JPH047976A - Butterfly arithmetic circuit - Google Patents

Butterfly arithmetic circuit

Info

Publication number
JPH047976A
JPH047976A JP2107520A JP10752090A JPH047976A JP H047976 A JPH047976 A JP H047976A JP 2107520 A JP2107520 A JP 2107520A JP 10752090 A JP10752090 A JP 10752090A JP H047976 A JPH047976 A JP H047976A
Authority
JP
Japan
Prior art keywords
multiplication
code
sign
butterfly
rpi
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.)
Pending
Application number
JP2107520A
Other languages
Japanese (ja)
Inventor
Toshio Shirasawa
寿夫 白沢
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ricoh Co Ltd
Original Assignee
Ricoh Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ricoh Co Ltd filed Critical Ricoh Co Ltd
Priority to JP2107520A priority Critical patent/JPH047976A/en
Publication of JPH047976A publication Critical patent/JPH047976A/en
Pending legal-status Critical Current

Links

Landscapes

  • Compression, Expansion, Code Conversion, And Decoders (AREA)

Abstract

PURPOSE:To unnecessitate the code bit of a multiplying coefficient and to decrease one bit from the number of bits inputted to a multiplier by converting all the multiplying coefficients in a butterfly arithmetic circuit so that all the coefficients can be positive values. CONSTITUTION:In the range of 0<r<1/4, namely, in the angle range of 0-pi/4 radian, a multiplying coefficient -cos(rpi)+sin(rpi) is a negative value. While paying attention onto this fact, a code is inverted in advance to cos(rpi)-sin(rpi). In place of inverting the code of this multiplying coefficient, -1 is multiplied and added to the multiplied result, namely, subtraction is executed in the next addition processing. Thus, an arithmetic result to be finally obtained is same as that of the butterfly arithmetic circuit of Wang. Therefore, a code processing in multiplication is unnecessitated and the code bit is also unnecessitated.

Description

【発明の詳細な説明】 (産業上の利用分野〕 本発明は、自然画像圧縮処理などで用いられる高速コサ
イン変換(FCT)のためのバタフライ演算回路に関す
る。
DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a butterfly calculation circuit for fast cosine transform (FCT) used in natural image compression processing and the like.

〔従来の技術〕[Conventional technology]

1次元高速コサイン変換の計算アルゴリズムとして、C
hen、、WangXMakhou 1などの方法が提
案されている。−例として、第2図(a)(b)に、C
henの8点FCTとその逆変換(IFCT)のアルゴ
リズムを示す。図中、黒丸は加算を示す。白丸はターミ
ナルであって、何ら演算を行っていない。各線の傍らに
記入されている値は乗算係数であって、その綿に入って
きたデータに当該乗算係数を掛けること示している。c
(rπ)はcos (rπ)の略、S(γπ)は5in
(γπ)の略である。また、−は−1の略であって、そ
の線に入ってきたデータの士符号を反転することを表し
ている。このChenのアルゴリズムによるときは、8
点FCT (IFCT)を16回の乗算と26回の加減
算で高速処理することができる。
As a calculation algorithm for one-dimensional fast cosine transform, C
hen, Wang X Makhou 1 and other methods have been proposed. - As an example, in Figures 2(a) and (b), C
hen's 8-point FCT and its inverse transform (IFCT) algorithm are shown. In the figure, black circles indicate addition. The white circles are terminals and do not perform any calculations. The value written next to each line is a multiplication coefficient, and indicates that the incoming data is multiplied by the multiplication coefficient. c.
(rπ) is an abbreviation for cos (rπ), S(γπ) is 5in
(γπ). Further, - is an abbreviation for -1, and represents that the sign of the data entering that line is inverted. When using this Chen's algorithm, 8
Point FCT (IFCT) can be processed at high speed with 16 multiplications and 26 additions and subtractions.

