JPH07306565A - Registration error correcting method - Google Patents

Abstract

PURPOSE:To make a registration error the smallest in limited smallest correction unit. CONSTITUTION:In order to correct Y with reference to K, -10/2<=K-Y<=10/2 is judged. Based on the judged result, Y=Y+ or -10, 0 is obtained and correction is performed. In order to correct M, (K-Y)/2 is added as information on previously corrected Y, -10/2>=(K-M)+(K--Y)/2<=10/2 is judged. Based on the judged result, M=M+ or -10, 0 is obtained and the correction is performed. Furthermore, in order to correct C, (K-Y)/2+(K-M)/2 is added as the information on the previously corrected Y and M, and -10/2>=(K-M)+(K-Y)/2+(K-M)/2<=10/2 is judged. Based on the judged result, C=C+ or -10, 0 is obtained and the correction is performed. By correcting YMC with reference to K, the registration error becomes small and color image quality is improved.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a multicolor image recording apparatus for obtaining a color image by carrying out multiple transfer of images of different colors onto recording paper or an intermediate transfer member to be conveyed, and more particularly to a resist for an image formed in each color. Of the registration error corrections, the present invention relates to a registration error correction method that makes maximum use of limited physical correction means.

[0002]

2. Description of the Related Art As an example of a registration error correction method applied to a color image recording apparatus, etc.
-271275. This is provided with detection means for detecting the passage of each color pattern image behind a plurality of image forming units, measures the passage interval time of each color, binarizes the electrical signal output from the detection means, and obtains the obtained position. This is a method for correcting registration based on information.
In this method, the reference clock is used for correction in the main scanning direction, and correction is performed in units of this clock. Here, it is disclosed that the correction accuracy is improved by increasing the frequency of the clock. Similarly, the sub-scanning direction, the magnification,
The correction of skew etc. is usually corrected in some standard unit. By the way, in the lateral direction correction, when the image enable timing is controlled from the reference clock number of times from the scan start timing, the correction is performed in a unit of one pixel. Further, when the scan start timing itself is corrected with the 1/4 reference clock, the correction becomes 1/4 pixel unit. In other words, by the correction means,
The amount of correction is limited.

[0003]

In the case of correction by the above method, when the registration shift of each color is shifted by an integral multiple of the reference clock (the minimum correction unit such as one clock is hereinafter referred to as the minimum correction unit), However, if the deviation is corrected, the registration deviation will be completely eliminated (no error) and it will not be a problem.
However, this is not usually the case, and an error of the worst reference clock weakness will remain if no measures are taken. Naturally, the larger the minimum correction unit, the greater the influence. Further, if the minimum correction unit is reduced in order to reduce the error, cost increase and design man-hour increase are inevitable.

FIG. 7A shows an example of an image position obtained by the detection system. Here, the deviation of Y with respect to K is −6, the deviation of M with respect to K is −4, the deviation of C with respect to K is 3, and K is the reference position of the image. In order to match this reference position, YMC has correction means. There is. Assuming that the minimum correction unit (minimum unit that can be physically corrected) is 10, when correcting YMC, first, in order to bring each color closest to the reference K, the difference between each color and K and the minimum correction unit are compared, and the correction is performed. Judge not to increase, decrease.

That is, in the algorithm shown in FIG. 8, it is first determined whether KY ≦ 10/2 is satisfied (S1
00) If it exceeds 5, it is increased as Y = Y + 10. If it is 5 or less, it is judged whether KY ≧ −10 / 2 is satisfied (S101). If not satisfied, Y = Y−1.
Decrease as 0. When KY is within the range of ± 10/2, Y is used as it is without correction. Thereafter, M and C are processed in the same manner. As a result, each color can be corrected as shown in FIG. However, with this method, a large error remains when considering the relative error of each color. In this example, the Y-M registration error is 8.

Therefore, an object of the present invention is to provide a registration error correction method capable of minimizing the error with a limited minimum correction unit. Another object of the present invention is to provide a registration error correction method capable of making an error appear small in human vision with a limited minimum correction unit.

[0007]

In order to achieve the above object, a registration error correction method according to the present invention as set forth in claim 1 is arranged such that one or more image forming units are arranged to detect registration. In a multicolor image recording apparatus equipped with a correction mechanism, a limited minimum correction unit is used, and correction is performed by taking into account the correction order from the position information of each color and the information to be corrected before, thereby making a relative deviation of each color. Is configured to be minimized.
Further, in the registration error correction method of the invention described in claim 2, in the invention described in claim 1, when the respective colors are corrected, the positional deviation of the pattern is utilized and the human visual sense is used to find the most misalignment of the respective colors. It is designed to be small. The registration error correction method according to the third aspect of the present invention is configured to use a neural network as a means for forming an input / output function that associates pattern position information with human vision.

