GB2236452A - Determining interpolation weighting coefficients for low ratio sampling rate converter - Google Patents
Determining interpolation weighting coefficients for low ratio sampling rate converter Download PDFInfo
- Publication number
- GB2236452A GB2236452A GB9015252A GB9015252A GB2236452A GB 2236452 A GB2236452 A GB 2236452A GB 9015252 A GB9015252 A GB 9015252A GB 9015252 A GB9015252 A GB 9015252A GB 2236452 A GB2236452 A GB 2236452A
- Authority
- GB
- United Kingdom
- Prior art keywords
- coefficients
- sampling rate
- output
- input
- sampling
- 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.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/0223—Computation saving measures; Accelerating measures
- H03H17/0227—Measures concerning the coefficients
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/06—Non-recursive filters
- H03H17/0621—Non-recursive filters with input-sampling frequency and output-delivery frequency which differ, e.g. extrapolation; Anti-aliasing
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/06—Non-recursive filters
- H03H17/0621—Non-recursive filters with input-sampling frequency and output-delivery frequency which differ, e.g. extrapolation; Anti-aliasing
- H03H17/0635—Non-recursive filters with input-sampling frequency and output-delivery frequency which differ, e.g. extrapolation; Anti-aliasing characterized by the ratio between the input-sampling and output-delivery frequencies
- H03H17/065—Non-recursive filters with input-sampling frequency and output-delivery frequency which differ, e.g. extrapolation; Anti-aliasing characterized by the ratio between the input-sampling and output-delivery frequencies the ratio being integer
- H03H17/0664—Non-recursive filters with input-sampling frequency and output-delivery frequency which differ, e.g. extrapolation; Anti-aliasing characterized by the ratio between the input-sampling and output-delivery frequencies the ratio being integer where the output-delivery frequency is lower than the input sampling frequency, i.e. decimation
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/06—Non-recursive filters
- H03H17/0621—Non-recursive filters with input-sampling frequency and output-delivery frequency which differ, e.g. extrapolation; Anti-aliasing
- H03H17/0635—Non-recursive filters with input-sampling frequency and output-delivery frequency which differ, e.g. extrapolation; Anti-aliasing characterized by the ratio between the input-sampling and output-delivery frequencies
- H03H17/0685—Non-recursive filters with input-sampling frequency and output-delivery frequency which differ, e.g. extrapolation; Anti-aliasing characterized by the ratio between the input-sampling and output-delivery frequencies the ratio being rational
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/01—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/01—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level
- H04N7/0135—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level involving interpolation processes
Abstract
A technique for reducing the number of coefficients in a low ratio sampling rate converter divides each input sampling rate interval S1-S2 into a plurality of bins i.e. smaller intervals. For each bin a set of conversion coefficients is computed. So long as output data samples lie within the same bin, the same set of coefficients is applied. When the output data samples move to another bin, the set of coefficients applied is changed. Thus a finite set of coefficients are computed rather than computing an exact set for each output data sample. Conversions between PAL and NTSC colour subcarrier sampling rates are described. <IMAGE>
Description
::P1:2 -:3 cl -cl COEFFICIENT REDUCTION IN A LOW RATIO SAMPLING RATE
CONVERTER
Background of the Invention
The present invention relates to sampling rate converters, and more particularly to a technique for reducing the number of coefficients in a low ratio sampling rate converter to minimize memory and maximize speed requirements where the ratio between the sampling rates is close to one.
Many different standards have been and are being created in both the video and audio worlds. With the sharp increase in the usage of digital technology in the television industry as well as the advent of high definition television, many new standards have recently emerged and many new ones are being proposed. one of the specifications in all these standards is the sampling rate, with different standards specifying different sampling rates. In one dimensional form the sampling rate specifies the spacing between adjacent samples, while in the two and three dimensional forms for television video it specifies the spacing between horizontal lines and the frame rate, respectively. Therefore there is a 2 growing need for sampling rate converters so that systems using different specifications can communicate with each other.
To convert the sampling rate the signal values at the locations between the samples at the first sampling rate have to be evaluated. For this purpose a kernel is used to interpolate between the samples. Ideally a sinc, or sin(x)/x, function is the kernel. However, as the sinc function is an infinitely long function, various different time limited functions are used instead. The kernel has to be long enough that N input samples are required to calculate one output sample. Therefore there are X sampled values of the kernel that become the coefficients to multiply with the input data values. If the relationship, or separations, between the locations of the input and output samples vary for different output samples, as is the case when the input and output sampling rates are different, then a different set of N coefficients, the sampled values of the kernel, is required for each output sampled. The number of such sets required is a function of input and output sampling rates. if the ratio of the input and output sampling rates is close to one, then a very large set of coefficients is needed.
