JPS6172474A - Shading correction system - Google Patents

Abstract

PURPOSE:To speed up a shading correction and to make it accurate by accessing the 2nd memory with the aid of the output of an A/D converter at the time of executing the optical scan of an original surface and shading information and obtaining a prescribed shading correction value. CONSTITUTION:Prior to writing shading information in an RAM5, a peak hold circuit 4 detects a previous peak level and adjusts gains of an amplifier 2 by an output (f) at the detected level. Thus the input level of an A/D converter 3 is made appropriate. Then a ROM6 is accessed with an original read-out data (c) and a shading information (d) read out of the RAM5, and a shading correction data (e) in accordance with the combination of the read-out data (c) and shading information (d) is appropriately read out according to the address.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a scanner that reads image information pixel by pixel by optically scanning a document surface, and particularly relates to a method for performing shading correction on the read image information.

In general, in scanners that read image information pixel by pixel by optically scanning the document surface, there is a problem with non-uniformity of light distribution characteristics in the optical system or with each element in the line image sensor consisting of COD etc. Since so-called shading occurs in which the output level of the image sensor fluctuates due to variations in sensitivity, etc., it is necessary to perform shading correction to keep the output level constant.

Conventionally, this type of shading correction requires a storage unit with a memory capacity compatible with the line image sensor,
A comparison section that compares the AD conversion output of the image sensor with the data stored in the storage section and outputs the larger signal of the two is provided, and the storage section is provided with a secondary signal based on the comparison result from the comparison section. The maximum output value for each image sensor element in the scanning direction is stored, and the correction signal is read out from the correction ROM using the accumulated data in the storage section as an address and given to the multiplier, where it is combined with the AD conversion output of the image sensor and a predetermined value. Shading correction is performed by performing a multiplication process of
(See Publication No. 9-27675).

However, in order to perform such shading correction, it is necessary to increase the resolution of the AD converter, that is, the number of bits, in order to increase the accuracy, and the precision of the output is determined by the number of input bits of the multiplier. It has become. In particular, the need for multipliers makes it difficult to perform high-speed processing.

The present invention has been made in consideration of the above points, and it is an object of the present invention to provide a shading correction method that allows high-speed shading correction to be performed with high precision by using simple means. .

Week 1! In order to achieve the object, the present invention includes a first memory in which shading information is stored, and a second memory that is accessed by an address corresponding to a combination of data read from the first memory and read data to be corrected. A second memory is provided with data corresponding to cases in which the read data to be corrected is corrected in association with the read data from the first memory, and the second memory The data read out is used as correction data.

Hereinafter, one embodiment of the present invention will be described in detail with reference to the accompanying drawings.

Figure 2 shows the output level characteristics in the main scanning direction of N CCDs in a line image sensor to explain shading. Due to optical variations, ■ Deterioration of the light source over time,
There are three factors: light source deterioration due to temperature characteristics, and (2) variation in COD sensitivity. In relation to Figure 2, A
Or, the B characteristic curve is due to the factor of (■) above,
The difference between the A characteristic and the B characteristic is due to the factor (■) above, and the enlarged unevenness of the curve is due to the factor (■) above. ,...is C
This shows the arrangement order of CDs.

FIG. 1 shows an example of a configuration for specifically implementing the shading correction method according to the present invention.

An amplifier 2 that amplifies the level of an output signal a read out in time series from a line image sensor 1 in which CCDs are arranged in a line in the main scanning direction, and converts the signal amplified to the required level into a digital signal C. An AD converter 3 that performs
A peak hold circuit 4 performs peak hold according to the AD conversion output, that is, original reading data C, and adjusts the gain of the amplifier 2 according to the peak hold signal f, and a shading information is written.
Predetermined shading information d according to read data C
The RAM 5 is read out and shading correction data is stored in advance.

Predetermined shading correction data e is read out according to the combination of read data C and shading information d.
It is composed of an OM 6 and a control section 7 that controls each section at predetermined timing.

