JPS60178584A - Averaging means of product and sum of digital input and digital output - Google Patents

Abstract

PURPOSE:To perform product sum averaging operation easily and reduce errors in arithmetic by performing analog-digital conversion after digital-analog conversion and arithmetic processing, and using an operational amplifier for the arithmetic processing. CONSTITUTION:Digital input signals I1-In are converted by analog-digital converters 1-1-1-n and then inputted to coefficient setters 2-1-2-n, which multiply the inputs by coefficients alpha1-alphan respectively. Outputs of the coefficient setters 2-1-2-n are summed up by an adder 3 and further multiplied and divided through a coefficient setter 4 by a coefficient for averaging, and the result is converted by an analog-digital conversion part 5.

Description

[Detailed description of the invention] The present invention relates to a digital signal calculation method.

Digital operations have replaced analog operations in most fields due to the characteristics of digital signals. Calculators are widely used in place of slide rules for multiplication and division, partly because they have fewer effective digits. Recently, calculators that can perform 8-digit calculations (addition, subtraction, multiplication, and division) have become available at low cost.

In the industrial world, for example, image signals are converted to digital signals and signal processing is performed in T, V, facsimile machines, printers, scanners, etc., and many circuits that handle digital signals are used. ■ has been published. .

The image signal is subjected to analog-to-digital conversion in a scanner or the like, and various calculations are performed on the 8-hit digital signal.

As an example of this calculation, the present applicant has proposed
, Japanese Patent Application No. 58-14621 and filed on February 10, 1988 as ``Method for Processing Contour Enhanced Signals in Double-Mode Scanning Recording Apparatus.''

The calculation of digital signals in the above application involves arranging sequentially extracted image signals two-dimensionally, multiplying and adding arbitrary coefficients to the center image signal and peripheral image signals of the arranged image signals, and averaging the resulting signal (hereinafter referred to as product sum). A signal obtained by multiplying the peripheral image by an arbitrary coefficient is subtracted from the average signal (referred to as an average signal), and the signal obtained by the subtraction is used as a contour enhancement signal.

Since this product-sum-average calculation uses a large number of peripheral image signals, it is realized by configuring a signal processing circuit using a large number of multipliers or look-up tables and adders.

However, when trying to obtain a product-sum average signal without error using arbitrary coefficients and a large number of peripheral image signals, the number of digital bits used must be increased in order to reduce the calculation error, and the number of peripheral image signals to be handled must be increased. The number of multiplication circuits and addition circuits must be increased in proportion to the number of circuits.

The purpose of the present invention is to provide a method for easily performing product-sum averaging and reducing calculation errors by performing analog-to-digital conversion after digital-to-analog conversion and arithmetic processing, and using an operand m unit in the arithmetic processing. shall be.

The key point is to convert the digital input signal to digital to analog (hereinafter referred to as DA conversion), multiply the analog converted signal by a coefficient using an operational amplifier, add it, multiply and divide it by a coefficient for averaging, and then - By performing digital conversion (hereinafter referred to as AD conversion), a product-sum-averaged digital signal is obtained. A more detailed explanation will be given below based on the drawings.

A block diagram of an embodiment of a digital input digital output product-sum-averaging circuit for carrying out the present invention is shown in FIG.

In FIG. 1, l is a DA converter, 2 is a coefficient setter, 8 is an adder, 4 is a coefficient unit for averaging, and 6 is an AD converter. The symbol l is a digital input signal, a is a coefficient for weighting each digital signal (hereinafter referred to as a weighting coefficient), K is a sum-of-products average coefficient (hereinafter referred to as an average coefficient), and 0 is a digital output signal.

For example, if the input signal I is an 8-bit digital signal, θ
If output signal 0 is also an 8-bit digital signal, it takes a value of 0 to 256. The weighting coefficient α is given arbitrarily and is 0 to 2 for an 8-bit digital signal.
A value of 55 is preset. The average coefficient is determined when the weighting coefficient α is determined, and is calculated and set in advance.

Input signal I, output signal O and their respective coefficients a.

K is configured so that the following relational expression holds true, and is determined to be an average coefficient.

If all the weighting coefficients a are a6, the product-sum average coefficient XO becomes 1/na6 from the second equation. Therefore, all input signals ■ are 1. Then, the output signal Oo is K(1-n·αOIO, which becomes 1. by substituting K(1=1/na6), and the sum of products is averaged according to the first equation.

If the specific weighting coefficient α is α5 (as?0), the number of set coefficients aB is N, and the remaining weighting coefficients are zero, the average coefficient 8 becomes 1/N from the second equation.

