JPH05219373A - Picture reader - Google Patents

Abstract

PURPOSE:To attain high resolution and high gradation of a picture by correcting output distortion of an output of a photoelectric conversion element array with respect to an output of an A/D converter. CONSTITUTION:A photoelectric conversion element array transfers sequentially an output charge of photoelectric conversion elements arranged in an array and outputs the charges in time series. A reflected light from an original 101 by a prescribed light source is reduced by a lens 102 and the image is formed to a CCD image sensor 103. The sensor 103 generates a charge in response to a received luminous quantity according to a drive signal of a timing control section 111 and outputs voltage signals in time series for each of even number picture element and an odd number picture element separately. The output is given to an analog processing section 104, in which amplification, level conversion and picture element synthesis are implemented and the result is converted into digital data by an A/D converter 105. Output distortion caused by a transfer efficiency of the photoelectric conversion element array is corrected with respect to the output of the digital data. Furthermore, the output sensitivity of the photoelectric conversion element array is corrected. Then a picture with high resolution and high gradation is obtained independently of the position on a picture.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image reading apparatus having a photoelectric conversion element array (CCD image sensor) such as a facsimile and a digital copying machine.

[0002]

2. Description of the Related Art A CCD image sensor guides an electric charge received by a Pn photodiode to an output section by successively transferring potential wells having a potential. When the charge is transferred to the well, 100% of the charge is not always transferred to the adjacent well, and a small amount of charge remains in the original well. The ratio of the charge transferred to the next well to the charge existing in the original well is called "transfer efficiency". This deterioration is due to the increase of crosstalk with the adjacent pixel, and the resolution (MTF) of the image signal obtained. To reduce the output distortion.

In particular, in recent years, digital copying machines have been improved in high speed, high image quality and high function, and the transfer efficiency is lowered by high speed driving of the CCD image sensor, multi-gradation of one pixel signal and deterioration due to high magnification image enlargement. This output distortion is becoming a problem because the visualization of images overlaps. When the transfer efficiency is 100%,
The image change from white to black is obtained as a complete change from the output value 100 to 0 as shown in FIG. 13, and if 4 times image enlargement is realized by linearly interpolating between pixels and inserting 3 pixels, FIG. As shown in.

[0004]

However, in the actual state where the transfer efficiency is deteriorated, the output value 10 as shown in FIG.
The output does not change completely from 0 to 0, and output due to crosstalk occurs. FIG. 16 is an image enlarged in the same manner as shown in FIG. 16, which requires 10 pixels or more to reach complete black. Regardless of the conventional device that binarizes this image signal to make it a visible image, in the future multi-gradation printer, this is observed as blurring. Also many C
The CD image sensor is provided with a plurality of transfer shift registers in order to reduce the number of transfer stages and increase the operation speed.
Since it is difficult to make the transfer efficiencies completely match in each shift register, for example, in a double channel CCD having two shift register groups, if transfer efficiency deterioration occurs in only one channel, it is indicated by ◯ as shown in FIG. Distortions appear alternately such that the pixel output has distortion and the pixel output indicated by Δ has no distortion. When the image is enlarged, it becomes as shown in FIG. 10, which is actually observed as a striped pattern. In addition, in FIG. 14, FIG. 16 and FIG. 18, a black circle indicates a pixel interpolated and inserted by image enlargement.

The present invention has been made based on such a background, and provides an image reading apparatus which can obtain high resolution and high gradation by correcting the image distortion caused by the above-mentioned transfer efficiency. With the goal.

[0006]

SUMMARY OF THE INVENTION The above object is to provide a photoelectric conversion element array for sequentially transferring output charges of photoelectric conversion elements arranged in an array and outputting them in time series, and an output of the photoelectric conversion element array as an analog signal. -Conversion means for digital conversion,
This is achieved by the first means including a correction means for correcting the output distortion caused by the transfer efficiency of the photoelectric conversion element array with respect to the output of the conversion means.

