JPH01194657A - Picture input device - Google Patents

Abstract

PURPOSE:To effectively eliminate the fluctuation and level difference of picture data due to defective shading by detecting the discontinuous elements of the picture data on boundary parts in plural reading areas, calculating the data to correct the discontinuity of the picture data, and eliminating the discontinuous elements on the boundary parts from the original picture data. CONSTITUTION:Light is converted into electric signals R1, G1 and B1 by CCDs 1r, 1g and 1b, the electric signals at every color are converted into digital data R2, G2 and B2 at every picture element by an A/D converters 2r, 2g and 2b, and sensitive unevenness is corrected by a shading correcting circuit 2. For one part of the signals R4, G4 and B4, which are processed in the above mentioned way, noise components are decreased by a smoothing circuit 5, thereafter, they are inputted as shading defective detecting signals R5, G5 and B5, the original unsmoothed signals R4, G4 and B4 are inputted as they are to a tilt correcting circuit 6 respectively, the defective shading is corrected, and picture signals R6, G6 and B6 at every final color are obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an image input device, and more particularly to an image input device that divides an image into a plurality of areas and reads the image.

[Conventional technology]

Image input devices used in facsimile machines and digital copying machines usually use a light-to-electric conversion element such as a COD. At this time, due to the characteristics of the optical system and conversion element, the sensitivity of the element to human power differs depending on the position even within the same element, so it is not possible to obtain a correct image signal as it is.For this reason, a shading circuit is usually installed to reduce the apparent uniformity. It is common to perform electrical correction to obtain sensitivity.

That is, read the white data and black data for reference,
A mathematical operation is performed so that these data become uniform data between each pixel. Normally, sampling of reference data for calculating this correction value is performed prior to reading the original image by, for example, reading a standard white plate installed at the home position of the original platen, and after that sampling, the actual original Perform reading.

Furthermore, in order to read a document image larger than the document size that can be read by a photoelectric conversion element, an image input device has been proposed that divides an i-striped image into multiple strip-shaped areas and sequentially reads the image in each area. There is.

[Problem that the invention is trying to solve] However,
When inputting a dog-sized document by scanning it multiple times, the following phenomenon occurs as the number of scans increases.

First, the mechanical displacement of the optical system due to the deflection of the document table, etc.
It increases as the distance from the home position where the shading data was sampled increases. Furthermore, as time elapses until a distant area is scanned, changes over time in the temperature of the input system, etc. also increase.

Therefore, due to such a phenomenon, the input sensitivity changes from the initial state, resulting in a situation where correct shading cannot be performed. In other words, in general, the input sensitivity of the reading range tends to lean to one side compared to the initial one, so the obtained image data has a discontinuous offset added at the seam of scanning, and the resulting image data differs from the actual image. It's coming.

In this way, boundary lines with discontinuous density occur on the image, resulting in significant deterioration of image quality. This results in so-called shading failure.

There are several ways to avoid this, as follows.

(i) The reading area of the manual element is widened and the image is manually read without dividing it.

(ii) Sample the shading data again before a shading failure occurs.

Among these methods, method (i) is a fundamental solution, but it is extremely difficult and expensive to manufacture an element large enough to read, for example, an A1-size original.

Furthermore, the subsequent processing circuitry involved will also be correspondingly large-scale, so this is not a practical means.

On the other hand, method (ii) does not pose any significant technical or cost difficulties, but has the disadvantage that it takes time to sample the shading data, which slows down the reading speed of the input device.

[Means for solving problems]

The present invention has been made in view of the above points, and includes image reading means that divides an image to be read into a plurality of regions and reads the image of each region, and detects discontinuous components of image data at the boundary portions of the plurality of reading regions. means for calculating data for correcting the discontinuity of image data according to the discontinuous component; and removal means for removing the discontinuous component at the boundary portion from the correction data and the read original image data. The present invention provides an image input device having the following.

