JPH04275776A - Picture reader - Google Patents

Abstract

PURPOSE:To reduce the capacity of a memory through the use of a reduction- type line sensor and to correct a gap between sensors without deteriorating picture quality by providing a means thinning one line data at every prescribed linen, and a means interpolating line data which is thinned on the output-side of a delay means. CONSTITUTION:Picture data is thinned for one line at every two lines in the thinning circuit 10 and it is inputted to a correction delay memory 11. Data delayed by the number of the prescribed line is outputted from the memory 11. In a line correction circuit 12, data are delayed by one line in FIFO memories 41 and 42. The outputs of the FIFO memories 11, 41 and 42 are selected in a selector 43 and two line data are outputted. Two of coefficients A-D are selected in a selector 44 and the prescribed coefficients are multiplied by adjacent line data in multipliers 45 and 46. Then, they are added in an addition circuit 47 and data which is line-interpolated is outputted.

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 in which the capacity of a memory for correcting the gap between line sensors in the sub-scanning direction is reduced when a color image is read using a reduced line sensor.

0002

[Prior Art] Conventionally, a color image reading device is shown in FIG.
As shown in (a), the original 2 placed on the surface of the platen 1 is irradiated with a light source (not shown), and the reflected light is reflected from the mirrors 3, 4, 5.
, an image is formed on a reduction type CCD line sensor 7 via an imaging lens 6 and read. The CCD line sensor 7 consists of three line sensors 7a, 7b, and 7c for R, G, and B parallel to the main scanning direction, as shown in FIG. 8(a). The original image is sequentially read by scanning in the sub-scanning direction, and R, G, and B image signals are output.

As described above, the CCD line sensors 7 are arranged offset in the sub-scanning direction, and data read at the same timing means data at different positions. In order to synthesize the data, it is necessary to delay the data from the preceding sensor to the rear sensor to match the timing. For example, the gap between each sensor is 168 μm, and the width of one pixel is 1
If it is 4 μm, there is a gap of 12 lines between each sensor, and a gap of 24 lines exists between sensors 7a and 7c. Assuming that images are read at a constant cycle regardless of the magnification, as the magnification increases, the scanning speed slows down and the amount of image data read for the same amount of sensor movement increases, which further widens the gap.
The % magnification is 24 lines between GB and 48 lines between RB.
If the magnification is 400%, the gaps will be 48 lines and 96 lines, respectively.

In order to perform such gap correction, when scanning is performed in the order of R, G, and B as shown in FIG. 9, four 262k×4-bit memories (8a to 8d), two memories (9a, 9b) of the same capacity were prepared for the G signal. For example, a pair of memories 8a and 8b and a pair of memories 8c and 8d each have a memory capacity of 262 kB, or approximately 52 lines.
A total of 104 lines of delay is possible for both pairs, so R
For signals, a pair of memories 8a, 8b and 8c, 8d can handle 96 lines at a magnification of 400%. Also, regarding the G signal, since the memories 9a and 9b have a memory capacity for 52 lines, it is possible to perform a delay of 48 lines when the magnification is 400%.

Now, considering the case where the magnification is 365%, the sampling density in the sub-scanning direction is 3.65 when it is 100%.
The gap between pixel columns between RB is 12×2×3.6
5=87.6 lines. Therefore, the gap correction memory shown in FIG. 9(a) is used to delay the 87 lines, and as shown in FIG. ...R
185, R186, R187, R188, R18
9..." is obtained. Next, by moving average, from R185 and R186 to R1
85.6, R186 and R187 to R186.6
, R187 and R188 to R187.6, and the output is delayed by 87.6 lines "...R185.6, R186.6,
R187.6, R188.6..." is obtained and the B signal output is "...B98, B99, B100, B101...
” I try to match the timing. Similarly, the timing of the G signal is adjusted by delaying it by 43.8 lines.