Therefore, the input signal I (the input signal I that is paired with the m distortion coefficient α8)
) are all IB, then from the first equation the output signal 08 is 9
N (IB I9, KB = 1 /N aB
Substituting 1. Then, the weighting coefficient a13(αB-0)
The average of the sum of the pairs of (ticks) is calculated.

FIG. 2 is a circuit diagram of a specific embodiment of the present invention.

100 is a DA conversion circuit, 200 is a coefficient setting circuit, 800
is an arithmetic circuit for addition and averaging coefficients.

The DA conversion circuit 100 and the coefficient setting circuit 20G have similar configurations, the details of which are illustrated within the dotted line frame.

The DA conversion circuit 100 will be explained. This circuit is DA
It shows a circuit that is generally used as a conversion circuit, and is not limited to what is shown as long as it suits the purpose. 101 to 108 are resistors, each having a resistance value inversely proportional to the weight of each pit of the digital signal.

The contacts 111 to 118 of the analog switch are opened and closed according to each bit information of the digital input signal 11. switch 1
19 is when the signal I is zero (digital signal 1 is 8 bits, for example, and an 8-person NOR circuit (
This is used to adjust the drift wM of the operational amplifier (hereinafter referred to as operational amplifier) 181, and is also used for subsequent drift adjustment of the operational amplifier. Resistor (R2) 121 is weight resistance 1
The basic value 28 R1 of 01 to 108 is set to 255 "l. Input signal I.

The relationship between and the output V of the operational amplifier 181 is as follows. Vo is the Hanawa quasi-voltage. Since the maximum value of the digital signal 11 is 255, the eighth equation becomes v=-IITo/255=-(5).

Next, the coefficient setting circuit 200 will be explained. This is DA
The conversion circuit 100 converts the digital signal 1 to the reference voltage Vo.
．． Previously, the output voltage was proportional to the input voltage (
It operates according to the digital weighting coefficient α1 with respect to the output voltage () of the DA conversion circuit 100. output to Vl
Then, similar to equation 8, we get: Assuming that the iJf coefficient αl has a maximum value of 255, the Kasu formula 6 becomes Vl--α1-v/255-(8). fj4
Substituting the fifth equation into the S8 equation yields.

As explained above, by the circuit 100 and the circuit 200,
A DA conversion circuit and a coefficient multiplication circuit are formed, input signal I
, -In for the required number n of circuits 100 and 200,
3. If the coefficient α is zero, the contact point 211~
Similarly to the DA conversion circuit 100, the switch 218 is open and the switch 219 is closed.

Next, we will discuss the calculation circuit 800 for addition and average coefficients in the second section.
The explanation will be made assuming n-=9 with reference to FIG. 801-80
9 is a resistor R5 having the same resistance value.

The coefficient multiplication circuit 20 (Ll~20
Outputs 0-9 are added via resistors R5801-809. Each output of the coefficient multiplication circuits 2011 to 200-9 is added in proportion to the feedback resistance value of the operational amplifier 881 and the resistance value of the resistors 801 to 809). The average coefficient in the first equation has 8 coefficients and 2. is set by a combination of. If the input signal of the resistors 801 to 809 is Vj, then the average coefficient K is.

is directed to the contacts 821 to 828, and the output VA of the operational amplifier 881 is expressed as. α1. ~． αg=255. It, ~, l9
= 255, the ninth equation gives vl, = N'0, and if the calculation is made with K, = 255 and -C, then the tenth equation becomes.

Then, the output Vl of the operational amplifier 881 becomes R8''R7...04) Then, @18 formula becomes, Substituting the 12th formula into the 15th formula, and substituting the 9th formula into the 16th formula, α1= If Ii = 255, then iα1・■1−9α・I becomes 1=1 g Ml = 255− K1. becomes 8V (+ −%,
If Kl = K2 = KB = 255, vB =
V□ . . . Q groans, and this value vb (=Vo) is outputted as a digital value 256 as an output value O via an AD converter.

Explain with specific numbers. Referring to Figure 2, n
"9 + al + ""'+ "9= Il +
~+I9”Kl *K2+Kl=255,
From the 5th equation, the output V of the operational amplifier 181 is v--V□... (E) From the 9th equation or the 8th equation, the output V of the operational amplifier 281! is vt=-v=Vo=f1) From the 12th formula, the output vm of the operational amplifier 881 is 55 Vm=-□・9・VQ=-vo... (a) K1・9 The output vB of the operational amplifier 881 is By substituting the coefficient into Equation 16, VB=vO. Each operational amplifier 181°281.8
81.881 All outputs are Vo or -y.