〔Example〕

FIG. 1 is an example of an electric circuit configuration of a human-powered image device to which the present invention is applied. The CCDs 1r, Ig, and lb convert the light into electrical signals RI, Gr, and B+ representing each color component of the color image, and the A/D converters 2r, 2g, and 2b convert the electrical signals for each color into digital data R2 for each pixel. ,2+8
Convert to 2. R,,G2. The data of B is corrected for sensitivity unevenness by the shading correction circuit 2, and is converted to R8°G3.
．． R3, and further undergoes correction of the spectral sensitivity of the optical filter in the manual masking circuit 3. The signals R4, G4, and R4 that have undergone the above processing are partially reduced in noise components by the smoothing circuit 5, and are then used as the shading defect detection signal R%.
+ GS + B! The unsmoothed original signals R4, G4゜R4 are input to the tilt correction circuit 6 as they are, and are corrected for shading defects, and are converted into final image signals Ra, Ga, Be for each color. Become.

FIG. 2 shows the mechanical structure of an image manpower device using the electrical circuit of FIG.

CCI) Lnit 7 is CCD1r shown in the first diagram.

CCD8 consisting of IG and IB. It is a unit consisting of a lens 9, etc., and is driven by a drive system in the main scanning direction consisting of a main scanning motor 11, a pulley 12, a pulley 13, and a wire 14 fixed on the rail 10, and moves the rail 10 in the direction of the arrow and vice versa. When moving in the forward direction, the image of the document placed on the document table glass 15 is read in the main scanning direction. Also, the rail 10 is the rail 16.17
It rides on the sub-scanning motor 18, pulley 19.
20, 21, 22, shafts 23, 24, and wires 25, 26, the drive system in the sub-scanning direction moves eight times in the direction of arrow B and in the opposite direction.

FIG. 3 is a simplified drawing of how the CCD 8 scans when reading a document using the CCD unit 7. The CCD8 moves from the home position HP in the main scanning direction.
It performs 6 scans, then returns to the home position HP in the main scanning direction to perform the next scan, and moves in the sub-scanning direction by the width of the image to be read, before starting the next scan. It should be noted here that the reading width of the CCD 8 and the actual reading width are wider in the former. Therefore, the line segment A at the boundary between adjacent scan amounts, e.g.
The points on B can be read out as image signals by scanning twice.

FIG. 4(A) shows a cross section of the document glass plate 15 in the sub-scanning direction and a shading correction output. In this figure, the deflection of the glass 15 is exaggerated to make it easier to understand, and is not the actual deflection itself. In addition, various displacements other than the deflection of the glass 15 can also be equivalently expressed as
It is assumed that the expression is added to the deflection of 5. Because of such displacement, the desired shading amount varies depending on the position in the sub-scanning direction. Even though this device is shown in FIG. 4(B) with different shading corrections performed in the sub-scanning direction, if fixed shading is performed only using the image data under the standard white plate RW,
Due to the deviation from the ideal shading amount, even if a document with uniform density is read, the output data will not be uniform depending on the sub-scanning position as shown in (C). For this reason, in the sub-scanning direction, density unevenness occurs within one scanning section, and steps occur at the boundaries of each scanning section, resulting in deterioration of image quality.

In particular, the latter discontinuity in density at the boundary affects as a very clear noise component when outputting the read image data by an error diffusion method or the like.

The reference figures are actual images that were manually generated, binarized and output using the error diffusion method, and (a) and (b) show a noticeable deterioration in image quality compared to (a).

The smoothing circuit 5 and the tilt correction circuit 6 of the circuit shown in FIG. 1 eliminate the above-mentioned shading defects in the sub-scanning direction. The smoothing circuit 5 is for reducing noise components eaten by the input signal, and performs appropriate weighting within a predetermined unit block, for example, a 3×3 pixel range, and outputs an average value.

The tilt correction circuit 6 uses this smoothed image data to detect the tilt of an offset due to a shading defect.

FIG. 5 shows an example of the configuration of the tilt correction circuit 6. The memory 27 stores smoothed image data of the boundary portion from one scan ago. The gate 28 passes the input data from the smoothing circuit 5 to the memory 2 at the end of the readout of the CCD 8, that is, at the timing when the data at the boundary between the current scan and the next scan comes in, and the gate 28 passes the input data from the smoothing circuit 5 to the memory 2 at the beginning of the readout of the CCD8, that is, at the timing when data comes in at the boundary between that scan and the next scan. Data is passed from the memory 27 to the subtracter 30 at the timing when the scan and the boundary data of that scan are received. The address control circuit 29 controls read/write addresses of the memory 27.