[0008]

According to the present invention, a limited minimum correction unit is used, and the correction order is corrected based on the position information of each color and the information to be corrected before is corrected. That is, after Y is corrected with respect to K, M is optimized with respect to the corrected midpoint of KY. Similarly, in C, the registration error can be minimized by the limited correction means by optimizing the corrected midpoints of KY and KM.

[0009]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a schematic view of a multicolor image recording apparatus to which the present invention is applied. In FIG. 1, the image forming apparatus 3
Consists of K, Y, M, C, and the photoconductor and laser beam R
Any combination of OS and LED ROS may be used. K is, for example, a device that forms a black image, and Y is
For example, a device for forming a yellow image, M for a device for forming a magenta image, C for a device for forming a cyan image, for example, are arranged at substantially equal intervals. ing.

The transfer member 7 has a transparent belt structure for transferring an image formed by each of the K, Y, M, and C image forming apparatuses, and has a driving roller 9a and a driven roller 9b facing each other.
Supported by. The drive roller 9a is driven by a dedicated drive motor (not shown) having excellent constant speed, and the driven roller 9b is rotated by the drive force thereof being transmitted by the transfer belt 7. Further, the transfer belt 7 has a function of carrying the transfer paper, and the transfer paper is carried from right to left in FIG. 1 similarly to the belt rotation direction. At this time, a suction corotron (not shown) is provided to suck the paper onto the belt member. On the downstream side of the image forming apparatus C, a CCD 1 as an image detecting means for reading the image on the transfer belt 7 formed by each image forming apparatus 3
, One at each end of the image area, two in total.

The mounting positional relationship between the CCD 1 mounted on the assay body 10 and its drive circuit and optical system such as the SELFOC lens is designed so that highly accurate positioning can be easily realized. The light source 2 produces background light necessary for the CCD 1 to detect an image on the transfer belt 7, and an LED, a halogen lamp or the like is used. CCD1
Any light source may be used as long as it can secure a sufficient amount of light. Further, the light source 2 secures an optimum image receiving state against deterioration of light amount of the light source itself, deterioration of transmittance of the transfer belt 7, deterioration of sensitivity of CCD, deterioration of transmittance due to contamination of optical system, and environmental change represented by temperature. In order to do so, the amount of light can be freely changed.

Next, the normal image forming mode will be described. When the leading edge of the sheet conveyed by the transfer belt 7 reaches the transfer point directly below the image forming apparatus K, the leading edge of the image formed by the image forming apparatus K reaches the transfer point directly below the image forming apparatus K. The paper feeding timing and the image writing timing are determined so that there is no deviation between the image and the paper in the sub-scanning direction (paper carrying direction). The sheet that has reached the transfer point is transferred with the image formed by the image forming apparatus K by a transfer corotron (not shown) or the like, and further reaches the transfer point directly below the image forming apparatus Y. The sheet that has reached the transfer point directly below the image forming apparatus Y is transferred in the same manner as it was transferred by the image forming apparatus K, and is transferred so as to be superimposed on the image transferred by the image forming apparatus K. Hereinafter, similar overwriting is performed in the image forming apparatuses M and C. After the transfer is completed, the paper is further conveyed by a belt, and when it reaches the vicinity of the driving roll 9b, it is separated from the transfer belt by a corotron, a stripper or the like for separating the paper from the transfer belt, although not shown. After that, it is fixed by a fixing device (not shown) or the like, and is discharged to the outside of the machine.

The interface board 4 comprises a K interface board, a Y interface board, an M interface board and a C interface board, and sends an image signal to the ROS of each image forming apparatus 3. The registration deviation correction board 5 is a registration deviation correction system for instructing the side registration correction and the lead registration correction to the interface board 4, and the memory / image processing board 1
The reference numeral 1 collectively handles the memory and image processing. The control board 6 manages movements of all of the boards 4, 5 and 11 and the entire apparatus.