さらに、コサイン変換の基本となる第3図(a)のバタ
フライ演算回路をWangの提案に従って第3図(bl
のような回路(以下、「W a n gのバタフライ演
算回路」という)に変形すると、図(alの回路では4
回の乗算と2回の加減算を要していたのが、3回の乗算
と3回の加減算で済むようになり、前記8点FCT (
I FCT)を13回の乗算と29回の加減算で実現で
き、計算に時間のかかる乗算を減らして更に高速化する
ことができる。なお、第3図(a) (bl中の口はタ
ーミナルであって、何ら演算を行っていない。
Furthermore, the butterfly arithmetic circuit shown in FIG. 3(a), which is the basis of cosine transformation, was modified as shown in FIG.
When transformed into a circuit like this (hereinafter referred to as "W a n g butterfly calculation circuit"), the circuit shown in the figure (al.
The previously mentioned 8-point FCT (
IFCT) can be realized with 13 multiplications and 29 additions/subtractions, and the calculation speed can be further increased by reducing the time-consuming multiplications. Note that the port in bl in FIG. 3(a) is a terminal and does not perform any calculations.

〔発明が解決しようとする課題〕[Problem to be solved by the invention]

コサイン変換の基本となるバタフライ演算回路を第3図
中)のWangのバタフライ演算回路のように変形する
ことにより、4回の乗算と2回の加減算を要していたも
のが、3回の乗算と3回の加減算で済む。この3回の乗
算係数は、第3図(blから明らかなように、 ■  cos (rπ)+5in(Tπ)■ −cos
 (γπ)+5in(γπ)■  5in(γπ) の3つである。ここで、γは1/4.1/8.1/16
.5/16などの0から172の範囲、すなわち角度値
(γπ)でいうと0〜π/2[ラジアン]の範囲の値で
ある。
By modifying the butterfly arithmetic circuit, which is the basis of cosine transformation, like the Wang butterfly arithmetic circuit (in Figure 3), what used to require four multiplications and two additions/subtractions can be reduced to three multiplications. It only takes three additions and subtractions. The multiplication coefficient for these three times is, as is clear from Figure 3 (bl), ■ cos (rπ) + 5in (Tπ) ■ -cos
(γπ)+5in(γπ)■5in(γπ). Here, γ is 1/4.1/8.1/16
.. It is a value in the range of 0 to 172 such as 5/16, that is, in terms of angle value (γπ), it is a value in the range of 0 to π/2 [radian].

ところで、前記3つの乗算係数のうち、■の乗算係数−
cos (rπ) +5in(γπ)はO<r<1/4
の範囲、すなわち角度値で0〜π/4[ラジアン]の範
囲では負の値になる。したがって、乗算係数は正負の符
号付きの値となるため、第3図(b)の回路をハードウ
ェア化するには、乗算係数に対して符号付きの演算機能
を持つような乗算器を用いる必要があり、回路規模がそ
れだけ大きくなってしまう。また、符号ピントに1ピン
ト必要なので、乗算の入力ビット数もそれだけ増えてし
まうという問題があった。
By the way, among the three multiplication coefficients mentioned above, the multiplication coefficient -
cos (rπ) +5in(γπ) is O<r<1/4
It becomes a negative value in the range of , that is, the angle value range of 0 to π/4 [radian]. Therefore, since the multiplication coefficient is a positive or negative signed value, in order to implement the circuit in Figure 3(b) in hardware, it is necessary to use a multiplier that has a signed arithmetic function for the multiplication coefficient. , and the circuit scale increases accordingly. Furthermore, since one pin is required for the sign pin, there is a problem in that the number of input bits for multiplication also increases accordingly.

本発明は、上記問題を解決するためになされたもので、
乗算係数をすべて正の値とすることにより乗算器をでき
るだけ簡単に構成できるようにしたバタフライ演算回路
を提供することを目的とする。
The present invention was made to solve the above problems, and
It is an object of the present invention to provide a butterfly arithmetic circuit in which a multiplier can be constructed as simply as possible by setting all multiplication coefficients to positive values.