A further object of the invention is to photoelectrically convert the outputs of the photoelectric conversion elements arranged in an array in order and obtain the output in time series, and to perform A-D conversion on the output of the photoelectric conversion element array. The transfer efficiency of the photoelectric conversion element array is calculated based on the conversion unit and the photosensitive element output value of one pixel arranged a predetermined number of elements behind the last end of the effective pixel row of the photoelectric conversion element array. This is achieved by the second means including a setting means for setting and a correction means for correcting the output distortion caused by the transfer efficiency of the photoelectric conversion element array with respect to the output of the conversion means.

A further object of the invention is to photoelectrically transfer the outputs of photoelectric conversion elements arranged in an array and obtain the output in time series, and to perform A-D conversion on the output of the photoelectric conversion element array. Conversion means, first correction means for correcting the output distortion of the photoelectric conversion element array due to transfer efficiency with respect to the output of the conversion means, and the photoelectric conversion means for the output of the first correction means. A third correction means for correcting the data based on the dark output of the conversion element array;
It is achieved by the means of.

A further object of the invention is to photoelectrically transfer the outputs of photoelectric conversion elements arranged in an array and obtain the output in time series, and to perform A-D conversion on the output of the photoelectric conversion element array. Conversion means, first correction means for correcting the output distortion of the photoelectric conversion element array due to transfer efficiency with respect to the output of the conversion means, and the photoelectric conversion means for the output of the first correction means. This is achieved by the fourth means including the second correction means for correcting the output sensitivity variation of the conversion element array.

[0010]

In the first means, the output distortion due to the transfer efficiency of the photoelectric conversion element array is corrected for the output of the A / D conversion means.

In the second means, the transfer efficiency calculated by the output value of the photosensitive element for one pixel arranged a predetermined number of elements behind the last end of the effective pixel row of the photoelectric conversion element array is caused. The output distortion of the photoelectric conversion element array is corrected.

In the third means, the first correcting means corrects the output distortion of the photoelectric conversion element array due to the transfer efficiency with respect to the output of the A / D converting means, and the second correcting means. Then, the data of the output of the first correction unit is corrected based on the implied output of the photoelectric conversion element array.

In the fourth means, the first correction means corrects the output distortion of the photoelectric conversion element array due to the transfer efficiency with respect to the output of the A / D conversion means, and the second correction means. Then, the output sensitivity variation of the photoelectric conversion element array is corrected with respect to the output of the first correction means.

[0014]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows a model of the charge shift, which is D ₁ , D ₂ ,
D ₃ ... Is the charge generated by the photo diode and is the true value of each pixel output. In the first shift, D ₁ is output as Q _{1 to} the outside of the CCD image sensor and D ₂ is D
_{2 1} and D ₃ are respectively shifted to D _{3 1} . In the second shift, D _{2 1} is output as Q ₂ and D _{3 1} is output to D _{3 2} .
D _{4 1} is internally shifted to D _{4 2} . If the transfer efficiency is 1 (100%), Q ₁ is D ₁ and Q ₂ is D
Matches exactly ₂ . Here, when the transfer efficiency is ε, the first to third shifts are as shown in FIG.

Then, these D ₁ , D ₂ ... D _n are converted into Q ₁ , Q
Solving at ₂ ... Q _n, the correction equation shown in FIG. 6 (1) is obtained.

By the way, in order to read an A3 original with a pixel density of 400 DPI, a CCD linear sensor of about 5000 pixels is required, and most of these are double channels (there are two transfer shift registers for the purpose of reducing the number of shift stages and high speed operation). , Each of which is responsible for the shift of the even-numbered pixel columns and the shift of the odd-numbered pixel columns), and a shift of about 2500 steps is required for the final pixel output. Depending on the operating frequency,
Generally, the total transfer efficiency TTE is 0.9 (90%) or more, and in this case, the transfer efficiency ε shown per shift is 0.99995 or more (ε≈0.9 1/250 ⁰ ). Is shown. Here, in equation (1), Q _{n -3}
Assuming that the coefficient value of ε = 0.999995, n =
The maximum value is 2500, and [1 / (0.99995) ² 500] ² 49 ₉ C ₃
(1−0.99995) ³ ≈0.00037. If AD conversion is 8 bits, the maximum value is 255.
Then, at this time, the calculation result of the term of Qn- ₃ is 0.09, which can be completely ignored. Generally, the quantization error of the AD converter is ± 1/2 LSB, that is, ± 0.5, and if the term that is less than or equal to half of this value is ignored, the term of Q _{n -3} becomes [1 / ε ^{25 0 0} ] ・_{2 4 9 9} C ₃・ (1-ε) ³ × 2
From 55 ≦ 0.25, ε ≧ 0.9999317 ..., which can be ignored when ε ≧ 0.999932.

