JPS6240873A - Picture reader - Google Patents

Abstract

PURPOSE:To restore the picture signal of a faulty bit, by providing a faulty bit correcting means which corrects the picture signal of a faulty bit by estimating the picture signal from the picture signals of surronding picture elements. CONSTITUTION:A shift register 101 makes picture signals A1, A2, and A3 from an on-dimensional line sensor 100 to correspond to the sensor cell positions of the sensor 100 and a faulty bit position 104 caused by the snsor 100 and optical path to the sensor 100 is specified by means of a faulty bit position specifying means 102. When the picture signal A2 of the faulty bit is corrected, a data selector 103 performs the correction by estimating the picture signal from the picture signals of surrounding picture elements. Accordingly, the means 102 informs the signal to the selector 103 by means of a select signal 105 at the timing when the picture signal corresponding to the faulty bit is inputted to the register 101 and the selector 103 restores the are picture element adjacent to the picture element before held by the register 101 as the picture signals of the faulty bit.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to an image reading device using an image sensor such as a COD, and in particular, the present invention relates to an image reading device using an image sensor such as a COD. This invention relates to improvements in image signal correction.

Margins Below [Prior Art] For example, problems specific to CCDI image sensors (imaging devices) include so-called shading and defective bits.

Shading is when a pure white document is read with a scanner using an image sensor, and when the video signal for one cycle of -) two scans of the image sensor is observed, the signal level appears to be distorted without being too sharp. This east appears even when reading an ordinary original, even if it is not a pure white original, and the line sensor outputs a video signal that includes shading. If this signal is converted directly into 2 (ifi or gradation data), the image quality of the output image will be reduced by 1.Therefore, it is necessary to correct the edge margin niting.The cause of shading is The light source, lens imaging characteristics (C094 rule), sensor variations, etc. are considered.

- On the other hand, a defective bit is one in which the output of some bits of an image sensor is extremely low, and that part does not function as a sensor. Causes include, for example, CO
If it is D or the like, it is thought that there is a defect in the - part of the chip itself, or a scratch in the optical system, etc.

Therefore, conventionally, when using an image sensor in an image reading device, 1. Correction of shading, detection of defective bits, restoration of image signals of defective bits, etc. are essential. Conventionally, complicated circuits have been used to detect the defective bits and especially to restore the image signal.

[Problems to be solved by the invention] However, it is difficult to solve the problem by using a complicated unique system to compensate for defective bits.
Providing a 't-shaped circuit has been an obstacle to downsizing and space saving of image reading devices. The present invention has been made in view of the drawbacks of the prior art as described in 1., and proposes an image reading device that is capable of restoring the image value 1 (1) of a defective bit with an extremely complicated configuration. Our goal is to propose an image reading device that pursues miniaturization by using an IG that shares parts of the circuit and the circuit for the purpose of the present invention.

1. In order to achieve the above-mentioned problems, the image reading device of the embodiment shown in FIG. 1 includes a one-dimensional line sensor 100, which is an example of an image sensor; An image value that associates the image signals AI, A2, A3 from the line sensor 100 with the sensor cell position of the image sensor 1) A special means such as a shift register lO1 and a line sensor lO
Defective bit location identification means 1 for identifying the location 104 of a defective bit due to the optical path to the O and line sensor 100
02, and when correcting the image signal A2 of the defective bit, estimate 1.02 from the image signals of surrounding pixels. The data selector 103 is an example of a defect correction means for correcting defective dots.

"Operation" In the configuration shown in FIG.
Due to the signal 105, the data selector 103 selects shift 1/
The image signal of the adjacent pixel one pixel before, which is held in the register 101, is restored as the image signal of the defective bit.

[Example J Hereinafter, the embodiments of the present invention will be summarized according to the attached drawings,
1111 will explain.

Reading operation>'rPJ2 Figure (a) shows the defective bit position d detection and shading correction iTE tl used in the image reading device of the embodiment.
Figure 3 illustrates the concept of movement n for r-determination. Figure 2(a)
The optical system 4 illuminates the white auxiliary plate l attached to the tip of the original platen 2 with the light whale 13 so that it is small in size, and the optical system 4 transmits the inverse Q4 light to the image sensor (in this example, a one-dimensional line (sensor) 3. The optical system 4 and the line sensor 3 move in the direction of the arrow in the figure to determine the shading correction IFμ and the position of the defective bit by the time the white sleeve +Ek1 is scanned 17.