〔課題を解決するための手段〕[Means to solve the problem]

本発明は、前記目的を達成するために、直交変換の基底
関数として余弦関数を用い、2個のデータに対して3回
の乗算と3回の加減算で演算結果を得るようにした高速
コサイン変換のためのWangのバタフライ演算回路に
おいて、0〜π/4[ラジアン]の角度範囲において係
数値が負となる乗算係数については予め符号反転してお
くとともに、該符号反転した乗算係数の次段の加算を減
算に置き替えたものである。
In order to achieve the above object, the present invention uses a cosine function as a basis function for orthogonal transformation, and provides a high-speed cosine transformation in which a calculation result is obtained by three multiplications and three additions/subtractions for two data. In Wang's butterfly arithmetic circuit for , the sign of the multiplication coefficient whose coefficient value is negative in the angle range of 0 to π/4 [radian] is inverted in advance, and the sign of the next stage of the multiplication coefficient with the sign inverted is It replaces addition with subtraction.

〔作 用〕[For production]

本発明は、O〜π/4[ラジアンコの角度範囲では、前
述した■の乗算係数−cos (Tπ) +5in(r
π)が負の値になることに着目したもので、本来は負の
値となる乗算係数を予め正の値に変換して乗算処理を行
い、次段の加算処理において加算に替えて減算するよう
にしたものである。減算処理は符号を反転して加算する
ことに等しいから、バタフライ演算回路から最終的に得
られる演算結果は、第3図(blの従来のWangのバ
タフライ演算回路と同じ値となるが、バタフライ演算回
路の内部においてはすべての乗算係数が正の値により処
理される。したがって、乗算係数に対して符号ピントが
不要となり、また、乗算器の入力側に対する正負の符号
処理が不要となる。
In the present invention, in the angular range of O to π/4 [radianco, the multiplication coefficient of the above-mentioned ■ - cos (Tπ) + 5 in (r
This method focuses on the fact that π) becomes a negative value, and the multiplication coefficient, which is originally a negative value, is converted into a positive value in advance and multiplication processing is performed, and in the next stage of addition processing, subtraction is performed instead of addition. This is how it was done. Since the subtraction process is equivalent to addition with the sign reversed, the final calculation result obtained from the butterfly calculation circuit is the same value as the conventional Wang butterfly calculation circuit shown in Figure 3 (bl), but the butterfly calculation Inside the circuit, all multiplication coefficients are processed as positive values.Therefore, there is no need for sign focusing on the multiplication coefficients, and there is no need for positive/negative sign processing on the input side of the multiplier.

〔実施例〕〔Example〕

本発明の1実施例を第1図に示す。 One embodiment of the invention is shown in FIG.

第1図(a)は、Tの値が0 < r <1/4(角度
値7πで0〜π/4[ラジアン])の場合のバタフライ
演算回路、また、第1図(b)は、γの値が1/4≦γ
〈1/2(角度値TπでO〜π74[ラジアン])の場
合のバタフライ演算回路の構成である。
FIG. 1(a) shows a butterfly operation circuit when the value of T is 0 < r < 1/4 (0 to π/4 [radians] with an angle value of 7π), and FIG. 1(b) shows the The value of γ is 1/4≦γ
This is the configuration of the butterfly calculation circuit in the case of <1/2 (O to π74 [radians] in angle value Tπ).

本発明は、第1図(a)に示すように、TがO<r〈1
/4の範囲、すなわち0〜π/4[ラジアン]の角度範
囲では、前述した■の乗算係数−cos(rπ)+5i
n(rπ)が負の値となることに着目し、この負の値と
なる乗算係数については、予め符号を反転してcos 
(γπ)−sin(γπ)としておく。そして、この乗
算係数を符号反転した替わりに、次段の加算処理におい
て乗算結果に−1を掛けて加算、すなわち減算するよう
にしている。
In the present invention, as shown in FIG. 1(a), T is O<r<1
/4 range, that is, in the angular range from 0 to π/4 [radians], the multiplication coefficient of the above-mentioned ■ cos(rπ) + 5i
Focusing on the fact that n(rπ) is a negative value, the sign of the multiplication coefficient that is a negative value is reversed in advance and cos
(γπ)−sin(γπ). Then, instead of inverting the sign of this multiplication coefficient, the multiplication result is multiplied by -1 and added, that is, subtracted, in the next stage of addition processing.

