. 2004 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1992531A (en) | 2007-07-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7822698B1 (en) | Spike domain and pulse domain non-linear processors | |
| CN103049243B (en) | True random-number generating method and device thereof | |
| CN105072359A (en) | CMOS (Complementary Metal Oxide Semiconductor) image sensor imaging system and method based on compressed sensing | |
| CN100440733C (en) | A/D converter with high effective quantization to signal | |
| CN107276932B (en) | Blind source signal separation method and device | |
| CN109936360A (en) | Pulse counting equipment and radiation detecting system | |
| Feizi et al. | Time-stampless adaptive nonuniform sampling for stochastic signals | |
| CN106788435B (en) | Quantify sampling noise-reduction method | |
| CN110677138B (en) | FIR filter based on error-free probability calculation | |
| CN113298778B (en) | Depth calculation method and system based on flight time and storage medium | |
| CN110297199B (en) | Frequency measurement method and system for cesium optical pump magnetometer based on full-phase FFT | |
| CN203894324U (en) | Phase difference detection device of fiber current transformer | |
| CN104113337A (en) | Streamline analog-to-digital converter | |
| US8531223B2 (en) | Signal generator | |
| CN101217665A (en) | A parallel line scanning method of video frequency | |
| Gharib | Conserved quantities and fluxes for some nonlinear evolution equations | |
| CN112911175B (en) | Noise reduction system for low-light-level imaging of CMOS image sensor | |
| CN113327261B (en) | Error-free resource image edge detection operator method based on random computation | |
| Jil et al. | Identification of systems with logarithmic quantizer | |
| RU105550U1 (en) | DATA TRANSFER INTERFERENCE SYSTEM | |
| CN113917966B (en) | Switched capacitor array chip direct current bias real-time correction method | |
| CN113472404B (en) | Method and device for optimizing digital domain beam forming based on condition generation countermeasure network | |
| Ye et al. | Stochastic Resonance AD Conversion and its Effect on Image Enhancement | |
| Baxtiyarovna et al. | ALGORITHMIC SUPPORT FOR PRE-PROCESSING DEVICES SPECTROMETRIC SIGNALS | |
| Daiyrbayeva et al. | Using wavelet transform in image processing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20081203 Termination date: 20111231 |