For example in video one sampling rate is four times the color subcarrier frequency, or 14.318180 M5amples/sec for NTSC and 17.734475 les/sec for PAL, another rate is the CCIR 601 standard of 13.5 M5amples/sec, and yet another rate is a PAL "line locked sampling rateO of 17.734375 MSamples/sec. Taking the two PAL sampling rates, the difference is only 100 Hz equivalent to a ratio of approximately 1. 000005639:1. In this particular case there are 3 709379 distinct output sample locations, and therefore 709379xN sample values of the convolution kernel. For N=10 this implies a calculation of more than 7x106 coefficient values. At video signal rates these values cannot be easily computed on the fly. To alternatively precalculate and store these coefficient values leads to a large memory requirement. Since each different output sample requires a different set of coefficients, memory access rate needs to be at the video output rate of 17.734475 MHz, i.e., the memory cannot be slow, and thus low cost.
What is desired is a method of reducing the number of coefficients in a sampling rate converter so that a slow, low cost memory may be used to prestore the necessary coefficients.
Summary of the Invention
Accordingly the present invention provides a technique for reducing the number of coefficients for a filter in a low ratio sampling rate converter by dividing the sampling interval of the input sampling rate into a finite number of bins and precalculating the coefficients for the bins to produce the output sampling rate. A finite input response (FIR) filter receives at its input at the output sampling rate a plurality of samples that were obtained at the input sampling rate. A coefficient memory contains the coefficients for each bin. When the output samples move to the next bin. a coefficient controller accesses the coefficient memory to output a new set of coefficients for the FIR filter. Since there are multiple samples per bin when the ratio is very nearly one, a slow, in ensive memory may be used as the 4 coefficient memory. Also for on the fly operation the coefficients may be calculated at a rate significantly less than the video rates, eliminating the need for the coefficient memory altogether.
The objects, advantages and other novel features of the present invention are apparent from the following detailed description when read in conjunction with the appended claims and attached drawing.
Brief Description of the Drawincr
Fig. 1 is a graph diagram representing data rate conversion between two sampling rates.
Fig. 2 is a graph diagram representing data rate conversion between two sampling rates using bins according to the present invention.
Fig. 3 is a block diagram of a variable coefficient FIR filter to perform data rate conversion according to the present invention.
Description-of the Preferl:ed Embodiment
To reduce the number of coefficients the problem is simplified by considering the fact that the accuracy desired in the output signal samples is of the order of b bits. For a sinusoid of frequency f Ez, the peak error, e, in the amplitude measurement, due to an inaccuracy of +/-dt in the sampling instant, t, is given by e = 2Asin(cfdt) = 2Acfdt (1) where A is the amplitude of the sinusoid. An accuracy of at least b bits implies that the maximum error, e, can be equal to Ax2-b.
Substituting for e in - equation (1) and rearranging the terms, dt = 1/(2cf2b) (2) In other words the time jitter, dt, is equivalent to an amplitude error of less than one quantization level using b bits.
From equation (2) for a sinusoid of 5 MHz, to obtain an accuracy of at least eight bits, the time jitter should not exceed +/-124.34 picoseconds. Similarly for a sinusoid of 5.5 MHz a desired output accuracy of at least 10 bits is equivalent to a time jitter less than +/- 28.26 picoseconds in the sampling instants.
Referring now to Fig. 1 the input data samples at a first sampling rate are multiplied by coefficients represented by dots on the respective superimposed kernels that are centered on the respective output data samples at a second sampling rate. Therefore each output data sample has a different set of coefficients. However as shown in Fig. 2 the duration of one input sampling period T may be split into T/(2dt) intervals, or bins. All the output sample values corresponding to sampling instants within one bin are approximated by the sample value corresponding to the center of that interval. As long as the locations of the output samples stay within the same bin, the same set of coefficients is used.
Therefore to obtain a ten bit accuracy for a sinusoid of 5.5 MHz sampled at a rate of 17.734375 MHz, the sampling period is divided into 2000 bins.
each bin is characterized by one set of coefficients so that only 200OxN values are required. For N=10 this implies a storage of only 20,000 coefficients, as compared to more than 7x106 required for the exact evaluation of the output samples. If each coefficient 6 requires two bytes of storage space, the overall memory requirement is reduced from more than 14 Mbytes to only 40 kbytes, achieving over two orders of reduction in the storage. Also the coefficients need be changed once for every 177 or 178 output samples.