The subtracter 30 is a circuit that detects a difference in level at the boundary between adjacent sub-scanning amounts by subtracting the data from the previous scan read from the memory 27 from the data HD input from the smoothing circuit 5. The correction value between adjacent pixels is detected by dividing it by the value corresponding to the number of pixels read out for one line of the CCD 8, and this result is stored in the latch 32 while the CCD 8 reads out one line. be done. On the other hand, the subtracter detects a step difference at the boundary between adjacent sub-scan amounts by subtracting the current scan data HD input from the smoothing circuit 5 from the pre-scan boundary data read from the memory 27. At 33,
The output of this subtracter 33 is transmitted to a selector 34 and a latch 35.
is added to the output of the latch 32 in the adder 36,
Correction data CD is calculated. This correction data CD is added to the image data MD from the input masking circuit 4 by an adder 37.
In addition, corrections are made. At this time, the correction data CD stored in the latch 38 passes through the selector 34 and the latch 35, and is again added to the output of the latch 32 to become correction data for the next pixel. That is, the content stored in the latch 32 is equal to the change in the correction value for each pixel.

In this circuit configuration, the variation of the correction value is fixed to a power of 2, but if the number of read pixels of the CCD 8 is not a power of 2, the latch signal LAT is changed at an appropriate timing.
If CH2 is given, a reading width of any length can be supported. For this purpose, the timing signal generation circuit 39
LATCHI, LA at the timing specified by U40
Generates various signals including TCH2. In addition, the CPU 40 controls the amount of software for the shifter 31 and the address control unit 29.
Also performs initial data settings.

Although FIG. 5 only represents the configuration for one of the three color signals of R, G, and B, this circuit may be provided for each color, or latches 32 and 35 may be provided for each color. The serial data may be processed sequentially.

In FIG. 1, the tilt correction circuit 6 is placed after the input masking circuit 4, but after the shading circuit,
It is also possible to place it in another part that processes multivalued data.

FIG. 6 shows the human power H from the smoothing circuit 5 to the slope correction circuit 6.
D, tilt correction data CD, and output data MD' obtained by adding both of them are shown. HS is a synchronization signal for reading out one line of the CCD 8, and the X point at the end of the CCD readout area in the i-th scan and the first Y point in the CCDM extrusion area in the i+1th scan are the same point on the document. Image data for . This is because the sub-scanning advances by the length of the COD readout area for each main scanning circuit.

Correction data CD for the i+1st scan, and, at the time of the Y point, are data A at the X point of the smoothed human data HD in the i-th scan to data B at the Y point of the smoothed input data HDI in the i+1th scan. is equal to minus
Thereafter, as the CCDI line is read out, the correction amount linearly approaches O and becomes 0 at the rear end of the readout area. By adding this correction data CD++t to the image data MD, the inclination of the image data and the level difference in the boundary area due to poor shading are removed. Note that FIG. 7 is shown assuming that the image data MD is equal to the smoothed input data MD.

Further, according to the circuit configuration shown in FIG. 5, the correction operation shown in FIG. 6 can be performed in real time in synchronization with data reading from the CCD 8.

FIG. 7 shows timing signals applied from the timing generation circuit 39 to each part of the circuit of FIG. 6. gate 3
8 is data H to the memory 27 when the SEL signal is low level.
D is opened in the writing direction, and memory 27 is opened when the level is high.
data) Open in the direction to read the ID. Selector 34
When the SEL signal is at a low level, the A input from the subtracter 33 is selected, and when the SEL signal is at a high level, the B input from the latch 38 is selected. The latches 32, 35, and 38 perform latch operations using latch signals LATCHI, LATCH2, and LATCH2, respectively. The correction data CD is the signal LATCH2
Accordingly, only the contents held by the latch 32 change. Therefore, by changing the timing of this signal,
The changing point of the correction data can be changed freely, and it can support reading widths of unspecified length, such as enlargement or reduction. The CPU 40 sets these timings. READ, WRIT
The signal E is the read/write timing for the memory 27.

As described above, it is possible to obtain image data in which the level difference at the boundary between adjacent sub-scanning amounts and the level difference between adjacent pixels caused by shading defects are corrected, and therefore the density level changes in the reproduced image. ,
Inconveniences such as differences in level can be eliminated.

