JPS5862973A - Color discrimination system - Google Patents

Abstract

PURPOSE:To record a picture with high quality, by making color data of a specific color thick line, erasing other color data with this thickened data, and discriminating the color. CONSTITUTION:Black+blue data are read based on a picture data SR1, and blue and red data are read based on data S1, S2 as a result of analog operation of data SR1 and SB1. Thus, the setting of color density adjustment of red, black and blue can be done independently of each color. After attenuating the noise of binarized blue data S11 outputted from a bonarizing device 17 at a compressor 43, the pulse width is enlarged with a thick-line circuit 44 and the result is applied to an inverter 46. The blue data are surely erased from a binary data S31 and even if position shift takes place, a black data BK can accurately be picked up.

Description

[Detailed description of the invention]

Favorite @ relates to a color discrimination method that can accurately produce a plurality of color data by performing color discrimination 11Jt from the original type vi: scientifically based on the optical book. Recently, there have been many image forming apparatuses that digitally record information by converting the original information in bits by a photoelectric conversion element such as a CCD into an electrical signal (according to No. 11). By the way, in general, such as documents, among manuscripts, the most important and
It is thought that IlsIlK is high for three color information: black, red, and back. As methods for determining these three types of color information, the methods shown in FIGS. 111 and 111 have been developed. To explain the first step, the original slv party m2 is irradiated, and its reflection -ytI is irradiated into a beam xylitter 6 like a dichroic tler by a III image lens 5 through a reflecting mirror 3 and an infrared absorption filter 4t. do. In this Be A slitter 6, the long-wavelength red light LR is reflected, and the short-wavelength blue light LB is transmitted and spectrally separated. The photoelectric converters 7 and 8 each have a plurality of elements arranged in a line. Therefore, the red light is detected by the dark V photoelectric converter 7 and the blue light *q**t# photoelectric converter 8 and converted into electrical signals. Photoelectric converter 7
and the respective love data BR and 5 detected in 8
int are sequentially outputted in time series using clock pulses from a clock generation circuit (not shown) and supplied to the one-color discrimination circuit 9. The second reason is an example of the configuration of the color discrimination circuit 9, and the above-mentioned image data 8R and 8B are each amplified &! After the amplitude is amplified by Sll and 12, the clan 1 circuits 13 and 14: voltage followers 15' and 16v are passed through to the monk data 8R1 and 8BIK, and then binarized 1117 and 1 like comparators.
This 2 becomes binary data DSR and D8B by 8.
The value data 08m1 and Daml are supplied to the decoder 519 and separated into colors IsJ, red data R1, 9-1 and IIK.
, blue data theory, color data of own data W is output. The dataer 19 is an inverter that inverts the pulse signal as shown in the figure, and an AND gate 2 that performs #iI layer removal.
It consists of 4 to 27. Here, the actual values of the image data 8R1 and SBl supplied to the binarizers 17 and 18 are shown in the table below and in FIG.
jI* (as shown in b) (11th!) Using II!t4 in the above table, the threshold of 8R1 ill' 5
By setting the II value t' of 50 mV and 811 to be 640 mV and binarizing it, color discrimination is possible from k. In other words, the 211 output of 8B1 is converted to DLR, and the 21 of 8!11 is
If the unified output is t-Daml, what is the result of the decoder 19 shown in Figure 1? The temples can be classified as shown in Table 6 below. (Table 2) , , . While tρ=Q, the red/blue signal output ratio of 8R1 is 670
; Since it is 281 out of 380, it is less affected by output fluctuation factors such as noise, and can be read as black. There are times when I confuse blue with white. In conventional color discrimination, accurate color cannot be obtained if noise enters the weighted signal of the photoelectric transformer due to the difference in the spectral sensitivity of the converter. An example of a disadvantage is that it cannot be distinguished, such as a combination of black halftones, black works or shading, red color being added to the edge of false information, or binarization. level 1
At 11 tF, red t

[The reproduced image became #1 with blurred, black and blue tints. Therefore, in view of the above points, the applicant has proposed a color discrimination method as shown in @4 button. When this t'rtg beak, 31 to 34 are amplifiers, and Makoto. The other parts are m1 diagram and m2 in above 4.