Therefore the coefficient memory may be a slow memory, further reducing the cost. Alternatively since the coefficients need be computed once for every 177 or 178 output samples, they now may be calculated on the fly, thus eliminating the storage requirement altogether.
A device for converting data between two sampling rates is shown in Fig. 3. An input data signal, such as a video signal sampled at 17.734375 MHz, is input to a first-in/first-out (FIFO) buffer register 10.
The input data signal is clocked into the FIFO register 10 at a first clock rate CM, and clocked out at a second clock rate CLK2 corresponding to the output sample rate, such as 17.734475 MHz. A clock generator 12 converts the input sample rate CLK1 into the output sample rate CLK2. A FIFO controller 14 controls the data into and out of the FIFO register 10. For precalculated coefficients a coefficient memory 16 stores the necessary sets of coefficients.
The coefficients used from the coefficient memory 16 are determined by a coefficient controller 18. The coefficients from the coefficient memory 16 and the data from the FIFO register 10 are output over respective buses for input to a FIR filter 20. The data is delayed by appropriate delay intervals established by delay lines 22, and then multiplied by the appropriate coefficients in multipliers 24.
Samples are then combined in summers 26, and the outputs from all the slimmers are combined in an adder 7 28 to provide the desired output data samples at the output sample rate.
Alternatively the coefficient controller 18 and coefficient memory 16 may be replaced with a microprocessor 30 that computes the coefficients on the fly as needed for every 177 or 178 samples for the example referred to above. Then the controller 18 becomes a computer and the memory 16 becomes the program and working memories for the computer.
Thus the present invention reduces the number of coefficients required for a sampling rate converter by dividing each input sampling period into a finite number of bins, calculating sets of coefficients for each bin, and applying the coefficients to a FIR filter to produce the output data at the output sampling rate, the coefficients being changed only when the output data falls into a new bin.
CLADS: 1. A method of reducing the nu of coefficients for a low ratio swpling rate converter ccup=sang the steps of: dividing each sampling interval for an input sampling rate into a finite number of intervals; 5 computing a set of filter coefficients for each interval; and applying the set of filter coefficients for each interval to input data to produce output data at an output sampling rate, the set of filter coefficients changing for each sample of the output data that falls in a different one of the intervals.
2. An apparatus for converting data from an input sampling rate to an output sampling rate having a low ratio comprising:
means for temporarily storing the data, the data being input to the temporary storing means at the input sampling rate and being output from the temporary storing means at the output sampling rate; means for computing a plurality of sets of coefficients corresponding to a plurality of intervals into which the sample interval for the input sampling rate is divided; and means for applying the sets of coefficients to the data from the temporary storing means such that the same set of coefficients are applied to the data so long as the data falls within the corresponding interval, the set of coefficients used by the applying means changing for each interval within which the data falls.
9 3. A method of reducing the number of coefficients for a low ratio sampling rate converter substantially as herein described with reference to and as shown in Figures 2 and 3 of the accompanying drawings. 5 4. An apparatus for converting data input substantially as herein described with reference to and as shown in Figures 2 and 3 of the accompanying drawings.
Published 1991 at The Patent Office, State House. 66171 High Holborn. London WC I R 41P. Further copies may be obtained from V- W--f WM 7147. Plited bv MultiDlex techniques lid, St Mary Cray. Kent.