With this configuration, if, for example, the reference white surface of the pressure plate is optically scanned prior to reading the original, shading information having characteristics as shown in A in FIG. 2 is obtained, and this information is stored in the RAM 5. Then, the sub-scanning progresses and reading of the document is started, but when the read data C at that time is corrected with the previous shading information, it is read out from the RAM 5 as a signal d and used. Further, at this time, the peak hold circuit 4 detects the peak level when the reference white surface was optically scanned before the operation of writing the shading information into the RAM 5, and holds that level. The gain of the amplifier 2 is adjusted to FlaI by the output signal f corresponding to the peak level at that time, and during shading, in order to correct in advance the variation due to the optical system described in (2) with rough accuracy, for example, when detecting the peak, as shown in Fig. 2 Even if it is at a level as shown in the medium B characteristic, after it is amplified, the gain is adjusted so that it becomes a level as shown in the A characteristic. As a result, the input level of the AD converter 3 is optimized, and conversion accuracy is improved. It should be noted that precise shading correction, including those due to variations in the optical system, is performed by the original shading correction according to the present invention. When the input level of the AD converter 3 is optimized in this way, the gain of the amplifier 2 is fixed to maintain that state, and from now on, writing shading information and reading image information of the original will be performed in that state. It is done. Next, when reading the original, the ROM 6 is accessed using the read data C and the shading information d read from the RAM 5, and shading correction data e is generated according to the combination of the read data C and the shading information d according to the address.
The sequence and timing control of the six or more parts from which the data are read out as appropriate, the control of sub-scanning, etc. are performed as appropriate in accordance with the signal group t output from the control unit 7. Note that the peak level may be detected not only in the AD conversion output but also in the output signal a of the line image sensor 1 or the output signal A of the amplifier 2.

The main points of the shading method according to the present invention can be listed in sequence as follows: 6 (1) After setting the gain of the amplifier that amplifies the image sensor output to the initial value, the reference white surface is optically scanned. A peak value corresponding to the obtained image sensor output signal is detected.

(2) Perform main scanning of the reference white surface at least once, and adjust the gain of the amplifier according to the obtained peak value.

(3) After that, the reference white surface is optically scanned at least once in the main scanning direction, and the shading information obtained at that time is written into the RAM.

(4) RA is synchronized with each main scan when reading the original.
The shading information is read from M, and the ROM is accessed in accordance with the combination of the read shading information and the read data of the document surface to obtain shading correction data.

FIG. 3 shows a specific configuration example of the amplifier 2 and the peak hold circuit 4. Here, a method is used to detect and hold the peak using a digital logic configuration, and as mentioned above, the gain adjustment of the amplifier 2 is performed with rough precision. Therefore, the read data of the upper 4 bits of the 6 bits of the AD conversion output is supplied to the peak hold circuit 4.

In the device configured as described above, the comparator 41 first compares the new input signal C with the peak value Cp held in the latch 43 until then, and outputs a select signal to the S terminal of the selector 42 according to the result. give. In response, the selector 42 selects either the new input signal C or the previous peak hold value CP and sends it to the latch 43. The selection condition at this time is c>cp
If so, signal C is selected, and if C≦cp, signal cp is selected. Thereby, the latch 43 updates and holds the new input signal C when it is larger than the previous peak value cp. The above operations are sequentially repeated for each main scan, and the peak values detected for each main scan are successively updated and held in the latch 43.

Next, the peak value detected for each main scan is transferred to the latch 44, and the latch output is given to the amplifier 2 as the gain w14 rectified signal f. Note that clocks t1, t3 and clear signals t2. t4 is given from the control section 7. Further, the gain adjustment signal f becomes each switching signal for the input resistance changeover switches (analog switches) SWt to SW4 of the amplifier 2, and the input resistance of the operational amplifier 21 is switched by the combination of on and off states of each of these switches SWI to SW4. let At this time, since the gain adjustment signal f is 4 bits, the input resistance value can be changed in 16 ways, and therefore the gain of the operational amplifier 21 is adjusted in 16 steps. Here, the 16-step gain is K/16. , to/15. /14, . . . K/3, K/2, K/l. At this time, the gain is /f (f is 16
level).

FIG. 4 shows an example of the configuration of the AD converter 3, in which the output signal of the amplifier 2 is converted into a 6-bit digital signal C. Imari reference voltages are expressed functionally as Vr+ and Vr-, ignoring their analog and digital properties.

e= (b − (Vr))/ ((Vr+)
(Vr-))me×64 (level). However, when C≦Vr-, c = l, C≧V
c=64 when r+ (or overflow signal generated)
It is.