It's the same as the picture, so Ql! Clear vf omitted. here. Assuming that the gains of the amplifiers 31-34 are in V order KG1 to G4, the subtraction 03 and the data st output from the winnowing terminal and the data st and 82
is given by the following equation. 81 Jin Gl-! 111-G3・all (lj82
=G2・8R1-04・8B1 (2)
Here, (13, (01 to G4t of formula 23 - incorrect selection as follows. Gl = '2.74.' 02 m-1,0-3 = 1
．． ()4771.91 This makes the actual 81.82
A for each color. Naag output values are shown in the table below and in Figure 5.
As expected, it's the best. (ig B table). I! Comparing Table 1 and Table 3 or Figures 3 and 6, in order to distinguish each color data V, the binarization process is performed using 81 and 82 rather than SR1 and 881. It is clear that 10,000 is much easier. Even if the level ratio is small for the 1 year old, it is 243! 1492=l
:6.4, so it is relatively unaffected by noise. Because it is red. Black, the color of the spine - Il! #IA! Iv can be done independently f, for example, because red is light and ζ - the red signal II!
When the color was changed by a degree, the discrimination of the other colors 5 was affected by #1, and a layer of discrimination was created. Furthermore, 1l actually has two photoelectric converters @ (COD) **
Because there is a discrepancy in the fireflies, lens bottle) 1111, etc., the meeting is off by that much! However, there is an interlayer in which an entertainment signal equivalent to ''ea pixels in the main scanning direction is output to the color signal of a negative color (for example, red) at the edge part of a certain color (for example, black). Also, in the S scanning direction, it is the same as 4. As a result, the thinning of the black edge (1hTo, black'') IaiI, omission, and 5゛-image will occur. Mechanical KWI4*v does. The optical path adjustment US range was reached and there was a slight drop. Me (for example, p' rate position) is in the boundary and L1 non-light El, jl17e & shi. . 1g6- shows the first embodiment of the color discriminating circuit of the color distribution apparatus according to the present invention, where 41 is 1111
A unit, 42 is a binarizer like a comparator, Kuzu is fine noise of binary data fill v Noise data for warming 812 times or aK pulse IIt Expanding thickening A-jl
,4. Oh □, it's up 4. ``t'i7b
* If we take the example shown in C(1) Sits and Section 4, the gain of the amplifiers 31 to 34 "01 to 04" is selected from the following KII. 0l-=02-3.0 ., G3-C2.38 Jiku 1 and 8
Since the value of 81 is the same as the wax in the above-mentioned Park-1 table, substituting the G17G4O value into the equation (11 and (4) of tb release, the analog output for each color sum of stsgg of 1!- The force value &- will be as shown in the table below and Figure 7. □ (Jllll 4 II) ′・
The current value of sl, that is, the slice level BL1' of d~: For example, -800mV Klt$: and binarized! 17t
C is set to 600 m'l/ and another binary conversion 918
As a result, the binarized red data 821'4til1 is output, and from this, separate busy slice levels BLl and 8mm2v are set, and the two castle data 811 and 812%'*') are output, so that Figure 7 is adjusted. In 813, λ of actual deda,
82] can contain only red data, 110,000, which is black data! Voltage 7 to separate fiIi
Data output from subordinate 15 8RI Y amplifier 4
Amplify with 1 and binarize with independent 211I conversion @42. That-ll! J, increase 11! 41 no xiao ost' analogy is 2
．． 0 and that. ! ! The actual location of my data 83 supplied to the digitization-42 is as shown in the table below based on the relationship between 113 MmGB and IRI. (Table S) For example, set these 83 principles as 1 oso Peng V. Binarize with 2-bit conversion @42. Therefore, the two pieces of data 831 that are output include information about black and evening. Next, the data inverted via the inverter 46v and the 211 data 831 described above are supplied to the AND gate 4BK and a logical calculation is performed to obtain actual information from the data 831 containing black and blue information. The data of the V and the gate are also black data BICVII. In this way, without using a common decoder, the (black + blue) data t's is calculated based on the K11i-41 data 8RIK, -1000Il image data 881 and 811f t'based on the analog-operated data 81 and 82. Te (blue), (
red) because it is configured to read data such as red,
In the dot and blue color images 1, it is possible to set the alignment, that is, the slice level, independently for each color (Fig. 7-1). Therefore, according to this embodiment, Each color per color *mm
GM is a very harmful bird and is suitable for use as an entry/exit device for color inkjet recording heads. ! l! As shown in the srs diagram, the binarizer 17 or the scum 4! IT ff-1fJ11vfEI11@43”
T:/After X reduction, this is the ttv thick Gan conversion circuit 4
By expanding the pulse width ° by 4 and supplying it to the impanitor 46, from the binary 2 data 1131! data t
'a* has a square and a sub-scanning compression-431. -4'1 Since it is used as a frame, the iI & i frequency is used as the noise a
In Figure S, the pressure in the assembly 60 is shown as the main scanning primary and secondary device +a.