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US37985089A | 1989-07-14 | 1989-07-14 |
Publications (3)
Publication Number | Publication Date |
---|---|
GB9015252D0 GB9015252D0 (en) | 1990-08-29 |
GB2236452A true GB2236452A (en) | 1991-04-03 |
GB2236452B GB2236452B (en) | 1993-12-08 |
Family
ID=23498974
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
GB9015252A Expired - Fee Related GB2236452B (en) | 1989-07-14 | 1990-07-11 | Coefficient reduction in a low ratio sampling rate converter |
Country Status (3)
Country | Link |
---|---|
JP (1) | JPH07120924B2 (en) |
DE (1) | DE4022387A1 (en) |
GB (1) | GB2236452B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB2245124A (en) * | 1990-04-11 | 1991-12-18 | Rank Cintel Ltd | Spatial transformation of video images |
JPH07120924B2 (en) | 1989-07-14 | 1995-12-20 | テクトロニクス・インコーポレイテッド | Sampling frequency converter |
GB2315625A (en) * | 1996-07-17 | 1998-02-04 | Roke Manor Research | Interpolating filters |
GB2356506A (en) * | 1999-09-25 | 2001-05-23 | Eastman Kodak Co | Multiple dimension interpolator |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3066241B2 (en) * | 1993-12-28 | 2000-07-17 | 株式会社エフ・エフ・シー | Digital filter and oversampling type analog / digital converter using the digital filter |
US6927826B2 (en) | 1997-03-26 | 2005-08-09 | Semiconductor Energy Labaratory Co., Ltd. | Display device |
JPH11307782A (en) | 1998-04-24 | 1999-11-05 | Semiconductor Energy Lab Co Ltd | Semiconductor device and its manufacture |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS59105712A (en) * | 1982-12-09 | 1984-06-19 | Fujitsu Ltd | Digital filter |
JPH0793548B2 (en) * | 1987-08-31 | 1995-10-09 | 三洋電機株式会社 | Sampling frequency conversion circuit |
US4837619A (en) * | 1987-10-28 | 1989-06-06 | North American Philips Corporation | Scan rate conversion apparatus and method |
JP2600236B2 (en) * | 1987-12-29 | 1997-04-16 | ソニー株式会社 | Sampling frequency conversion circuit |
GB2236452B (en) | 1989-07-14 | 1993-12-08 | Tektronix Inc | Coefficient reduction in a low ratio sampling rate converter |
US5023825A (en) | 1989-07-14 | 1991-06-11 | Tektronix, Inc. | Coefficient reduction in a low ratio sampling rate converter |
-
1990
- 1990-07-11 GB GB9015252A patent/GB2236452B/en not_active Expired - Fee Related
- 1990-07-13 DE DE19904022387 patent/DE4022387A1/en active Granted
- 1990-07-13 JP JP2186981A patent/JPH07120924B2/en not_active Expired - Fee Related
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH07120924B2 (en) | 1989-07-14 | 1995-12-20 | テクトロニクス・インコーポレイテッド | Sampling frequency converter |
GB2245124A (en) * | 1990-04-11 | 1991-12-18 | Rank Cintel Ltd | Spatial transformation of video images |
GB2315625A (en) * | 1996-07-17 | 1998-02-04 | Roke Manor Research | Interpolating filters |
GB2315625B (en) * | 1996-07-17 | 2001-02-21 | Roke Manor Research | Improvements in or relating to interpolating filters |
GB2356506A (en) * | 1999-09-25 | 2001-05-23 | Eastman Kodak Co | Multiple dimension interpolator |
Also Published As
Publication number | Publication date |
---|---|
DE4022387A1 (en) | 1991-01-24 |
JPH07120924B2 (en) | 1995-12-20 |
DE4022387C2 (en) | 1992-11-19 |
JPH0372711A (en) | 1991-03-27 |
GB9015252D0 (en) | 1990-08-29 |
GB2236452B (en) | 1993-12-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US5023825A (en) | Coefficient reduction in a low ratio sampling rate converter | |
AU634605B2 (en) | A high speed interpolation filter for television standards conversion | |
US6141671A (en) | Asynchronous digital sample rate converter | |
US7262716B2 (en) | Asynchronous sample rate converter and method | |
CA1193014A (en) | Direct digital to digital sampling rate conversion method and apparatus | |
US5621404A (en) | Digital-to-digital sample rate converter | |
US7254281B2 (en) | Method and apparatus for electronic image processing | |
EP0837561B1 (en) | Sampling frequency converting apparatus | |
JPH06268477A (en) | Sampling frequency converter | |
EP0305864B1 (en) | Improved sampling frequency converter for converting a lower sampling frequency to a higher sampling frequency and a method therefor | |
US5182633A (en) | Video sample rate converter | |
US5159339A (en) | Sampling rate converter for signals having a non-integer sampling ratio | |
EP2374125A1 (en) | Multi-staging recursive audio frame-based resampling and time mapping | |
EP0450817B1 (en) | Digital phase detector arrangements | |
GB2236452A (en) | Determining interpolation weighting coefficients for low ratio sampling rate converter | |
EP0706262B1 (en) | Filter selection circuit for digital resampling system | |
JPH11122080A (en) | Device for reducing data speed | |
EP0103355B1 (en) | Analog to digital converting system for a video signal | |
US5459525A (en) | Video signal converting device and noise eliminator used therein | |
US5537154A (en) | Edge compensation method and apparatus of image signal | |
GB2240684A (en) | Cascaded digital signal interpolator | |
CA2102936C (en) | Combined aligner blender | |
JPH0568156B2 (en) | ||
JPH07264626A (en) | Sampling frequency converting circuit | |
JPH05291881A (en) | Digital filter device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PCNP | Patent ceased through non-payment of renewal fee |
Effective date: 20030711 |