[Second Embodiment] FIG. 8 shows a tilt correction circuit 6 with a different circuit configuration.
An example is shown below. The memory 27, gate 28, address control circuit 29, subtracter 30, and latch 32 are used to calculate the difference in data from the previous scan on the boundary line of the main scan, similar to the circuit configuration shown in FIG. This difference is multiplied by the value of the counter 46 in a multiplier 47, divided by a shifter 48 for alignment to obtain correction data CD, and correction data CD and manual data MD are added in an adder 49 to perform correction.

The counter 46 first sets the number of pixels to be read out from the CCD and counts down to 0 at the end of CCD reading, thereby updating the correction data for each pixel.

The circuit shown in FIG. 8 may also be provided for each of the three colors R, G, and B, and by providing a latch 32 for each color, it can accommodate serial inputs of R, G, and B. FIG. 9 shows various timing signals set by the CPU 51 and generated by the timing generation circuit 50. In addition to setting the timing, the C0U 51 also sets the initial value of the address control section 29, the initial value of the counter 46, and the shift amount of the shifter 48.

Further, the gate 28 passes data HD from the smoothing human power 5 to the memory 27 when the select signal SEL is at a low level, and passes data from the memory 27 to the subtracter 30 when the select signal SEL is at a high level. When the latch 32 receives the LATCH signal, the subtractor 3
Latch the output result of 0. Counter 46 is C0UNT
Performs counting operation according to the signal.

[Third Embodiment] FIG. 10 shows the configuration of an electric circuit of a human-powered image device having a different configuration. CCD52r, 52g, 5
Electrical signals R+, G of the image for each color read in 2b
+, B+ are variable gain amplifiers 53r, 53g, 53
The digital signal Rs is amplified or attenuated by A/D converters 54r, 54g, and 54b.
, G3. Converted to B3. Shading correction circuit 5
The signals R4, 4, 84, which have been corrected for non-uniformity due to photo-electrical conversion of the CCD by 5, are corrected for the spectral sensitivity of the CCD by the input masking circuit 56, and then R5+G s *
It is output as Bs.

On the other hand, after the noise component of the obtained signal R5+8+BS is reduced by the smoothing circuit 57, the signal at the rear end of the CCD reading area passes through the gate 58 and is stored in the memory 59, and the signal at the leading end of the reading area is stored in the memory 59. Smoothing circuit 5
7 to the subtractor 618. The subtracter 61 receives the signal H at the tip of the readout area, which is the output of the smoothing circuit 57.
D and the signal HD' at the rear end of the readout area read from the memory 59 and stored during the previous scan,
It is held in latch 62.

The value of the counter 63 is added to this held value by the multiplier 64.
and calculates the tilt correction signal CD. After converting this signal CD into an analog signal using the D/A converter 65,
Variable gain amplifier 53r.

Input as each gain control signal of 53g and 53b, CC
By correcting the amplitude of each output of D52r, 52g, and 52b, the tilt of image data due to defective shading is corrected.

The correction data is updated by the counter 6 as it is read out from the CCD.
This is done by counting down or counting up the contents of step 3. The address control circuit 60 controls the reading and writing addresses of the memory 59. Timing signal generation circuit 6
7 repeatedly generates various signals at timings set by the CPU 66. The CPU 66 also has a counter 6
3, the initial value of the address control circuit 60, etc. are set.

In this embodiment, the timing signal given from the timing signal generation circuit 67 is generated and operated in the same manner as in the second embodiment shown in FIG.

In FIG. 10, the same shading slope correction is performed for each color of R, G, and B, but it is of course possible to provide an independent circuit for each color and apply the correction separately.

In this embodiment, when performing correction in the first and second embodiments, correction data is added to the image data to approximately correct shading defects, whereas the variable gain amplifier is used to add the correction data to the image data and correct the correction data. By multiplying by , it is possible to perform correction that does not rely on approximation.