[0006]

[Problems to be Solved by the Invention] As described above, in gap correction in conventional image reading devices, a memory capacity that can accommodate the delay at the maximum magnification is prepared, and this correction memory is used to read the preceding R data, Delay G data,
The remaining fractions were determined from adjacent pixels by moving averaging. Such a gap correction method requires a large-capacity memory, especially in accordance with the maximum magnification, which increases the cost of the apparatus. In order to prevent such an increase in memory capacity, it has been proposed to extract image data from a necessary portion of an image in one line of main scanning according to the magnification ratio, and then delay it in a delay memory to perform gap correction. (Unexamined Japanese Patent Publication No. 1-141460)
However, this method cannot handle cases where the main scanning magnification is equal to the magnification and the sub-scanning magnification is enlarged.
Replace the original shown in Figure 10(b) by turning the original 90 degrees so that the main scanning and sub-scanning are reversed, and extract the necessary part of the image in one main scanning line. I had no choice but to do this.

[0007] The present invention is intended to solve the above-mentioned problems, and uses a reduced-type line sensor to reduce memory capacity, do not degrade image quality, and can accommodate any reduction/enlargement in space between sensors. An object of the present invention is to provide an image reading device capable of performing gap correction.

[0008]

[Means for Solving the Problems] The present invention provides a plurality of line sensors parallel to the main scanning direction that output different colors, arranged at predetermined intervals in the sub-scanning direction, and a gap in the sub-scanning direction between each line sensor. In a color image reading device having a delay means for correction, means for thinning out at least one line data for each predetermined line is provided on the input side of the delay means, and means for thinning the thinned line data on the output side of the delay means. The feature is that a line data interpolation means for performing interpolation is provided, and further includes a first switching means for replacing each signal provided at a stage before the thinning means, and a first switching means for switching each signal provided at a stage after the line data interpolation means. 2nd to swap signals
A switching means is provided to switch the signals to be delayed during forward scan and back scan, and line thinning and line interpolation are performed when the image reading magnification is equal to or higher than a predetermined value. In addition, the plurality of line sensors are R, G, and B line sensors, and a B line is placed in the last row during forward scanning.
Alternatively, a G line sensor is arranged, and the read signals of other line sensors are subjected to delay processing for the B or G line sensor.

[0009]

[Operation] The present invention thins out the data input to the gap correction memory, for example, line by line, inputs it to the correction memory at 1/2 the density of the conventional method, and uses adjacent line data to output data from the correction memory. By generating data, performing interpolation to double the number of lines, and performing correction for the remaining fraction by moving averaging, gap correction can be performed with half the memory capacity of the conventional method. Then, if data thinning and interpolation are performed only when the magnification is 200% or more, for example, the memory capacity can be reduced with almost no effect on image quality since the data is thinned out when the sampling density is very high. can do. Note that the rate of thinning is not limited to each line, but may be performed for each plurality of lines, and the number of lines to be thinned out is not limited to one line, but may be two or more lines. Further, if the delay operation is performed after extracting necessary pixels in one main scanning line during zooming, the memory capacity can be further reduced.

0010

[Embodiment] Fig. 1 is a diagram for explaining gap correction in the present invention, Fig. 2 is a video signal processing circuit diagram to which the gap correction of Fig. 1 is applied, and Fig. 3 is a diagram showing specific correction of the gap correction. FIG. 4 is a diagram showing the principle of gap correction, and FIG. 5 is a timing chart of gap correction.

First, the overall configuration in which the gap correction of the present invention is used will be explained with reference to FIG.

FIG. 2 is a block diagram for explaining video signal reading and shading correction of the present invention.

In the figure, the line sensor 20 has R, G,
B consists of each sensor 20a, 20b, 20c, R and B
There is a 24-line gap between and, and a 12-line gap between G and B.

Now, to explain the reading of R,
The signals read by the sensor 20a are divided into odd-numbered signals and even-numbered signals and are extracted from both sides. In other words, a reduced CCD
Line sensors have a large number of pixels per line, and in order to increase reading speed and reduce transfer loss, odd-numbered and even-numbered charges are carried and output from both the left and right sides. After sampling and holding the read video signal, the AGC22
The gain is adjusted by the AOC 23, the offset is adjusted by the AOC 23 to correct between the odd and even signals, the signals are combined by the multiplexer 24, the offset is adjusted, and A/D conversion is performed. The detection signal is 6.75 MHz when divided into odd and even numbers, but when combined, it becomes a signal with double the frequency of 13.5 MHz.