It becomes.

Next, n ”” 4 + ”l+〜* α4 ”” Il
＋〜＋I4 ” 255(”5 e 〜+ αfJ
” 0 ), operational amplifier from formula 15 for formula 12!1
The output VA and VB of 81,881 are g=9X28
=252. Setting 1=4X28=112°Ks=255 gives 55 VA = -112,9・4 Vo = -1,0011
9047V.

Vl(vO).Therefore, the sum average of the four-man power side numbers has been achieved.

The average coefficient Kl + K2 may be 4.9, but due to the nature of analog computing units, Kl, Kll, and Kl are set to values close to 256, and the output value of the operational amplifier is also set to the reference voltage V.

It is preferable that the value be closer to , since the error will be smaller.

It goes without saying that α, Kl 9Kg 1Kll may be set to 9 bits or more.

Next, the coefficient “1 + αB + “7+”9”8 L
α2r ”4 *α6.aB=128.a6=265 will be described. n=9.II, ~, Ig =
If it is 255.

The output V of the A/'[) converter 100 is Vo,
Output vt, s, y of coefficient setter 200-1.8, 7.9
, e is 132 V l , 8.7 , 9 ''' 82 ・255 ' from the 9th formula
255'255VO"=255 V (] Similarly, the output V2.4 of the coefficient setter 200-11.4, 6.8.
6.8 is 28 VB from @9 formula, 4.6.8” 255 V.

Similarly, the output v6 of the coefficient setter 200-5 is 5 from equation 9.
5 vIs:■vO. From formula 12 and formula 15, operational amplifier 881°8
The output VA+VB of 81 is more than K1=100. 2=180. 8 = 179 weeks, Js = -o, light + Vc Ve = ■'c v
-τ；71゜apl'tliAVJ前'c Enter-h-L
t ci4 bed, f r (Z a yCl, 51J-・
, 1-"*cLv+)^4Please enter the upper side number zfb*;4IQtd111Aqn#drunk(post1;gporttv
>.

An application example using this invention will be explained. As mentioned above, the effect of this invention becomes more pronounced as the number of inputs increases.

One of the purposes of this invention is to obtain the product-sum average id number when the image information of the crowd h() is stored in a storage device such as a disk memory. As shown in FIG. For example 0
) When displaying on LT, all image signals are used to display the entire image, and even when obtaining a reduced image, the center image signal is obtained with a ratio according to the required magnification. For example, set the sampling ratio in the main scanning direction and sub-scanning direction to 8.
Pair 82nd is 2:81st is 8:8...For the 1st sub-scanning line, it is 2:21st is 8;29th is 2:2...
If we get the center 1w image number as .., then 2/28-I
If the image signal from the storage device is the original size (m), a 215 (40%) reduced image can be obtained. A 4x enlarged image will be obtained.It goes without saying that the image signal from the storage device may be an image signal from a scanner, etc.Even if the magnification in the main scanning direction and the sub-scanning direction is different. It's good.

Another application example using this invention will be explained.

By appropriately distributing the weighting coefficients, it is possible to obtain an optical filter effect similar to that using a layered mask. The filter effect signal is appropriately used for the center image and surrounding image signals.

In digital performances, addition is performed for each double sign, so if the number of signals to be added is n, (n-1) adders are required.In this invention, since the resistor takes the place of The practical arrangement of the parts is not large.

In digital calculations, errors due to rounding up or down of the smallest digit are added during multiplication, and a large number of cumulative errors occur.However, since analog values are handled, the error due to addition is around 8-bit calculations. It becomes considerably less.

Since multiplication uses a proportional amplifier, it is generally cheaper than a digital multiplier (1) including the A converter. If a lookup table were to be used instead of a multiplier, actual wiring would be required to provide an external signal to create the lookup table for the modified sum of the intensification coefficients, but this is not necessary in the present invention. I don't. If the look-up table is fixed, the result will be 71 (tct).