Regarding the term of Q _n -2, [1 / ε ² 500] ₂ 49 ₉ C ₂ (1-ε) ² × 2
From 55 ≦ 0.25, ε ≧ 0.999926 ..., which can be ignored when ε ≧ 0.999983.

From the above, the equation (1) is expressed as ε ≧ 0.99999.
In the case of 3, the equation (2) shown in FIG. 7 and ε ≧ 0.999
At 932, it can be approximated by the equation (3) shown in FIG. 7.

By such approximation or adaptive approximation by the value of ε, the scale of hardware can be reduced, and
It becomes possible to perform the operation without degrading the correction accuracy.

Effective pixels D _{1 to} D as in the model of FIG.
_When D _n pixels are arranged with the light shield pixels for 4 pixels separated from the last end of _{n − 5} , D is output to the output Q _n corresponding to D _n.
As mentioned above, there is almost no effect of _n-5 . Also, D
_{n - 4} to D _{n -} Since the _first value is 0, the n-th shift and (n + 1) th shift is as shown in FIG. 8, Q _n
And Q _{n +1} are as shown in FIG. 9, and the equation (4) shown in FIG. 10 is obtained.

From the above, the transfer efficiency ε is D _n in FIG.
Obtained by observing the output Q _n and the next output Q _{n + 1} corresponding to the (called TTE pixels).

Differences due to variations in the transfer efficiency of the CCD line sensor and variations in the phase and waveform of the driving clock among the devices are obtained by observing the TTE pixels in this way, and the value of the transfer efficiency peculiar to the device is obtained. Alternatively, it is possible to obtain an image output signal which can be completely absorbed by correcting with the approximate expressions (2) and (3) and which is not affected by the temporal change without variation between the apparatuses.

The CCD image sensor has a charge transfer function for the purpose of improving the total transfer efficiency (defined by Q _n / Q _n + Q _{n +1} from Q _n and Q _{n +1} described above) and enabling high speed operation. Some have multiple analog shift registers for use. For example, a device having two systems of shift registers called a double channel generally has odd pixel columns (D ₁ , D ₃ ,
D for ₅ ...) and to separate the transfer of the even pixel columns _{(D 2, D 4, D} 6 ...).

In this case, since crosstalk due to a decrease in transfer efficiency occurs between odd-numbered pixels and even-numbered pixels, it is necessary to perform the correction operation independently for both channels.

The transfer efficiency ε is obtained by observing the TTE pixel even after the image reading operation is started, and two parameters 1 / ε ⁿ and 1 / ε ⁿ (n-1) in the equation (2) are obtained.
(1-ε) is calculated from the first stage to the n-th stage according to the number of transfer stages and stored in a register or a memory by a microprocessor or the like, so that the correction calculation at the time of image reading operation can be performed. Can be realized by multiplying twice and adding once, and by implementing this as hardware, real-time correction operation can be realized at a speed synchronized with the pixel read frequency.

Normally, in digital copying machines, facsimiles, etc., correction by the output of the CCD image sensor in the dark (offset correction) and pixel sensitivity, light amount unevenness correction (shading correction) are performed, but the output distortion due to transfer efficiency is corrected. After the correction is performed and the true pixel output is obtained, the correction order of performing these correction operations is correct.

FIG. 3 is a block diagram according to one embodiment.