<Configuration> FIG. 2(b) shows the circuit logic for detecting defective bits. In the figure, 5 is a log (
7 is a defective bit 'rIl fixing section; 8 is a shading RAM that stores information regarding the shading correction (5 and defective bits); 9 is a correction section that corrects image data and shading.

If the output 15 from the line sensor 3 is an A/D converted digital level, the LOG converter 5 can convert the brightness level into an image signal of the density level. The image signal 16 at the density level is determined by a defective bit determining section 7 whether or not it is a signal containing a defective bit. and,
Information on whether or not the sensor cell of that bit is a defective bit (defective bit information), and if that bit is not a defective bit, the shape ink complement + IE star (shaping data) is stored in the shape ink RAM 8. do. The defective bit information and shading data are loaded into the shading RAM 8 by the RAM activation signal 12. In this embodiment, the defect information and shape ink data are collectively referred to as shading correction data. In this way, after storing the shading compensation (I data) in the shading RAM 8, the supplementary IE section 9 performs shading compensation on the image signal 16 using the shape ink data stored in the shading RAM 8, while -・Read the shark manuscript.

Further, for the output from the general document due to the cell determined to be in defective focus, an image signal corrected by the method described later is obtained.

Defect pin]・The determination unit 7 is shown in FIG. 3(a) as an example.
It is constructed by an OR circuit or the like so as to avoid the following problems.

Assuming that the density level image value 1416 is n-bit D1 to Do, Ds++ih in FIG. 3(a) is defective bit information, and D, 02, . . . , Dx are shape ink data.

<Determination of supplementary data> Determination of shading supplementary IF data in this embodiment will be explained. When the image signal 16 of ~D1 is input to the defective bit determination unit 7, if the line sensor 3 and the optical system are normal, the bits Dn and Dn near the MSH of the image signal 16 from the white supplement iT and plate 1 are input. ...Dx is "O°'
Therefore, only DA, D2, D3, . . . , DX, which are close to LSB, are output as density levels corresponding to shading.

In addition, how many bits of 4h are treated as shading data and the remaining 1-bit bits are used for determining defective bits, that is, how to determine X, depends on the gradation of the output image data. This is determined depending on conditions such as the nature of the shading RAM and the number of customers of the shading RAM, and is not limited to this embodiment.

Therefore, in this embodiment, if the bit Dn near the MSB
, Dnl, Dn-2.+, Dx (7) If any of them becomes "°1", that bit is determined to be a defective bit or a bit not subject to shading correction. In this way, in this embodiment, the bits that are not targeted for shading compensation 1F are also regarded as defective bits.

The bits that are not subject to the shading compensation IF are bits whose light is scattered due to scratches on the sensor surface glass, etc., and which have an extremely higher density level than the shading data that should normally be processed. In this case, even if shading correction is applied to the image data, the absolute output of the input image is small, making it difficult to obtain image data with good gradation, or the data of that bit may be distorted from page to page due to light scattering on the glass surface. Because there is no guarantee that the data is correct, data that is not subject to shave-even correction is also treated as a defective bit.

According to the OR game (in FIG. 3(a)) 20, either D n ” D x of the image signal 16 at the density level becomes 1
”, the sign of the defective bit is the most negative bit of the shading correction data 10) Ds+6+” 1
” and the F-order bit D near the LSH of the image signal 16
I, D2, D3, . . . , DX are used as shading data as they are. That is, the shading correction data 10 consisting of 4 defective bit information and shape ink data consists of the lowest (-) bit) DSIGN and the lower bits DI, D2,
Consisting of D3..., Dx, shading RAM
8 is stored.

Further, the defective bit can also be detected by the method shown in FIG. 3(b). That is, if the image value number 16 is
・Compare by constant Kan Ai 1 White 21 and Comparator 22] 7
, if the image signal 16 is larger than the threshold value 21, that bit is determined to be a defective bit, and the 10 constant ipoi D S
Let IGN be °゛l゛′. Then, the - part of the shading data remains as defective bit information T) s1+;
It is stored in the shading RAM 8 together with N. In addition, in the methods shown in FIGS. 3(a) and 3(b), only the toy bits are used as shading data, but 1. Since the ☆ bit is determined to be a defective bit when it is 1°", the shading data of the d-order bit is not needed at that time, so the shading data may not be stored in the shading RAM 8.