Further, in FIG. 4, the analog switch SW5 is switched according to the control signal t5 from the control unit 7, and the reference voltage Vr- becomes 2 due to the control signal t5 during the writing period of the shading information to the RAM 5, and during other periods. Inside, a reference voltage Vr1 is connected to GND. This is done for the purpose of increasing the accuracy when writing shading information, as will be described later.

Further, FIG. 5 shows a specific configuration example of the RAM 5 and ROM 6 portions.

In this configuration, first, the shading information RAM5
When writing data to C, data C corresponds to the output signal for each pixel of the line image sensor 1 when the reference white surface is optically scanned.
is input to the Din terminal of the RAM main body 51. that time,
Addressing of the RAM main body 51 is performed by an address control signal t6 given from the control section 7 for each data input in pixel units. At the same time, a write signal t7 is applied to the W terminal of the RAM main body 51 in synchronization with this.

As a result, shading information for each pixel in the main scanning direction is written to the RAM main body 51.Next, each read data C for each pixel in the main scanning direction when reading the original is given to the ROM 6, and the pixel The address-control signal t is synchronized with the clock.
6, the RAM body 51 is accessed and the previously written shading information d is read out and temporarily held in the latch 52. The latch 52 has a clock t
8 is provided as necessary so that the RAM main body 51 and ROM 6 can be used even if the access time is long. Further, when the latch 52 is provided, the timing of access in the RAM main body 51 is controlled in order to synchronize the read data C and the shading information d.

Next, the read data C of the original and the shading information d read out from the RAM 5 are used to create an RO in which shading correction values corresponding to various combinations of these data are stored in advance.
M6 is addressed, and a predetermined shading correction signal e is read out from the ROM 6 accordingly. Here, each data C1d consists of 6 bits, and RO
In the range of addresses AO to All of M6, read data C is made to correspond to addresses AO to A5, and shading information d is
is made to correspond to addresses A6 to All.

In this embodiment, a line image sensor 1 in which 2000 CCDs are arranged in the main scanning direction is used, and the RAM
5 with a capacity of 2K x 13 bits, and RO
Each M6 with a capacity of 4 x 6 bits is used.

Next, the accuracy of the shading information d will be explained below with reference to FIG.

Figure 6 shows the density L/ for each pixel 1 to N in the main scanning direction.
/< indicates the characteristics of

When the gain r of amplifier 2 is set to the initial value, that is, all switches 5Wl-3W4 are turned off, the gain of amplifier 2 is /16, and when the reference white surface is optically scanned in that state. Let B be the read data. Assume that the peak value of the read data B is Bp and the minimum value is Bm, and for example, Bp=27 and Bm=18 are obtained.

That is, in this case, the read data B is 1 of 18 to 17.
It will be in the 0 level range. FIG. 7 shows a comparison table when levels expressed in decimal numbers are expressed as digital values of 6-bit 1-bit and 4-bit binary. From FIG. 7, if BP"27 is expressed as a 6-bit digital value, Bp=011010, and the gain adjustment signal f at that time is f=0 with the upper 4 bits of the peak value Bp.
110 (level 7), and therefore the gain of amplifier 2 at that time is adjusted to /7.

Next, when optical scanning for writing shading information was performed on the reference white surface, the amplified rf! 2 input signal a
The level is as shown in the B characteristic in FIG. 6, as in the case of optical scanning for gain adjustment. Strictly speaking, some difference will occur due to unevenness in the light reflectance of the reference white surface, noise in the electrical system, etc.

However, since the gain of amplifier 2 that amplifies the read data B of the reference white surface at that time is set to /7, the output signal of amplifier 2 at that time is BXK.
/7. Therefore, the amplifier output Bp' for Bp=27 is BP' =BPXK/7=27/7, and the amplified output Bm' for Bm=1.5 is Bm'==nmXK/7=18/7. Become. The actual value of K may be set to approximately 16. This can be achieved by appropriately selecting the input resistors RO to R4 and the feedback resistor Rf in the configuration shown in FIG.

Also, setting it to about 16 is because when the B characteristic in Figure 6 is corrected (or it can be considered as normalization) as the A characteristic by the gain adjustment, Bp' is the highest. 0 that is intended to be at level 64 or below 64 and close to it For example, the above-mentioned Bp', B
At each position of m', Bpl =27/7Φ61.7 Bm' ==18/7-41.1.