このように変形すると、符号反転した乗算係数の乗算結
果は本来の乗算値と符号が逆になるが、次段の加算処理
において〜1を掛けて加算、すなわち減算されるので、
最終的に得られる演算結果は、第3図(b)のWang
のバタフライ演算回路と同一の値となる。したがって、
乗算における符号処理が不要になるとともに、符号ビッ
トも不要になる。
When transformed in this way, the multiplication result of the sign-inverted multiplication coefficient will have the opposite sign to the original multiplication value, but in the next stage of addition processing, it will be multiplied by ~1 and added, that is, subtracted, so
The finally obtained calculation result is the Wang
The value is the same as that of the butterfly arithmetic circuit. therefore,
Sign processing in multiplication is not required, and the sign bit is also not required.

前記以外の1/4≦r<1/2の範囲、すなわちπ/4
〜π/2[ラジアン10角度範囲では、■の乗算係数−
cos (γπ) +5in(rπ)は正の値となる。
A range of 1/4≦r<1/2 other than the above, that is, π/4
~π/2 [In the radian 10 angle range, the multiplication factor of ■ -
cos (γπ) +5in(rπ) is a positive value.

したがって、角度値がこの範囲の場合には、第1図(b
)に示すように、符号反転することなく、従来のW a
 n gのバタフライ演算回路(第3図(b))と同一
のままとする。
Therefore, when the angle value is in this range,
), the conventional W a
ng remains the same as the butterfly arithmetic circuit (FIG. 3(b)).

〔発明の効果〕〔Effect of the invention〕

以上述べたところから明らかなように、本発明によると
きは、バタフライ演算回路内のすべての乗算係数が正の
値となるように変換されるので、乗算係数に対して符号
ビ・ノドが不要となり、乗算器の入力ビツト数を1ビッ
ト減らすことができる。
As is clear from the above description, according to the present invention, all the multiplication coefficients in the butterfly calculation circuit are converted to positive values, so there is no need for sign bits for the multiplication coefficients. , the number of input bits of the multiplier can be reduced by one bit.

しかも、乗算器の入力側に対して正負の符号処理も必要
としないため、乗算器の構成が簡単となり、高速コサイ
ン変換における演算速度の向上とノ\−ドウェア量の低
減を図ることができる。
Moreover, since no positive or negative sign processing is required on the input side of the multiplier, the structure of the multiplier is simplified, and it is possible to improve the calculation speed and reduce the amount of hardware in high-speed cosine transformation.

【図面の簡単な説明】[Brief explanation of drawings]

第1図は本発明の実施例を示す図、 第2図はChenの8点FCTと8点IFCTのアルゴ
リズムを示す図、 第3図はW a n gのバタフライ演算回路を示す図
である。 特許出願人     株式会社 リ コ(bl 0くγく1/4偽工[蓚 fo) 1−発明色亥克伊1 第1図 (b) Wang めハクフラづ慣輩−回狗ト 第3図
FIG. 1 is a diagram showing an embodiment of the present invention, FIG. 2 is a diagram showing Chen's 8-point FCT and 8-point IFCT algorithms, and FIG. 3 is a diagram showing W a n g's butterfly calculation circuit. Patent applicant Riko Co., Ltd. (bl 0kuγku1/4 fake factory [蓚FO) 1-Invention Color Yi Katsui 1 Figure 1 (b) Wang Mehakufrazu Conventional Figure 3

Claims (1)