As shown in 1- below, defective bit information and shape ink data are stored in the shape ink RAM 8. However, once the defective bit information becomes "l", even if the cause is a scratch on the shading correction plate 1, it will be treated as a defect in the line sensor itself.Therefore, in this embodiment, as shown in FIG. As shown in a), even if the sensor bits have defective bit information tν of °゛l', they should not be regarded as defective bits unless they can all be determined as defective bits by scanning the shading correction plate l. Make it. Fourth
Figure (a) shows how the line sensor 3 scans the white correction plate 1 and, for example, the DSIGN for column 2 changes.
- indicates. DSIGN up to line 3 is 0°, but lines 4 and 5 are 1'.

becomes. However, the bit in column 2 whose DSIGN is determined to be "0" even at -.degrees is not regarded as a defective bit.

<Storing Shape Ink Information> In order to do this, the shading RAM G8 actually has a configuration as shown in FIG. 4(b). In the figure, numeral 30 is an address control unit including a RAM address counter cap, and the shading supplement 1F data 10 is the first one.
Defect pin that records - bit MSB (D5IGN)]・Shading data RA that stores information RAM 31 and shading data (DI...)
Address signal 37 is output to M32. 33 is RAM3
This is a RAM control unit that generates a write permission signal of 1.32. The capacity of the RAMs 31 and 32 is equal to the length of the line sensor 3.

The operation will now be explained with reference to FIG. 4(b). The defective bit information DSIGN and shading data (Dz...
) out of 10 shading correction data consisting of 1M5
B Defect information DSIG) I is defect information RAM3
1, and other pins (DI...) of the auxiliary IF Cheetah 1O are input to the shading data RAM 32. The address control unit 30 stores a RAM 31 for each image transfer lock CLK.

32 is counted up to the number of pixels of the line sensor 3, and the water 1i synchronization signal HS is
Clear the internal address counter every YNC. Using this address value, defective bit information and shading data corresponding to each bit of the line sensor 3 are stored in the RAM 31,32.

By the way, the RAM energizing signal 12 input to the RAM control unit 33 becomes "1" when reading the white correction plate l.
Defective bit information RAM31, shading data RA
It is input to the E end f of M32 to energize them. Also, R.A.
The RAM activation signal 12, defective bit information DSIGN, and tree 47-synchronization signal HSYNC are manually input to the M control unit 33. Therefore, the signals output from the RAM control section 33 will be explained.

First, the signal 35 is transmitted for one water period period, that is, one line of the white correction plate 1, immediately after the -1 force signal 12 becomes "1" for the first line of the white correction plate 1. becomes "1" during the time it takes to read.In this way, the first
All bits of shape ink data corresponding to the line are i
! ). For lines 1 and above, when the defective bit information DSIGN is 0'', that is, even when it is not a defective bit, it is 1°, and the writing data RAM 32 is written in the 11th line state. Make it.

On the other hand, the output signal 36 of the RAM control unit 33 first changes to "1" during the 1 water II synchronization period immediately after the RAM inquiry signal 12 becomes "1" for the first line.

Then, when the defective bit information DSIGN is 0゛°, that is, when it is not a defective bit, it becomes 1'', and the information DSIGN-0 indicating that it is not a defective bit is stored in the defective bit information RAM.
Write. When DSIGN is l'', signal 36 is 0''
Therefore, the contents of the defective bit information RAM 31 are not changed.

By the operation described in 1- above, whenever a defective bit is not determined when reading multiple lines of the white correction plate, the defective bit information DSIGN=O is stored in the defective bit information RAM3.
1, so even if it is determined to be a surface defect bit due to scratches or dirt on the white correction plate 1, other lines (
If a defective human is not determined at the final line (substantially the final line), the human's sensor is treated as normal.

This is because even if handled in this way, shading correction can be performed using shape ink data. In this example, the defective bit information and shape ink data are
can be changed while scanning, but another method is to count the number of lines that are DSIGN, and if that number exceeds a predetermined number, the DSIGN will be changed to 1''' from then on.
There is also a person V who says that even if there is a line, it cannot be regarded as a defective bit without rewriting RAM 31, 32.