Reference numeral 101 denotes an original, which reflects light reflected by a light source (not shown) by a lens 102 to form an image on a CCD linear image sensor (CCD) 103. CCD 1
In accordance with the drive signal of the timing control unit 111, 03 generates charges according to the amount of received light, and outputs them in time series as voltage signals separately for even and odd pixels. Reference numeral 104 denotes an analog processing unit that amplifies the output of the CCD 103, converts the level,
Even / odd pixels are combined (mixed) and this is
The / D converter 105 converts the digital data. 113
Is a control unit, which includes a microprocessor, ROM, RAM, and I
It is composed of peripheral circuits such as / O and operates in the modes described below. Here, it is assumed that the CCD 103 is modeled in FIG.

The first operation mode control unit 113, prior to the actual image reading operation,
First, the light source is turned on, and a white reference plate or a blank document is imaged on the CCD 103. Switch S1, S3
Is turned on and S2 and S4 are turned off, the pixel count value in the main scanning direction from the timing control unit 111 is given to the memory 112 as an address, and the output of the ADC 105 is stored in the memory 112 as data.

Second operation mode When the ADC output for one main scan is stored, S1 and S3 are turned off and S2 and S4 are turned on to make the memory 112 readable / writable by the control section 113. Control unit 11
3 is an address 2N- where the odd TTE pixel output Q _{2 n} -1 of the CCD 103 and the next odd output Q _{2 n +1} are stored.
From the contents of 1 and 2N + 1, the transfer efficiency ε _{0 DD} of the odd-numbered column shift register is obtained by the equation (4) with ε _{0 DD} = 1-{(1 / n) · (2N + 1) / (2N-1)}. Note that () represents the value of the data stored at that address.

Similarly, even TTE pixel output Q _{2 n} and next even output Q _{2 n + 2} are stored at addresses 2N and 2N.
Based on the contents of +2, ε _EVEN = 1-{(1 / n) · (2N + 2) / (2N)} is obtained.

At this time, if the output of the CCD 103 has a value even in the dark (≠ 0, offset), Q
The output per _{2 n + 9, Q 2 n} + 1 0 as dark output value, the contents of the address 2N + 9,2N + 10 stored, as shown in FIG. 11, it is necessary to calculate.

Further, the first operation mode and the second operation mode are repeated a plurality of times, and from the average thereof, ε _{0 DD} , ε
_{Obtaining EVEN} will improve the accuracy even more.

Third operation mode S1 and S3 are turned off, S2 and S4 are turned on, and the second operation mode is set.
According to the values of ε _{0 DD} and ε _EVEN obtained in the operation mode of, the calculation parameter (1 / ε ⁿ ) (n-1) of the equation (2)
(1-ε) and 1 / ε ⁿ are calculated up to n = 1, 2, ..., N−5, and when the result is ε _{0 DD,} memory addresses 1, 3,
5, ... 2N-11, 2, 4 when using ε _EVEN
6 ... 2N-10 are stored respectively. The two parameters are set in the upper word and lower word of the same address, respectively.

Fourth operation mode S1 is turned on and S2 to S4 are turned off. Memory 11
2 is read-accessed with the pixel count value in the main scanning direction from the timing control unit 111 as an address, and the calculation parameters preset in the third operation mode are sequentially read. (1 / ε ⁿ ) (n-1) (1-ε) is supplied to the multiplier 107a, and 1 / ε ⁿ is supplied to the multiplier 107b.
An adder 108 calculates (output value of multiplier 107b)-(output value of multiplier 107a). Delay circuit 106
The delay circuit 106, the multipliers 107a and 107b, and the adder 108 operate as shown in FIG.
Then, the correction operation is performed.

The above is the correction operation only within the effective pixel range, and there is no problem even if correction is performed beyond this range. Reference numeral 109 denotes a dark-time output correction unit, which performs an operation of removing an offset when the output of the CCD 103 exists. Specifically, it is realized by storing the pixel output when the black reference plate is read or the pixel output when the light source lamp is off, and subtracting from the data after deterioration correction by transfer efficiency. A shading correction unit 110 corrects unevenness in illuminance distribution due to a light source, variations in pixel sensitivity, and the like based on output data when a white reference plate is read.