The 8-bit data converted by the A/D converter 26 is corrected by a gap correction circuit 27 so that the timing matches the B signal. Therefore, the gap correction circuit 27
In this case, 512 KB of memory is used, and since the gap between G and B is half of that, 256 KB of memory is used.

The shading memory 29 reads and stores data when the reading device comes directly under the white reference plate according to instructions from the CPU 31. At the time of copying, the read image data is divided by the data stored in the memory 29 in the division circuit 28 to perform shading correction. Since the shading-corrected data is reflectance data, it is converted into desired data using the LUT of the ENL 30.

The configuration of the gap correction circuit is shown in FIG. The FIFO memories 11a, 11b or 13a, 13b are delay memories similar to those explained in FIG. The structure is designed to reduce the number of items. A thinning circuit 10 is provided before the correction delay memories 11a and 11b, and thins out one line of data every two lines and inputs it to the correction memory. Therefore, the correction delay memory can be used if the memory capacity is 1/2 that of the conventional one. The data delayed by the correction delay memory can be synchronized with the B signal by generating and interpolating the thinned data using adjacent line data in the line interpolation circuit 12 and outputting the resulting data. Note that since the G signal can be handled by two FIFO memories, line thinning and interpolation are not particularly performed.

Next, a specific example of gap correction will be explained with reference to FIGS. 3, 4, and 5.

In FIG. 4, the image data is processed by the thinning circuit 1.
0, one line is subtracted every other line and input to the correction delay memory 11, and the data delayed by a predetermined number of lines is output from the memory 11. In the line interpolation circuit 12, FIFO memories 41 and 42 each delay one line, the outputs of the FIFO memories 11, 41, and 42 are selected by a selector 43, and two of them are output, and adjacent line data are Selector 44 selects coefficient A.
, B, C, and D are selected, multipliers 45 and 46 respectively multiply them by predetermined coefficients, and the adder circuit
and line interpolated data is output.

Now, in order to correspond to FIG. 9(b), a case where the magnification is 365% will be explained. As mentioned above, 36
In the case of 5%, a pixel column gap of 87.6 lines occurs between RB, and it is necessary to adjust the timing by delaying R by 87.6 lines. Therefore, first, thinning circuit 1
0, every other line is subtracted for 1 line, the FIFO memory 11 is used to delay 43 lines, and the corrected output is shown in Figure 3 (a).
) as shown in “...R185 , R187 , R18
9...'', FIFO memory 41, 4
2 data and combination, for example R185 and R187
Accordingly, R186', R187 and R189
R188' is generated respectively, and this is interpolated to generate "...R185, R186', R187, R18
8', R189, R190'...'' is output. By interpolating the line data thinned out in this way, the number of lines is doubled and restored to the state before thinning, and further R18
5.6 from the data of R186', R187
186.6...The remaining fraction is corrected using the moving average, and the output is delayed by 87.6 lines.
...R184.6, R185.6, R186.6,
R187.6, R188.6, R189.6...
” and B data B97, B98, B99, B100
... and match the timing.

Note that in FIG. 3(a), the number of lines is doubled by line interpolation, and then fractional correction is performed, but as shown in FIG. 3(b) and (c), the number of lines is thinned out one line at a time. It is also possible to obtain data that has also undergone fractional correction directly from the line data.

That is, FIG. 3(b) shows a case where 0≦n<1, where n is the fraction of the number of correction lines, and Ri and Ri
+2, the interpolated data on the side closer to Ri is {(2-n)Ri +nRi+2}/2Ri
The interpolated data on the side closer to +2 can be directly obtained by {(1-n)Ri + (1+n)Ri+2}/2. In addition, in the case of 1≦n<2, Fig. 3 (
As shown in c), when the fractional data is closer to Ri, it is {(3-n)Ri + (n-1)Ri+2}/2, and when the interpolated data is closer to Ri+2, it is {(2-n) Ri
+nRi+2 }/2.

Next, the gap correction shown in FIG. 4 will be explained with reference to the timing chart shown in FIG.