Shading correction> Shading correction 11 information such as white correction plate reading correction bit and shape ink data? ! )is! In -, it becomes possible to read a general manuscript. That is, shading RAM8
When reading an image, it is switched to read-only mode by the RAM inquiry signal 12. The defective bit information and shape ink data that have been read out are stored in the sleeves 11 and 1.
7, it is input together with the image signal 16 to the correction section 9 in FIG. The correction section 9 has a configuration as shown in FIG. In the figure, numeral 42 indicates the signal distribution portion, and the output 11 of the shading RAM 8 is distributed to the most I-f*pit·(DSIGN) 45 indicating a defective bit and shading density data 44. The shading 81 ship data 44 is input to a subtraction circuit 40 together with the density image signal 16, and shading correction 11 is performed. This shape ink compensation IF- compensates for variations in the characteristics of the line sensor 3 or the characteristics of the lens. In this embodiment, the shape ink compensation i1E is performed at the a1 level, so when performing shading compensation 1 [but shading compensation IF] for the image signal at the luminance level in the subtraction circuit 40, The subtraction circuit 40 of FIG. 5
Use a division circuit instead! It goes without saying.

<Compensation for defective bits 11-> The normal shading compensation IF is performed as shown in (-),
Next, correction of the image signal output from the sensor cell determined to be defective bin i will be described. The image signal 46 subjected to shedding correction by the subtraction circuit 40 is input to a shift register 41 and latched by an image transfer lock. Therefore, since the shift register 41 extends the image signal 4' by one pixel, the image signal 4 which is the output of the shift register 41
7 is an adjacent pixel of one 11f1 of image A, -8 and 46. Therefore, the output 45 of the section 42 is l'''
In other words, when the image signal 46 to be input to the shift register 41 is an image signal from a sensor with a defective bit, the data selector 43 selects and outputs the previous image signal 47 by the select signal 45. Ru. When there is no defective bit, that is, the select signal 45 is °°0.
``'', the image signal 46 of the pixel of interest is output.

(Modified example of supplementary iE section) Correction section 9 in FIG. 2(b), especially shift in FIG. 5 for correction of defective bits) show. This modification is shown in Figure 6 (
As shown in a), when the image signal is from a defective pixel I, the defective bit is compensated for using the data of six pixels surrounding the pixel that responds to the defective bit. In Figure 6(b),
51 to 53 are line buffers, one line buffer γ
has a speed of 8 times corresponding to one line of the line sensor 3. Figure 6 (
54 to 64 in b) are all shift registers, and in particular, the outputs of the shift registers 54 to 62 correspond to the pixel configuration shown in FIG. 6(a). Therefore, the shift register 58 corresponds to A9 in FIG. 6(a).

Now, when the 1111 image data is set to A9 in FIG. 6(a), the determination signal of the defective bit corresponding to the pixel of interest is shifted].
It is output from the register 64. shift register 63.6
4 is because the output synchronization with the shift registers 54 to 62 is increased by 11.

11 The equalization unit 65 calculates the surrounding six pixels Ax around the pixel of interest,
Receive input of A3, A4, A5, A8, As,
Each pixel is multiplied by a coefficient and an 11L average value is output. The line buffer 51 outputs the image signal after shading correction of the line before the pixel of interest, the line buffer 52 outputs the image signal after shading correction IL of the line of the pixel of interest, and the line buffer 53 outputs the image signal after shading correction of the line of the pixel of interest. An image signal after line shading correction is output. The outputs of the line buffers 51 to 53 are input to shift registers 54, 57, and 60, and shifted by shift registers 55, 56, 58, 59, and 61, and 62 with image transfer locks.

Now, when the target pixel A9 is not determined to be a defective bit, the selector 66 outputs the image signal after shading correction II, that is, the output of the shift register 57 as is. If the pixel of interest A9 is determined to be a defective bit, the output of the shift register 64 is the determination information and becomes the select signal of the selector 66, so the output of the i-balancing unit 65 is selected and output by the selector 66. Ru. 1i equalization section 6
5, the following processing is performed.

Equalized output Z = (A+B+÷ A3B3+ A4B4+ A3
B5+ A6B6+ AsBs )/(81+ B
3 + B4 + F35 + 86 + Be) In other words, a IIi equalized output weighted to the surrounding six pixels is obtained.

(In the case of area sensor) In the above-mentioned embodiment, the line sensor was explained in [2]. However, the present invention is not applied only to line sensors. First, let's talk about detecting defective bits when using an area sensor.
The capacities of the defective bit information RAM3L and the shading data RAM shown in FIG. 3(b) are increased by the number of sensor cells.