[0038]

According to the first and second aspects of the invention, since the image distortion caused by the transfer efficiency is corrected, a high resolution and high gradation image can be obtained without depending on the position on the image.

In the third aspect of the invention, the device recognizes its own transfer efficiency and performs the optimum correction operation for each device, so that a high-quality image with no variations can be obtained.

According to the fourth aspect of the present invention, since the correction operation is performed by the approximate calculation, the number of calculations per pixel is significantly reduced, and the small size and high speed operation are possible.

In the fifth aspect of the present invention, since the device cuts and approximates the higher-order term of the correction calculation formula at the optimum position based on the value of the transfer efficiency recognized by the device, the optimum approximation calculation according to the state of the device can be performed. In addition, high speed operation can be realized.

In the sixth aspect of the present invention, a high quality image having no difference between the channels can be realized by independently performing the correction operation corresponding to each transfer shift register of the CCD.

According to the seventh aspect of the invention, by calculating the calculation parameter in advance before the image reading operation based on the value of the transfer efficiency which the apparatus self-recognizes, the apparatus can be downsized and the high speed operation can be realized.

According to the eighth aspect of the present invention, the number of output bits of the AD converter is set to be larger than the number of output bits of the correction circuit, so that the accuracy of correction calculation is improved and a high quality image can be obtained.

In the ninth aspect of the invention, the size of the device can be reduced by making it monolithic.

According to the tenth and eleventh aspects of the invention,
A high-quality image can be realized by correcting the output (offset output) in the dark and the shading correction after correcting the image deterioration caused by the transfer efficiency.

[Brief description of drawings]

FIG. 1 is an explanatory diagram showing a model of charge shift.

FIG. 2 is an explanatory diagram showing a model of charge shift.

FIG. 3 is a block diagram of the image reading apparatus according to the embodiment of the present invention.

FIG. 4 is a conceptual diagram of a CCD.

FIG. 5 is a diagram showing mathematical expressions.

FIG. 6 is a diagram showing mathematical expressions.

FIG. 7 is a diagram showing mathematical expressions.

FIG. 8 is a diagram showing mathematical expressions.

FIG. 9 is a diagram showing mathematical expressions.

FIG. 10 is a diagram showing mathematical expressions.

FIG. 11 is a diagram showing mathematical expressions.

FIG. 12 is a diagram showing mathematical expressions.

FIG. 13 is an explanatory diagram showing an image change from white to black when the transfer efficiency is 100%.

FIG. 14 is an explanatory diagram showing a fourfold image enlargement.

FIG. 15 is an explanatory diagram showing an image change from white to black when transfer efficiency is deteriorated.

FIG. 16 is an explanatory diagram showing a 4-fold image enlargement.

FIG. 17 is an explanatory diagram showing an image change from white to black when transfer efficiency deterioration occurs in only one channel in the duffle channel CCD.

FIG. 18 is an explanatory diagram showing a fourfold image enlargement.

[Explanation of symbols]

 101 Document 102 Lens 103 CCD 104 Analog Processing Unit 105 A / D Converter 106 Delay Circuit 107a, 107b Multiplier 108 Adder 109 Offset Correction Unit 110 Shading Correction Unit 111 Timing Control Unit 112 Memory 113 Control Unit

Claims (11)