Scanning in the sub-scanning direction and scanning in the main scanning direction is performed in response to a trigger signal from the CPU, and data is captured. In the present invention, lines are thinned out and interpolated only when the magnification is 200% or more, and data is thinned out every other line as shown in FIG. Output. The output of the FIFO 11 is data in which odd lines are thinned out, for example, 0 line, 2 line, 4 line, 6 line, 8 line. FIF this output
Read it twice with O41, delay it by one line, fill in the thinned out line data with the same data, and write "00224".
Line data is output like "46688...". This data is further delayed by one line in the FIFO 42, and by selecting the selector 43, it is combined with the data on the 2nd line of the FIFO 11 and the data on the 0th line of the FIFO 41. For example, by selecting coefficients A and B, the "0'" line is created. generate data. Further, the selector 43 extracts data on the second line of the FIFO 41 and data on the 0th line of the FIFO 42, and uses the coefficients C and D to generate data on the "1'" line. After that, FIFO11 and FI
Select the data of FO41, FIFO41 and FIFO42 sequentially and select "0', 1', 2', 3', 4', 5',
Line data is interpolated and output as shown in "6', 7'...".

[0025] Also, if the magnification does not reach 200%,
As shown in FIG. 5(c), the output of FIFO 11 is not thinned out, so it becomes "0, 1, 2, 3, 4, 5, 6...", which is delayed by one line in FIFO 41, and then Delay by one line using FIFO42. In this case, each adjacent line data is obtained from FIFO 11 and FIFO 41, so there is no need to use the data from FIFO 42. For example, if there is a fraction such as 150.5% expansion rate, F
Using the data of IFO11 and FIFO41 respectively, "0', 1', 2', 3', 4', 5', 6', 7'...
..." and outputs by interpolation.

FIG. 6 is a diagram illustrating another embodiment of the present invention.

As shown in FIG. 6(a), a delay FIFO
The data delayed by a predetermined line in step 11 is further transferred to FIFO5.
0 to delay one line, output of FIFO11 and FIF
Inter-line data is generated by combining with the output of the FIFO 11, which is fed back and further delayed by one line, and combined with the output of the FIFO 11 to generate corrected line data.

As shown in FIG. 6(b), the input to the FIFO 11 is thinned out one line at a time, and the FIFO 11 contains "0,
2, 4, 6, 8...'' is written, and this data is read twice to generate "00224466..." as the output of the FIFO 11. This is delayed by one line in the FIFO 50, the data of the 2nd line of the output of the FIFO 11 and the data of the 0th line of the output of the FIFO 50 are used to generate the data of the "0'" line, which is fed back through the adder circuit 54, FIF again
Delay by one line at O50, combine this "0'" line data with the 2nd line data of the FIFO 11 output to generate "1'" line data, and repeat the same process thereafter. ``1', 2', 3', 4''
', 5', 6'...'' line data can be obtained as a correction output.

FIG. 7 is a diagram showing another embodiment of the present invention. In the above embodiment, gap correction was explained when scanning in the order of R, G, and B. However, if you try to read an image also during backscanning, the R signal and B signal
The signals will be swapped. Therefore, as shown in FIG. 7, switches 60 and 61 are used to exchange the R and B signals to perform thinning processing, and then the signals are input to the FIFO, and after interpolation processing is performed, they are further exchanged using switches 62 and 63, and the R, G,
By outputting B without changing the order, gap correction can be performed in the same way as in forward scanning. Note that the switch switching signal may be generated by a sub-scanning synchronization signal, or may be set and output from the CPU.

By the way, since the computation for interpolation processing is a kind of low-pass filter, the MTF deteriorates due to the computation. For example, the G sensitivity of a CCD sensor is close to the brightness sensitivity of the human eye, and the R, G, B color system is
* When converting to a color system, it directly affects L (lightness). Therefore, it is preferable to perform interpolation processing for R and B, centering on G, and not perform interpolation processing for G, thereby performing color shift correction with less deterioration in definition. In addition, if there is a large difference in the MTF of R, G, and B before interpolation processing (such as a sensor with a large pixel size for B to increase the sensitivity of B), interpolation processing centered on G may be performed as described above. If this is done, the difference in resolution between R, G, and B becomes too large, and for example, even if the document is black, it may be determined to be color. Therefore, R is centered around B, which has the lowest resolution.
, G to reduce the resolution of R, G, and B, so that the output ratios of R, G, and B are at the same level when reading black. As a result, R, G,
Interpolation processing with little difference in B resolution can be performed. In addition, since pictogram separation focuses on the contrast (brightness) of image data, the resolution of G is important, and it is appropriate to perform interpolation processing for R and B with G as the center. Now, to focus on the saturation of each pixel, R
, G, and B, it is suitable to perform interpolation processing on R and G, centering on B, which has a small difference in color shift between them. Therefore, it is desirable to switch the colors for interpolation processing depending on the purpose.