[Fourth Embodiment] FIG. 11 shows the tilt correction circuit 6 shown in FIG. 8 described in the second embodiment adapted to correspond to the case where RGBa color data is serially input. . The difference from FIG. 8 is that latches for latching the output of the subtracter 30 are provided for each color of RGB, and these latches 72r and 72g.
, 72b is provided, and the timing generation circuit 79 is modified to generate timing signals LATCHR, G, and B. Color selection signal C3ELO, C3 of selector 73
The ELI may be generated by the timing generation circuit 79 or may be provided from outside the slope correction circuit 6. The selector 74 selects C3EL, 1. When both O are at high level, latch R is at high level, C5ELI is at high level, and when C5ELO is at low level, latch G is at low level, C5ELI is at low level, C3E
It is assumed that each latch B is selected when LO is at high level. These timing signals are shown in FIG.

As can be seen from this embodiment, when performing tilt correction processing on serial data, latch circuits 72r, 72g, 72
By simply adding b and selector circuit 74, most of the other circuits can be used in common. This is particularly effective when the CCD readout signal is obtained in RGB serial form, such as when using a color sensor in which R, G, and B filters are periodically arranged on multiple light-receiving elements arranged in a line. It is. Of course, not only the second embodiment but also the first embodiment and the third embodiment can have a serial compatible circuit configuration.

As described above, according to the configuration of the above embodiment, in an image input device that reads an input image by dividing it into a plurality of regions, by correcting the tilt component of image data due to shading defects, data for shading can be read many times. It is possible to remove discontinuous changes in the scan connection portion of an image without performing sampling, and improve the image quality. In this embodiment, full-color image reading has been described, but it goes without saying that the present invention can also be applied to apparatuses that perform monochrome image reading or multi-color reading such as two-color or three-color image reading.

〔effect〕

As explained above, according to the present invention, when an image to be read is divided into a plurality of regions and the image of each region is read, discontinuous components of image data at the boundaries of the plurality of reading regions are detected and discontinuous components are detected. Data for correcting discontinuities in image data is calculated according to the component, and discontinuous components at boundaries are removed from this correction data and the read original image data, so fluctuations and steps in image data due to poor shading can be efficiently corrected. It can be removed well and good image data can be obtained.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an image input device implementing the present invention;
Fig. 2 shows the mechanical configuration of the image input device, Fig. 3 shows the operation during image reading, Fig. 4 shows the mechanism by which shading defects occur, and Fig. 5 shows the mechanism by which shading defects occur. 6 is a diagram showing the operation of the tilt correction circuit, FIG. 7 is a diagram showing the timing signal generated by the timing signal generation circuit in FIG. 5, and FIG. 8 is a diagram showing the timing signal generated by the timing signal generation circuit in FIG. FIG. 9 is a diagram showing the timing signal generated by the timing generation circuit of FIG. 8, and FIG. 10 is a circuit example of the image manpower device of the third embodiment. FIG. 11 is a block diagram showing a configuration example of the tilt correction circuit of the fourth embodiment, and FIG. 12 is a diagram showing the timing of the circuit in FIG. 11. lr, Ig, lb. 52r, 52g, 52b++++zz++z+l::(1
) 2r, 2g, 2b...A/D converter 3...Shading correction circuit 4...・・・・・・・・・Input masking circuit 5・・・・・・・・・・・・・・・・
・・・・・・・・・・Smoothing circuit 6・・・・・・・
・・・・・・・・・・・・・・・・・・・Tilt correction circuit 7・・・・・・・・・・・・・・・・・・・・・・−
CCD unit 8・・・・・・・・・・・・・・・・・・
・・・・・・・・・・・・・・・CCD15・・・・・・
・・・・・・・・・・・・・・・・・・・・・Manuscript glass 27.59・・・・・・・・・・・・・・・・・・・
・・・・・・Memory 28.58・・・・・・・・・・・・
・・・・・・・・・・・・Gate 29.60・・・・・・
・・・・・・Address control circuit 40, 51, 66
．． 80・・・・・・・・・CPU39.50,6
7.79... Timing signal generation circuit 53r, 53g, 53b... Variable gain amplifier male
Figure R: Darkness) Figure 7 Male Figure q H3 Coryl Male/Figure 2

Claims (1)

[Claims] an image reading means for dividing an image to be read into a plurality of regions and reading an image in each region; a means for detecting a discontinuous component of the image data at a boundary between the plurality of reading regions; and a means for detecting a discontinuous component of the image data in accordance with the discontinuous component. An image input device comprising means for calculating data for correcting continuity, and removal means for removing discontinuous components at the boundary portion from the correction data and the read original image data.