Therefore, by adjusting the gain of the amplifier 2 by the peak hold value Bp of the input signal B shown by the B characteristic in FIG. 6 when the gain of the amplifier 2 is at its initial value, the amplified output for the input signal shown by the B characteristic in the figure will be It will be corrected like the A characteristic.

Note that this is not limited to the case of FIG. 6; for example, in the case of Bp=40, Bp=100111, f=1001 (level 10)
Therefore, in this case, Bp' = BpxK/
f=64, and the gain of amplifier 2 is /f=
Adjusted to 16/10.

In this way, the gain of the amplifier 2 is adjusted according to the peak hold value Bp, and the amplified output Bp' is corrected to a level close to 64 at or below 64.

As shown in FIG. 6, the input signal having the B characteristic is corrected to have the A characteristic, thereby improving the level resolution. That is, in the case of B characteristic, B p -B m + 1 = 10 levels, and in the case of A characteristic, Bp' - Bm' +1 = 22 levels, and the resolution is improved from 10 levels to 22 levels. However, the resolution in this case is 6 in the AD converter 3.
We are thinking at a digital level of bits, or 64 levels.

Next, improvement in resolution when writing shading information will be described below.

As mentioned above, the reference voltages Vr+ and Vr- in the AD converter 3 are (Vr+)=:Vl, (Vr-)=
Set in V2. Vl is V2<B according to the relationship shown in Figure 6.
m' is an appropriate value that satisfies Bm', and here, when shading due to deterioration of the light source, variations in the optical system or CCD, etc. is within the specified range, it is determined from the minimum possible value of Bm' in characteristic A in Fig. 6. is chosen as a small value,
The relationship is V2=V1/2. That is, from the relationship of the y-axis in FIG. 6, V1=64 level, V
2=32 levels.

If the output level of the AD converter 3 with respect to Vl and V2 is illustrated, it will be as shown in the yl axis shown in FIG.
It is expressed at level 4. here.

y 1 = 2 y −64 (32< y≦64) yl
=1 (y≦32) yl=64 or overflow (/>64).

By establishing the relationship (Vr-)=V2=Vl/2 in this way, 32 levels on the y-axis correspond to 6 levels on the yt-axis.
4 levels, which means that the resolution has doubled. From another perspective, if the same 6 bits are expressed in binary, it will be 100,000 to 111 on the y-axis.
32 levels of 111, ooooooo ~ 11 on the yt axis
A characteristic will be expressed at the 64th level of 1111,
On the y-axis, only 5 out of 6 bits contribute to the resolution, whereas on the yt-axis, all 6 bits are used.

For example, in Figure 6, Bp' (y=61.7) is
If it is one coordinate, Bp' (yl) = 2X61.7-64 = 59.4, and Bm' (y 1) = 2X41.1-64 = 18.
2, so Bp' (y 1) Bm' (y 1)+1
=P42.

That is, in the B characteristic shown in FIG. 6, Bp-8m+1=1
By correcting the 0 level to the A characteristic, the resolution is increased to Bp'-Bm'+1/2 levels, and further, the resolution is increased to 42 levels by AD conversion using the yl axis.

The yl axis shown in FIG.
This shows that the resolution of the AD converter 3 needs to be 128 levels, that is, 7 bits, in order to obtain a resolution equivalent to that of the yl axis when performing AD conversion with r- set to GND.

Furthermore, if an amplifier 2 that corrects the B characteristic to the A characteristic
In order to obtain the same resolution as the yl axis when writing shading information using the B characteristic without adjusting the gain of the Is required. In addition, when the number of bits is increased, a problem arises in that the SN ratio per ILsB deteriorates. According to the present invention, however, shading information can be taken in with high resolution without increasing the number of bits of the AD converter 3 as described above.

In this way, when writing shading information, the AD converter 3
By controlling the reference voltage Vr- at , the resolution can be improved. In that case, the relationship (Vr-) = V2 = Vl/2 is only when writing shading information, and when reading the original, Vr-
will be connected to GND.

Next, a process in which the ROM 6 is accessed using the read data C and shading information d when reading a document and predetermined shading correction data e is obtained will be described below.