*o2 Show a specific example, where = 4 7 ritsu l
1 ahead 1 -4+17
y a y 7 (FIF) 51-54t' cascaded, ryst, sm-1 fence Sho 58-62973 (4
) and the related Q output signal Q1. Q2 and Q3 will be supplied to the t'andgate group. Also, F'P51.52 and (
Q output 4111t91.92m to σQ4t' and gate 68, FF51, 53 and (to Q output signal Q1, 93!hiQ4t'7y V?-)57
1C, FF52*sm'tUma)QmRikirinv4'z
, Q3 and Q4V (and &-) 58 are supplied, and based on the output signals of these AND gates 55 to 58, a logical operation of the OR gate is performed. It is supplied from the undo 60j1f to the honorary input terminal 'F- of the ry61.
-8YNC1 is supplied, the count data of the counter CT1 becomes OK, and this color C'l
'1' is introduced into its clock trend CKK and counts the garden lock 'pulse xcP3t'. The output terminal Q-ρ of the power, power, and power CTl,
The ζ minute sag1-*ez゛ outputted from the motor is supplied to the t'nandogenito 63', and the output No. 41 T1 is supplied to the -1 human power terminal of the AND gate, and the inverter 6
Both outputs of AND gate ω and 64 which supply the signal (Tl, )− inverted by 6 to the universal power terminal of AND gate ω and 4! Based on ORGATE 66. Logic signal DMI obtained from Q output terminal of Fi'6B
Both 12V AND gates 61 and 64' supply the terminal to the other terminals. This main scanning pressure a! 143
main compressed data DCM from all
The operations 1 to 4 are performed by FFs 51 to 54 in the above configuration.
This is done by commonly supplying the clock pulse CP3t' to the counter 68 and the counter C'rlK. Figure w41 wLtQl shows the signal waveforms of each part in Figure 8. Blue data 81 is currently obtained from $@MkT.
1tl - Assume that the signal is as shown in Figure 9. In response to the clock pulse CP3 shown in m-, the clock number 1. -DMRI-Peng 2 and DCM i' become the signals shown in tcJ-hi in FIG. Here, the four ANDGATE groups ~ss:jl# and the ORGATE s9HA'[) majority decision 111E path t'*I,
There are a lot of data (more than 3 out of 4) that are at Akin's level, and t'1 is that. Is the data that you compared, overall, at the theoretical level? Enter the data of one χ size to be taken. In this way, the main scanning yearbook and
00M43 A K Yotsuya, 94 Eha! line roar 17
Along with compressing the data to 216 bits of 2 & bit w%. Devil's Majority Decision llK-,,7-riCk15. vi
P, Pressure at the 5th Figure 10' Kfi *6 The pressure---
3 of 7. Here, a specific example of the sub-scanning compressor 43B is the shift register 5RIK* of the main compressed data DCM & 215 pits. Its output signal 71'1216 bit shift register
. 5R1 IC, and its output is supplied to No. 11 72t' 216-bit shift register 8R3, and these cascaded 1
1g sny: 3 shift registers l1R1,,sR
2 and 811B primary output (1-!73t-%,
Ru. Earth pressure, compression data DCMt and ° gate 74, 7
5 and 76tC, blade type part 7 of shift register 8R1
1 to the AND gates 74, 75 and 77, the output signal 721' of the shift register 8R2, and the AND gate 74,
76 and 77: Shift register north 3 output 4
No. 1 73t-Antoge 5-Door 5. ．． 76 and 77-
Supply each. This is 64 AND gates 74~7
Based on the output signal of 7, an m-sum operation is performed by an OR gate 78, and the output logical destruction signal D8R1t' is supplied to the AND game 79, and also supplied to the shift register 8114K of gta gt, and the output signal 80T and It supplies the gate 79 as well as the 216-bit shift register BBSVC, and its output 411-1181t'andgoo)nvc supply "fba-'r゛:/Vgate?7
9 to 611! 1m horizontal output sen* No. 08121'
A 216-bit shift register 8116 (part of the first line MLI) is supplied and sampled compressed data 1112V is obtained as the output of the hermitage. A clock pulse CP3t- that controls the operation t's of the main column scan compression @am is applied to the tarlock terminal CK of the 16-bit counter CT2.