The timing of writing the information to be written into these RAMs is either changed for each line as in the above-described embodiment, or the timing is changed by changing the timing of writing the information to be written into these RAMs, or by using the white correction plate 11. If RA is scanned by the number of cells in the vertical direction of the area sensor, then RA
All you have to do is rewrite M.

In this way, it has become clear that the method applied to the one-dimensional line sensor described above is also applicable to the one-rear sensor.

Next, the defective bit correction 1 [′
To explain the case where the r method used for a one-dimensional line sensor is applied, it is necessary to refer to the 8 pixels surrounding the pixel of interest A9, which is considered to be the output from a defective bit cell, and calculate and correct it as shown in the following formula. .

Averaged output Z ” (AIBd” A2B2+ A3B3” A
4B4+ A5B! 1+ A6B6+ AtB1
+ AsBe ) / (B+ + B7 + B:l + B4 + B5
+ B6 + 8+ + 88) In this way, even in the case of an area sensor, there is no need to make any changes to the configuration of the above embodiment.

<Other Modifications> In the two embodiments described in 1-1, the detection of defective bits is
There is a defect at the ship level, but then there is a defect at the brightness level]
Figure 7 shows the detection of .

As with the density level, the 1st order bit (D1 to DX)
is manually applied to the NAND gate 70. If the data does not apply to the 1st floor of the shading sleeve, the brightness level will be low.NA
Any of the inputs to the ND gate becomes "o", and the output of DSIGN becomes °°l', which is treated as a defective bit. Further, Di to Dn are stored as shading luminance data in the shading RAM together with the defect determination data.

Also, similarly to method D shown in FIG. 3(b), the brightness level data is compared with a certain threshold value P, and when the brightness level data is smaller than the threshold value P, it is determined that the bit is defective and the defective bit information is Set DSIGN to °゛l''' and save the %11 degree data as is to shape ink RAM 8 along with DSIGN.
It is also possible to store the information in .

"Effects" As explained above, according to the present invention, it is possible to restore the image signal of defective bits with a simple configuration. The part that is incorporated into the box becomes a mouthpiece, and further miniaturization becomes possible.

[Brief explanation of drawings]

FIG. 1 is a diagram explaining the basic operation of an embodiment according to the present invention. FIG. 2(a) is a diagram showing an optical path for detecting a defective bit in the embodiment. FIG. 2(b) is a circuit diagram of the embodiment. Figures 3(a) and 3(b) are diagrams explaining the determination of shading correction data, Figure 4(a) is a diagram explaining the basic operation for determining defective bits, and Figure 4(b) is a diagram of an example of the internal circuit of the shading RAM, FIG. 5 is a circuit diagram of the supplementary section 11, FIG. ) is a circuit diagram of a modification of the sleeve)-[ section, and FIG. 7 is a diagram for explaining the determination of defective filler I when detecting defective bits based on the luminance level. In the figure, 1... white supplementary IF board, 3... line sensor,
5...LOG converter, 7...defect pin) 'Fil
Fixed part, 8... Shading RAM, 9... Correction part, 30... Address control part, 31... Defective bit information RAM, 32... Shading data RAM, 3
3... RAM control unit, 40... Accelerator, 41...
Schiff] register, 42...4" E number is part, 4
3...Data selector, 51-53...Line memory, 54-62...Shift register, 65...IL
This is the equalization section.

Claims (6)

[Claims] (1) An image signal specifying means for associating an image signal from an image sensor with a sensor cell position of the image sensor, and a defective bit position specifying means for specifying the position of a defective bit caused by the image sensor and the optical path to the image sensor. and defective bit correction means for correcting the image signal of the defective bit by estimating it from the image signals of surrounding pixels when correcting the image signal of the defective bit. (2) The image reading device according to claim 1, wherein the surrounding pixels are image signals of pixels adjacent to each other in the main scanning direction or the sub-scanning direction of the image sensor. (3) The image reading device according to claim 1, wherein the surrounding pixels are image signals of six or eight pixels adjacent to each other in both main and sub-scanning directions of the image sensor. (4) A shading correction means is further provided, and the defective bit correction means uses an image signal after shading correction of surrounding pixels is used. image reading device. (5) The image reading device according to claim 4, wherein the image signal is represented by density, and the defective bit correction means performs the correction by adding and subtracting the density data. (6) The image reading device according to claim 4, wherein the image signal is represented by luminance, and the defective bit correction means performs the correction by integrating or dividing density data.