In FIG. 5, the original reading data C consists of 6 bits, and this data is designated as Da. Also, the shading information d is 6 bits and corresponds to the yl axis in Fig. 6,
Let this data be Dyl.

Further, the level on the y2 axis corresponding to the data Dyl is assumed to be DV2. Furthermore, Dc=(dc5, dc4, -, d
co) (dc5=MsB), Dy 1=(dy5, dy
The correspondence between each bit of 4, . . . , dye) (dy5=MSB) and the address AO-All in the ROM 6 is as follows.

dco=AO d c 1 = A 1 dc5=A5 dyo=A7 dyl=A7 dy5=All This correspondence is hereinafter expressed as an address (Dyl, Dc).

Further, the shading correction data stored in the ROM 6 is assumed to be Da.

In the above, ROM6 at address (Dy l, Dc)
The data De in is stored in the ROM 6 as a result of calculation using the formula: De=(DcXI28/Dy2) (rank). Here, the data De has a 6-bit configuration. Also, the data DC given as an address is 6 bits, but
Dc used in the above calculation does not need to be 6 bits. Furthermore, r′)y2 and 12 in the calculation formula
There is no restriction on how many bits the number 8 can be expressed in. In other words, each element in the calculation formula and its calculation result are calculated with the required precision or number of bits, such as 8 bits + 8 bits = 16 bits, and the upper 6 bits of the result are used as data. It is stored in the ROM 6 as De. At this time, for the lower bits below the 7th bit, numbers above a certain number are rounded up, and numbers below a certain number are rounded down and discarded.

For example, if Bp' is expressed on the y2 axis in Figure 6, then y2=
By 27.

Bp' (y2)=2X61.7=123.4.

Let Np be the position in the main scanning direction that is now Bp', and the read data at the position Np when reading the original is, for example, De=
Consider the address of the ROM 6 and the read data De at that time when the level is 38.

First, the address (D71.Dc) of ROM6 is 111010Dc=3 in Dy1=59.4 (level)
8 (level)=100101, so the address titoioto.

101=(EC5), I. Here, (EC5)H is a hexadecimal expression.

That is, when Dc=38 in NP, address (EC5)H in ROM6 is accessed.

Contents D of ROM6 corresponding to that address (EC5)
Since e is De=DcXl 28/Dy2 and Dc=38 and DY2=123.4, De=38X128/123.4÷39(level)=l
It becomes OO110 (binary 6 bits).

In other words, data 100110 should be written in advance to address (EC5), 4 of ROM6).

As described above, according to the shading correction method according to the present invention, the resolution of shading information can be improved by setting the gain of the amplifier 2 so that the B characteristic becomes the A characteristic in the relationship shown in FIG. 6, and by introducing the yt axis. The resolution of AD conversion 3 (
Even if the number of bits is small, shading correction can be performed with high accuracy.

Further, according to the present invention, it is possible to obtain shading correction data Da having arbitrary density characteristics when implementing the above-described shading correction method.

This will be explained below.

Figures 8(a), (b) to (c) respectively show examples of various agricultural conversion characteristics (gamma characteristics), and show the correspondence of the output density level to the input density level when the original is read. There is.

Now let the function of the read data Dc of the manuscript be F (DC) +
As the contents of address (Dyl, Dc) in ROM6.

Consider De=F (Dc) x 128/Dy 2.

Here, if F (Dc) = Dc, the contents of De at that time are the ROM6 at the time of shading correction described above.
The data content is the same as that of , and the density characteristic in that case becomes linear as shown in FIG. 8(a). At that time, F
If (Da)=k1·DC (kl is a constant), kl gives the slope of the straight line in the characteristic shown in FIG. For example, if F (Dc) = k 2 ・Dc'' (k2 is a constant), it becomes possible to obtain the squared concentration characteristic shown in FIG. 8(c). That is, F (D
If De is calculated using the above equation for the function c) and the result is stored in the ROM 6, shading correction data De having arbitrary density characteristics can be obtained.

Now, a case will be described below in which a color image of a document is read.

FIG. 9 shows the arrangement of each CCD sensor section in the line image sensor 1', and each sensor section is arranged in the order of R, G, and B in the main scanning direction. Here, R is red, G
is a sensor section that is sensitive to green and B is sensitive to blue, respectively.
Each of these sensor sections is obtained, for example, by disposing each color filter on the light-receiving surface of a CCD. These R
The spectral sensitivity characteristics including color filters of the -G and B sensor sections are shown in FIG.