In the above explanation, the thinning and interpolation processing for the R signal has been explained, but when data is thinned out, the image quality inevitably deteriorates. Due to the sensitivity of the human eye, even if the accuracy of the B signal is slightly lower, it is not noticeable, so if you replace R and B and thin out the B signal, you can further prevent the image quality from deteriorating. can.

In the above embodiment, the thinning circuit 1 of FIG.
0 described the case where only line thinning is performed, but if the thinning circuit 10 cuts out the necessary portion of image data from the image data in one line of main scanning, the delay memory capacity for gap correction can be reduced. Is possible. By extracting the image data in one main scanning line in this way, there is no need to perform line thinning if the memory capacity is sufficient even without line thinning. Therefore, the CPU should make a judgment based on the area required for main scanning, sub-scanning magnification, and memory capacity, and perform sub-scanning line thinning and interpolation in addition to main-scanning image data extraction only if necessary. In this way, by combining main scanning image data extraction and line thinning,
Even in the case where the main scanning magnification is equal to the magnification and the sub-scanning magnification is enlarged as described in Section 0, line thinning can be performed without replacing the document.

[0033]

Effects of the Invention As described above, according to the present invention, 200%
When the magnification is above a certain level, such as when enlarging, line data is thinned out line by line and delayed, and the thinned out data is interpolated to match the timing of each data.
It is possible to perform gap correction by reducing the delay memory capacity, and since the sampling density increases when the magnification is higher than a predetermined magnification, even if data is thinned out, the image quality will not deteriorate significantly and the image quality will be maintained. It is possible to reduce the memory capacity while maintaining the memory capacity. In addition, by combining main scanning image data extraction and line thinning, it is possible to further reduce the delay memory capacity for gap correction. This can be done without replacing the original.

[Brief explanation of the drawing]

FIG. 1 is a diagram for explaining gap correction in the present invention.

FIG. 2 is a video signal processing circuit diagram to which gap correction is applied.

FIG. 3 is a diagram showing a specific correction method for gap correction.

FIG. 4 is a block configuration diagram of gap correction.

FIG. 5 is a diagram showing the principle of gap correction.

FIG. 6 is a diagram illustrating another embodiment of the present invention.

FIG. 7 is a diagram illustrating another embodiment of the present invention.

FIG. 8 is a diagram illustrating image reading by a reduction type line sensor.

FIG. 9 is a diagram illustrating conventional gap correction.

FIG. 10 is a diagram illustrating conventional gap correction.

[Explanation of symbols]

10... Thinning circuit, 12... Line interpolation circuit, 11, 41
,42,43...FIFO.

Claims (5)

[Claims] 1. A plurality of line sensors parallel to the main scanning direction that output different colors are arranged at predetermined intervals in the sub-scanning direction, and a delay means is provided for correcting the gap between each line sensor in the sub-scanning direction. In the color image reading device, the input side of the delay means is a thinning means for thinning out one line of data at least every predetermined line, and the output side of the delay means is a line data interpolation means for interpolating the thinned out line data. An image reading device characterized by being provided with. 2. The apparatus according to claim 1, further comprising: a first switching means for interchanging each signal provided at a stage before the thinning means; and a first switching means for switching each signal provided at a stage after the line data interpolation means. 1. An image reading apparatus characterized in that a second switching means is provided to switch the signals to be delayed during forward scanning and during back scanning. 3. The image reading apparatus according to claim 1, wherein line thinning and line interpolation are performed when the image reading magnification is equal to or higher than a predetermined value. 4. The image reading apparatus according to claim 1, wherein the thinning means extracts a necessary area from the output image of the line sensor. 5. The plurality of line sensors are R, G, B
The image reading device according to claim 1 or 2, wherein the image reading device is a line sensor, and a B or G line sensor is arranged in the last row during forward scanning, and the B or G line sensor performs delay processing of the read signals of other line sensors. .