supply ``Ii-un'' C10 is the clock pulse C1
'3t - total, its 0 frequency divided signal QA * divided frequency signal Q
'B* and Tanebu QC are supplied to the AND gate edict, and the time element * of the horizontal output is obtained. Same J9! The clear terminal CLR of counter q1 and the clock terminal e of another 16-ugly counter CT3 are connected to the seed part BYNC2'.
supply to sc. Counter CTa is #9141 aY
N('! is the number of stitches, and then the divided pIIJ QAIX frequency division signal QB. The frequency division signal gQc and the frequency division signal QDt' data DE
CK is supplied, and its decoded output is the 4th second time signal T82.
v get. Furthermore, the clear terminal CI of the 16-way counter CT3 is outputted via the inverter 831'.
, RK is supplied, and this (No. 11 is collected as 3 o'clock wound number Tj1B.).
Commonly supplied to the clock terminal CK of 2 and SR3.
It rang. Antogame with gi time wound Tal and second time wound number τ82? 84, and its output No. 41 lA31 is supplied to the clock terminals of both shift registers SR4 and 81B. Supplied to K Zenidoshi. In addition, the first wound number T81 and @3
t1[[[m? a! I and V7y are supplied to gate 85. Its output gI signal lA32 shifts to the clock terminal CKK of register 886. 4.1. Here, -clock pulse CPS is a signal for main scanning m, and is a pulse signal generated by 1 bit @fc. In addition, there is 411 "1" for the 8YNe21t IIJ run, and it is a signal that causes the 1 counter CT31 to count up on the 1 line @. The clock pulse Xaps and the synchronization signal aYN (!2) that control the main scanning and sub-scanning V are
is the main compressed data ochivt, 54,5 bit □,
It is compressed into numerical data 812. four and gates 74
~77, consists of OR gate 78, and when the mm path U pigeon is compressed to 7, the co-condensed data ■ 3 shifts 5 registers 8
'R1, north 2 o-, am l goes to high logic level.・Also, the AND gate 79 is a circuit with a data voltage of 5tl-. Therefore, the argument f! One circuit LOG and an AND gate 79 perform the majority voting operation of the tree. In this way, the sub-scanning pressures SaW and BK are determined. Sub-compressed data by bulk eggs every 12 lines 81
21: Obtain a Iigll signal waveform of each part in the #410 sound (shown here, the period of the Rizonotsuku pulse CP - t' ``cp
Then, the period of the synchronizing signal 8YNC1 "8Yl is lLL"
i32 desk again. 391 signal BYNc2t) period v
'r8Y2 To-rb and the period T of the 148 signal T81
T81 is 8'rc, , 2nd time wound number TB2 period t
', -1ch'8Y2, 3rd time 4II 'rgsa7m
1st term TT & 3 will be 12 TIIY□. . Thickening circuit 44 in IIe- to porridge 12-. Here, 101 is binary value 1i1i+bond data 1112v1 line delayed j! Yangang Road, 1o2
Updated data 812 from Renden-Danroku 101.
1. Delay circuit that causes line sfl, 10:H-! The above-mentioned trench data 812, E112m and II 12
B t-calculate the sum of pulses and increase the number of pulses in the sub-scanning direction (
This delay circuit 101 and !70 are OR gates that output 0.
2 and the OR gate 103 perform widening in the sub-scanning direction. , 7;
(For example, 1 bit) The delay circuit is a Daft flip-flop or a delay circuit. 105 - its delay circuit 104. Output data 812B
¥, Ri. Predetermined bit (for example, 1 bit) 1 per C vine pulse CP
! For example, a delay circuit such as a D-wrinkle flip-flop t1 is used to delay the image. 106 is the above-mentioned jI image data g12c. 812D and a1i! Data 8 whose pulse width has been expanded (hereinafter referred to as thick line) in the main scanning direction by wv@ring summation
! It is an OR gate that outputs 3V, and this delay circuit a110
4*j,bi101o7game)'106! ')! :i'
Thickening in the scanning direction ・13-(At~(Force, thickening of the sub-scanning direction in the 12th factor WCWJ″r) Data 812 is the 1318th
(If the signal is as shown in Bj, the same atAJ [
According to the clock pulse CPK shown in FIG. Delay 1 depending on the width of the thick line
The number of hems 101 and 102 can be increased or decreased.゛14th '6! d(4) ~ field has rust lz&! that makes the line thicker in the main scanning direction. 4 showing the signal waveform of the valley in J
One of the valley pulses of the data 812C, which is made bold in the sub-scanning direction by 4#tlII from L'th Ohage-1-103, is 14W (assuming that it is colored by a signal similar to that shown in the town).