□1 However, the spectral sensitivity of each color is different on the scanner side, which reads the color image of the document, and even on the printer side, which records the color image based on the read data, the ink Each color has its own unique spectral characteristics. Therefore, from the viewpoint of improving color reproducibility during recording, R, G,
It is necessary to perform different gamma corrections on each color data separated into B. For example, the slope of the straight line in FIG. 8(a) is made different for each of R, G, and B.

Figure 11 shows a specific configuration example for processing each color data of R, G, and B.
61. ROM62 for G data, RO for B data
It consists of M63 and ternary counter 8. Further, the ternary counter 8 counts the clock [5] from the control section, and sends a chip select signal 3 to each ROM 61 to 63.
m, 3m+1.3m+2 (m=o, 1, 2, . . . ), and is reset by a clear signal tlo from the control section. Here, ROM61-6
3 and Dc, Dyl correspond to those in FIG.

With this configuration, if the arrangement of each sensor section in the main scanning direction of the line image sensor 1' is as shown in FIG. Original reading data DC
From the shading information Dyl, only the R data is read into the ROM 61, and in response to the chip select signal 3m+1, only the G data from the original reading data DC and the shading information Dyl is read into the ROM 62, and the chip In response to the select signal 3m+2, only B data from the original reading data Dc and shading information Dyl is stored in the ROM 63.
will be loaded respectively.

Therefore, if the ROMs 61 to 63 are pre-stored with data based on the calculation results according to the formula below, each having its own unique gamma correction characteristics, not only the shading correction but also the predetermined gamma correction can be applied to the processed data De.
(R) and De (G)-De (B) are obtained, respectively.

De=F (Dc,)Xk/Dy 1 (k is a constant)
Note that when reading a color image of a document by separating the colors, instead of arranging the sensors in the order of R, G, and B as described above, for example, a red filter, a green filter, and a blue filter are installed for each main scanning line. It is also possible to take a method of performing color separation by performing three scans through the filter, and in such a case, instead of using the ternary counter 8 in FIG.
For example, a decoder or the like may be provided that can sequentially select 3.

In the shading correction method according to the present invention, it is possible to perform high-speed shading correction with high precision using simple means, and gamma correction can also be performed at the same time without any separate processing. It has this excellent advantage.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a configuration example for concretely implementing the shading correction method according to the present invention, FIG. 2 is a characteristic diagram showing the shading pattern S in data read from a document by a line image sensor, and FIG. FIG. 3 is a circuit diagram showing a specific configuration example of an amplifier and a peak hold circuit in the same embodiment. FIG. 4 is a circuit diagram showing a specific configuration example of the AD converter in the same embodiment, FIG. 5 is a circuit diagram showing a specific configuration example of the RAM and ROM in the same embodiment, and FIG. 6 is a circuit diagram showing a specific configuration example of the AD converter in the same embodiment. A characteristic diagram showing the output level state of the AD conversion output in the example, Fig. 7 shows the level expressed in decimal notation in 6 bits and 4
Figures 8(a), 8(b), and 8(c) are characteristic diagrams showing examples of various density conversion characteristics (gamma characteristics), respectively. Figure 9 is a diagram showing the arrangement of each sensor part in a line image sensor, Figure 10 is a diagram showing the spectral sensitivity characteristics of each sensor part for R, G, and B, including color filters, and Figure 11 is a diagram showing the R, G, and B sensor parts. , B is a block diagram showing a specific configuration example for performing shading correction and gamma correction for each color data.

Claims (1)

[Claims] 1. Means for adjusting the gain of an amplifier for amplifying the image sensor output signal according to a peak level corresponding to the image sensor output signal when a reference surface is optically scanned prior to reading a document by a scanner; , means for writing the output signal of the AD converter when the reference surface is optically scanned with the gain of the amplifier being adjusted into the first memory as shading information, and the A when the document surface is optically scanned.
The second memory in which shading correction data is stored in advance is accessed using the output signal of the D converter and the shading information read from the first memory to obtain predetermined shading correction data. Shading correction method. 2. The shading correction method according to item 1 above, characterized in that the gain of the amplifier is adjusted so as to have a value close to either one of the upper and lower reference voltages in the AD converter.