[Brief explanation of the drawing]

FIG. 1 is a block diagram of this invention, and FIG. 2 is a detailed one-way diagram of this invention. DESCRIPTION OF SYMBOLS 1...Digital analog converter, 2...Coefficient setter, 8...Latitudinal adder, 4...Coefficient setter, b...Analog digital tomoki device, 100...Digital converter, 200
...Coefficient setting circuit, 80()...Arithmetic circuit for hai l- and coefficient, 101~108.201~208,811~
1318.841-848, 861-868... Weight resistance, 111-119.211-219.821-82
8.851~858.871~878...Analog switch contact, 121, 221, 801~BOn...
・Resistance, 181.281.881.881... Operational amplifier, ■... Digital input signal, α... Digital weighting coefficient, K... Product-sum average coefficient, Kl + K2 r
K8: summation average combination coefficient, vc: reference voltage,
■...Operational amplifier output voltage, O...digital output signal. Figure 1 Figure 2 (A) and Figure 2 (B) Procedure 1lli city letter (spontaneous)] - Display of 11 cases Showa 5911 Patent Application No. 84018 Relationship with the uniform small f1 Patent applicant 11, 4-chome, Horiyoudori Terinouchi Umaru, Kamigyo-ku, Kyoto City Name (?l) 1 Dai Nippon Screen Manufacturing Co., Ltd., 1-1 Tenjin Kitamachi Representative 540 II Address 2-7 Nojinrin, Higashi-ku, Osaka Contents of the Wei Amendment [l] The scope of the patent claims will be amended as follows: (1) The product value of a plurality of digital input signals and a coefficient value arbitrarily given to each of the plurality of digital input signals as an analog signal. A digital input/digital output product-sum averaging method whose characteristics include adding the respective analog signals, performing an averaging operation, and converting the averaged value into a digital signal. (2) After converting a plurality of digital input signals into analog signals, each analog signal is multiplied by the above-mentioned arbitrary coefficient value and the product value of both is obtained as an analog signal. digital input digital input signal (I averaging method. (3) I converted the above arbitrary coefficient value into an analog signal
After that, each analog signal is multiplied by a plurality of digital input signals, and the product value of both is obtained as an analog signal. Method. (4) If the non-zero values among the coefficient values arbitrarily given to each of the plurality of digital input signals are the same, the digital output is the sum average of the digital input signals given the coefficient values other than zero. The digital input digital output product-sum averaging method according to claim 1, wherein the averaging calculation is performed so that the following is obtained. [2] On the 2nd line of page 2, there is a statement that says “...in the signal calculation method...
” is changed to “...one-word calculation method...”. [8] In the 4th line of the same line, e, ``I am broken....J'' is changed to ``I am broken...''. [4] On the 5th line, “...the number of significant digits is small...”
Toraru is assumed to be ``...has a large number of significant digits...''. [5] In the 6th and 7th lines of page 8, replace the phrase ``from... with a signal obtained by multiplying @marked peripheral pixels by an arbitrary coefficient...'' with ``・
. . . From the center image signal to the product-sum average signal...". [6] In the 8th line of the same line, change "...to the contour enhancement signal..." to "...to the contour enhancement signal...". [7] In the 20th line of the same line, change the phrase ``...and perform the calculation process'' to ``...and perform the calculation process (multiplication, sum, division)''. [8] On page 6, line 17, replace “Route 1” with “Route 1”.
Let's give an example.

[9] In the 18th line, replace "...things and limitations..." with "...
Limited to items/J. Let's say it says on page 7, line 12 of CIU. mark) ta page 20th line r VA= -0,994Vo
Let l be rV, = -0,994VoJ. ['121 The handwritten text from line 20 to line 3 on page 14 has been corrected to type as shown below. r·v,=-o99ivo vB=vo. In Fig. 2, input ■1 and weighting coefficient α1 (■1 and α2
・・１． and α. ), the same conclusion can be obtained as is clear from equation 9. J (14) On page 15, line 10, it says, “…set it to zero as appropriate…”
・" is changed to "...Tin (zundam or regularly) zero K...". [15゛] Page 16, lines 10 to 11: ``...If it is fixed, it is not possible to take an arbitrary weighting coefficient.
``...'' If we think of it as a fixed weighting factor, we would have to have many lookup tables and switch between them in order to obtain an arbitrary weighting factor...''. JI2 diagram (B)

Claims (2)

[Claims] (1)? Digital-to-analog conversion is performed for each of the M number of digital input signals, each of the plurality of analog signals is multiplied by a plurality of coefficient values arbitrarily given separately, and the respective multiplied signals are added by addition or proportional addition using an operational amplifier, A method for calculating the sum-of-products average f of digital human power output, characterized in that the signal obtained by proportional calculation of the added signal is converted into a digital output signal. (2) Make one or more of the arbitrarily given coefficient values the same as a non-zero value, set zero or one or more coefficient values as 岛, and output digital output (if the coefficient M is a non-zero value) 2. The digital input/digital output product-sum averaging method according to claim 1, wherein a proportional calculation is performed to obtain a sum average of t digital input signals given numerical values.