Same l[i! Depending on the clock pulse CP shown in I(2):
Each pulse of data B12D, 812E and 813 shown in C, l112- is @InIC1 (C1~1
The signal shown in m is thus obtained.In this way, 1g6a! From the thickening circuit axis indicated by lIK, the thickness becomes thicker in the sub-scanning direction and the main scanning direction. In other words, data 4813 expanded by a predetermined amount including the number of pulses and the pulse width is output. Then, the inverter 11131' inverts it and supplies it to the AND gate 48, and the AND gate mallet performs a logic operation with the data 831 containing (7 blue + points) to deactivate the blue data and black data BKt. Output. As a result, the gastric data to be deenergized is expanded by 1 ψI, and as a result, the blue day and evening can be deenergized to 11 tones, causing misalignment of the photoelectric converter screen and the like. Data lit'correct m-discriminate, output is t'0・
7"T"-11')Jl')"L%i Iekuro, ?#lAl
The conventional point of constant l1! ! 7 can be resolved at will 8. May:, 1i6WJK7ic, Ki-sho? A neutral input is made from the AND gate 48, and the bs data III is thickened by the circuit I.
By supplying the data to a thickening circuit 45 having a configuration similar to that shown in FIG. supply The AND gate 49 performs a logical AND operation on the inverted data s16 and the red data 82M supplied from the binarizer 18, cancels out the black data, and outputs the red data It'. Therefore, if some black information remains in the red data 821 output from the binarization 1118, the black data vm'Is can be turned off, and the positional shift of the photoelectric converter etc. can be avoided. Also, it is possible to take out the red data 8 by separating it into the correct VcII@.Therefore, the red data 8 can be taken out separately, so the red data 8 can be taken out separately.
The traditional problem of missing parts can be completely solved! ! ! 1 indicates a second detailed example of a color discrimination circuit II) using the present invention, where 110 is an amplification circuit 81 for amplifying the amplitude of I/all supplied from the giltage follower 16; 111)! The data supplied from the increase @ 9110 is converted into 84v binary value! ! It is -. In addition, 112 is an analog game that performs horizontal calculations on the data 1141 supplied from the 111I @111 and the data 816 supplied from the inverter 47, deactivates the black data, and outputs the red data. A self 7 Other parts are g
Since it is similar to the detailed example 111111 shown in sw, the sharp @ is omitted. If you make a motion different from the 111th m11m, the image puri 11B1i! The amplitude is increased at 111141111◇. The gain of the amplification is, for example, G6.
Divided by 2.0, 14 is G6/1111. The actual value of 84 is as shown in the table below (see table 111).
(No. 6 113 This 84 jllllllI [v example is 1270 a+V
t&tF L, te. Since it is binarized by the 2iI converter 111, the binarization is from Oboro. Binary data 841 supplied to the AND gate 112 includes red and black information. -Man, J11i data 1! If it becomes thicker! The pulse width and number are expanded by 1111g45, and the data 816v and goo) 112 is further inverted by inverter 47. Therefore, the black information is extracted from the binary data 1141 containing the red and black information described above.
! It is possible to output accurate red data 8 from the anime game 112. The actual data 8 output from the one color discrimination circuit is JI[I
When used as area specification for mm5i collection of II,
Since the solution gI degree is not required, it can be used without special deactivation of other colors, so the binarization time #! 17
The data all output from the controller is supplied as the back data B as is. The same reason is also used to supply the blue information to the thickening circuit a through the pressure SSO when the blue information is turned off. However, since the red data R requires a high scale device, pressure a is not applied when the black information disappears. of course. When using the blue data cell as data to be reproduced, such as characters or II diagrams, rather than specifying an area, it is necessary to deactivate the color of the cell (for example, black) as was done in the color discrimination of the red data R. is suitable. . FIG. 16 shows another specific example of the thick-lined circuit 44 or 45, where "C" and "12G" are delay circuits that delay the 2-bit data Sν or 8K by one line, and the other parts are #8.