[Claims] 1. A photoelectric conversion element array for sequentially transferring output charges of photoelectric conversion elements arranged in an array and outputting them in time series, and conversion means for analog-digital converting the output of the photoelectric conversion element array. An image reading apparatus comprising: a correction unit that corrects output distortion caused by the transfer efficiency of the photoelectric conversion element array with respect to the output of the conversion unit. 2. The method according to claim 1, wherein Q _n (n = 1, 2)
...) is the A / D conversion output, D _n (n = 1, 2 ...) Is the corrected output, and ε is the transfer efficiency, D _n = (1 / ε ⁿ ) · {Q _n − _{n − 1} C ₁・ (1-ε)
_{· Q n - 1 + n -} 1 C 2 · (1-ε) 2 · Q n - 2 ... +
(-1) ⁿ -2 ・_n -1 C _n -2 ・ (1-ε) ⁿ -2 ・
An image reading apparatus characterized by performing an operation of Q ₂ + (-1) ^{n -1} · (1-ε) ^{n -1} · Q ₁ }. 3. A photoelectric conversion element array for sequentially transferring outputs of photoelectric conversion elements arranged in an array and obtaining the output in time series, and a conversion means for A / D converting the output of the photoelectric conversion element array. A setting for calculating the transfer efficiency of the photoelectric conversion element array based on the output value of the photosensitive element for one pixel arranged a predetermined number of elements behind the last end of the effective pixel row of the photoelectric conversion element array, and setting the transfer efficiency by itself. An image reading apparatus comprising: a conversion unit and a correction unit that corrects an output distortion caused by a transfer efficiency of the photoelectric conversion element array with respect to an output of the conversion unit. 4. The method according to claim 3, wherein Q _n (n = 1,
2 ...) is the A / D conversion output, D _n (n = 1, 2 ...) Is the corrected output, and ε is the transfer efficiency, D _n ≈ (1 / ε ⁿ ) · {Q _n − (n− 1) (1-ε)
Qn- ₁ + [(n-1) (n-2) / 2] (1- [epsilon]) ²
An image reading device characterized by performing an approximate calculation such as Qn- ₂ . 5. The method according to claim 3, wherein Q_n(N = 1,
2 ...) is A / D converted output, D_nCorrect (n = 1, 2 ...)
Post-output, when ε is the transfer efficiency, D_n= (1 / εⁿ) ・ {Q_n−_n-1C₁・ (1-ε)
・ Q_n-1+_n-1C ₂・ (1-ε)²・ Q_n-2… +
(-1)^n-2・_n-1C_n-2・ (1-ε)^n-2・
Q₂+ (-1)^n-1・ (1-ε)^n-1・ Q₁} Approximate calculation is performed and the transfer set by the setting means is performed.
According to the value of the transfer efficiency, the number of terms in the calculation performed by the correction means
By determining the necessary and sufficient
An image reading device characterized by performing an approximate calculation. 6. The correction unit according to claim 1, wherein the correction unit independently corrects the output distortion caused by the transfer efficiency corresponding to the plurality of transfer shift registers included in the photoelectric conversion element array. Characteristic image reading device. 7. The apparatus according to claim 3, wherein prior to the actual image reading operation, the calculation parameter of the correction unit is calculated according to the transfer efficiency calculated by the setting unit, the calculation result is converted into data, and the calculation parameter is read at the time of image reading. An image reading device characterized in that the correction means performs a correction calculation using this data. 8. The method according to claim 1, wherein the output bit number m of the conversion means and the output bit number n of the correction means have a relation of m> n (m, n ≠ 0). Image reading device. 9. The image reading apparatus according to claim 1, wherein the photoelectric conversion element array, the conversion unit, and the correction unit, or the conversion unit and the correction unit are monolithic. 10. A photoelectric conversion element array for sequentially transferring outputs of photoelectric conversion elements arranged in an array and obtaining the output in time series, and a conversion means for A / D converting the output of the photoelectric conversion element array. A first correction means for correcting output distortion of the photoelectric conversion element array due to transfer efficiency with respect to the output of the conversion means, and an output of the first correction means,
An image reading apparatus comprising: a second correction unit that corrects data based on a dark output of the photoelectric conversion element array. 11. A photoelectric conversion element array for sequentially transferring outputs of photoelectric conversion elements arranged in an array and obtaining the output in time series, and conversion means for A / D converting the output of the photoelectric conversion element array. A first correction means for correcting output distortion of the photoelectric conversion element array due to transfer efficiency with respect to the output of the conversion means, and an output of the first correction means,
An image reading apparatus comprising: a second correction unit that corrects a variation in output sensitivity of the photoelectric conversion element array.