【特許請求の範囲】[Claims] 直交変換の基底関数として余弦関数を用い、2個のデー
タに対して3回の乗算と3回の加減算で演算結果を得る
ようにした高速コサイン変換のためのWangのバタフ
ライ演算回路において、0〜π/4[ラジアン]の角度
範囲において係数値が負となる乗算係数については予め
符号反転しておくとともに、該符号反転した乗算係数の
次段の加算を減算に置き替えたことを特徴とするバタフ
ライ演算回路。
Wang's butterfly calculation circuit for high-speed cosine transformation uses a cosine function as a basis function for orthogonal transformation, and obtains the calculation result by three multiplications and three additions/subtractions for two data. The multiplication coefficient whose coefficient value is negative in the angular range of π/4 [radian] is sign-inverted in advance, and addition in the next stage of the sign-inverted multiplication coefficient is replaced with subtraction. Butterfly calculation circuit.
JP2107520A 1990-04-25 1990-04-25 Butterfly arithmetic circuit Pending JPH047976A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2107520A JPH047976A (en) 1990-04-25 1990-04-25 Butterfly arithmetic circuit

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2107520A JPH047976A (en) 1990-04-25 1990-04-25 Butterfly arithmetic circuit

Publications (1)

Publication Number Publication Date
JPH047976A true JPH047976A (en) 1992-01-13

Family

ID=14461282

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2107520A Pending JPH047976A (en) 1990-04-25 1990-04-25 Butterfly arithmetic circuit

Country Status (1)

Country Link
JP (1) JPH047976A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO1998041026A1 (en) * 1997-03-12 1998-09-17 Matsushita Electric Industrial Co., Ltd. Encoding method, encoder and recording medium, and decoding method, decoder and recording medium

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO1998041026A1 (en) * 1997-03-12 1998-09-17 Matsushita Electric Industrial Co., Ltd. Encoding method, encoder and recording medium, and decoding method, decoder and recording medium
US6744928B1 (en) 1997-03-12 2004-06-01 Matsushita Electric Industrial Co., Ltd. Variable-length decoding method and apparatus employing code length and a recording medium
US7050644B2 (en) 1997-03-12 2006-05-23 Matsushita Electric Industrial Co., Ltd. Coding method and apparatus for input image data and a recording medium performing such method

Similar Documents

Publication Publication Date Title
Wang et al. A fast multi-scale retinex algorithm for color image enhancement
CN109785854A (en) The sound enhancement method that a kind of empirical mode decomposition and wavelet threshold denoising combine
Dubey Quaternion Fourier transform for colour images
AU2001223770A1 (en) Method of clipping signal amplitudes in a modulation system
Bovik et al. Basic linear filtering with application to image enhancement
JPH047976A (en) Butterfly arithmetic circuit
CN110190831B (en) Mixed norm non-negative adaptive filter
Dougherty The dual representation of gray-scale morphological filters
Sanz et al. Iterative time-limited signal restoration
Curto et al. Subnormality and k-hyponormality of Toeplitz operators: A brief survey and open questions
EP0466252A2 (en) A method and apparatus for restoring convolution degraded images and signals
Gao et al. High dynamic range image rendering with a luminance-chromaticity independent model
JPH0645948A (en) Orthogonal converter and reverse orthogonal converter
Cesini et al. An iterative method for restoring noisy images
Mardan et al. New fast Walsh–Hadamard–Hartley transform algorithm
US6308194B1 (en) Discrete cosine transform circuit and operation method thereof
JPH0645947A (en) Orthogonal converter and reverse orthogonal converter
JPS5966790A (en) Operating circuit
JPH0138697Y2 (en)
JPH05143634A (en) Correlation arithmetic unit
Choudhary et al. Analysis of Deep Learning Algorithms for Image Denoising
CN118151885A (en) Model conversion method, electronic device, and storage medium
US20030061252A1 (en) Non-constant reduced-complexity multiplication in signal processing transforms
Walvoord et al. Matched filter approximations for suppressing noise amplification
Neeti et al. Image enhancement based on log-gabor filter for noisy fingerprint image