The configuration is similar to that shown in FIG. 12. As shown in FIG.
The signal is then supplied to other delay circuits 101 and 102 and an OR gate 103. Therefore. Thick line data l113f output from the OR gate 106
Since the output data 813 or 815 can be delayed by one line as a whole with respect to the input data 1112 or BKK, the output data 813 or 815 can be matched with the write timing of an llll1 image recording device or the like. Note that the thickening circuits I and 6 are used to make the deactivation of the Wr constant color more effective and to enhance the effect, so even without this thick line data or 45, a constant deactivation young fruit can be obtained. It is clear that As stated above, according to the present invention 91, the color data of a specific color (for example, evening) is made thicker, and the sacrificial color data (for example, black + blue) is energized by this thickened data. Since color discrimination is performed by this, it is possible to use a regular saw to turn off a specific color. Therefore, the problem of black thinning or missing black thin black squares can be solved, and it is also possible to reliably remove red recordings from black edges. in this way,
According to the book addressed to f11, Hara SK is extremely loyal.11i 4
11 can be obtained, and by using this color discrimination output signal for recording, it becomes possible to produce a high-quality 111-ki sickle.

[Brief explanation of the drawing]

#! Figure 1 shows one image area using the conventional color discrimination method.
t's internal configuration diagram, Fig. 2, and color discrimination same as those in Fig. 1. Figure 3 is a block diagram of another conventional color discrimination circuit. Figure 5 is a block diagram of another conventional color discrimination circuit. Figure 6 is a block diagram of a color discrimination circuit which is a GL embodiment to which the present invention is applied, Figure #I7 is an analog diagram of Figure #I6, and Figure #I9 is a diagram showing Imm flaw signals of analog subtraction. (Q is the 11 prefectures that represent the signal waveforms of each part in Figure 8.
Figure O is a block diagram showing an example of the -j scan compressor part of the Wang Liuki in Figure 6, and Figure 11 (1) shows the signal waveforms of each part in Figure 10. sea% showing
FIG. 15 shows a block W of a color discrimination circuit according to a second embodiment of the present invention. The sixteenth factor is a block diagram showing another example of the thick-lined circuit in FIG. 6 or FIG. 5. 1...Hara Shizu, 2...Hikari I1.3...Reflection Prisoner, 4.
...Infrared absorption filter, 5...M4a lens, 6...
・Beam splitter, 7.8...Photoelectric transformation Bokusai, 9...
・Color discrimination circuit, 11, 12...v#1-device%13
, 14... Clamp circuit, 15, 16... Voltage follower, 17, 18... 2 Wart generator, 1
9...Decoder, 2+3-23...Inverter,
24-27...AND gate %31-34...Amplifier, 35.3p...Wave counter, 41...L ■Support device,
42...2 saponifier, 43.i-≠Jl device, Komu...
・Main scanning pressure return device, 43g・Sub-scanning pressure return device, 44.4)
... Thick line circuits 46, 47... Inverter, 4
8, 49...And Gate, 01-54 m 68
”・Free lag flop, 55-58.60.63°6
4, 74-77, 79, 82, 84, 85・
...and gate. 59, 66.78... or game←, 65, 83
...Inverter, 101, 102...Delay circuit, 103...OR gate. 104, 105...Flip-flop (slow f1 circuit), 106...Or game), 110...Amplifier, 111...2 division! #, 112...AND gate, 120...Delay circuit, SR, 8R1°D8
R, 81, 81], 8B, 8Bl, D8B, 82.8
2]...color @49. 83, S:31.812.
813.814.815.1116.84. 841...color signal, R...red data, BK...black data. B...back data, W...white data. Applicant: Canon Co., Ltd. Figure 3

Claims (1)

[Claims] The color-separated original light is converted into a plurality of electric signals by a photoelectric conversion means, a color signal of at least one color is obtained from the electric signals by a color discrimination means, and a characteristic of the color signal is A color characterized in that the width and number of pulses of a running color signal are expanded by a thickening means, and the color signals of colors other than a specific color are deactivated by the expanded color signal to perform color discrimination. Discrimination method.