Next, the operation of the registration shift correction system will be described. The registration shift correction is executed by entering a dedicated correction cycle preset in the apparatus. When entering the correction cycle, the control board 6 issues a command to each board,
The interface board 4 plays a role of a pattern generator that outputs a pattern for measuring registration error. Further, the registration deviation correction board 5 is transmitted from the interface board 4 to the image forming apparatus 3 and prepares to sample the registration deviation measurement pattern output from the image forming apparatus 3.

When the correction cycle starts, first, the registration deviation measurement pattern output from the interface substrate K by the image forming apparatus K is transferred onto the transfer belt 7 as a record 8K. From the interface board K to the image forming apparatus K
After the registration misalignment measurement pattern is transmitted, the interface board Y and the image forming apparatus Y continue with a time difference according to the distance between the transfer points of the image forming apparatuses K and Y.
A pattern for measuring registration error is transmitted. The pattern for measuring the registration deviation formed by the image forming apparatus Y is transferred onto the transfer belt 7 as a record 8Y. Record at this time 8Y
The pattern No. is a pattern in which a pattern for measuring the registration shift is overwritten on the already transferred record 8K. Similarly, the patterns for measuring the registration deviation formed by all the image forming apparatuses are overwritten, and the patterns are completed as the recording 8C on the transfer belt 7.
The device is made in advance so that the completed recording pattern 8C for measuring the registration shift reaches directly below the CCD 1 by the transfer belt.

The registration deviation correction board 5 monitors at least one of the pattern output timings for measuring the registration deviation of the interface board 4 by sampling the image data from the CCD 1. From the output timing of at least one interface board, the time when the pattern for measuring the registration shift reaches just below the CCD 1 is determined. That is, the sampling start timing and the sampling end which are necessary and sufficient for sampling the registration deviation measurement pattern from the pitch between the image forming apparatus 3 for forming the registration deviation measurement pattern output from the interface board and the CCD 1. You can figure out the timing.

At the sampling start timing, the registration shift correction substrate 5 starts to capture the image signal from the CCD 1 into the high speed memory, and at the sampling end timing, the capture is finished. At the same time as the end of the acquisition, before the end of the sampling of the pattern for measuring the next registration shift, the image position is determined from the acquired data by, for example, the center of gravity method, and used as the main image position address, for example. Store in memory. By repeating this operation several times, several fixed image position addresses can be obtained for each image forming apparatus. Here, in order to improve the accuracy of the fixed image position address, these several fixed image position addresses may be averaged for each image forming apparatus.

Next, a correction value for correcting the registration deviation between the image forming apparatuses is calculated for each of several registration deviation correction parameters by an algorithm determined in advance from the image position address determined for each image forming apparatus. Moreover, it is calculated for each image forming apparatus. By setting the calculated correction values from the registration shift correction substrate 5 to each image forming apparatus 3, each interface substrate 4, etc., it is possible to provide image quality without registration shift.

There are various types of misregistration patterns due to the process direction, the lateral direction, the magnification, and the skew, which are usually caused by a combination of these. Various deviations are read by the measurement pattern shown in FIG. 2 and which pattern is measured. Then, correction is made according to the measured pattern. Here, in order to facilitate understanding as an example, a case where only a deviation in the process direction occurs will be described with reference to FIG. In FIG. 2, l ₂ is a preset value (expected value) from the reference position, l _2K is the deviation of K obtained from the measurement, and similarly l _2Y is the deviation of Y. When Y is added to K, the image forming timing of Y must be corrected by l _2K −L _2Y . For example, if 1 _2K −L _{2Y corresponds} to one pixel, the correction parameter in the process direction is 1. Similarly, each correction in the lateral direction, magnification, and skew is calculated to determine all correction parameters.

When the calculated correction value is set in the interface board, if Y is displaced from K by one pixel in the lateral direction in the above example, for example, both KY and image enable are enabled at the 100th clock from the scan start timing. If the timing is used, Y is delayed by one pixel, that is, correction is performed to set the image enable timing at the 99th clock from the Y scan start timing. As one of the registration error correction algorithms, it is important how to reduce the registration error of each color by using the image position address. These will be described in detail below.

(Embodiment 1) FIG. 3 shows a correction example in which the corrected color information is added before the color to be corrected. FIG. 4 shows an algorithm for correcting the color to be corrected by adding the corrected color information. In FIG. 4, first, in the correction of Y with respect to K, K
It is determined whether −Y ≦ 10/2 is satisfied (S200), 5
When it exceeds, increase as Y = Y + 10, and again 5
In the following cases, it is determined whether or not K−Y ≧ −10 / 2 is satisfied (S201), and when not satisfied, Y = Y−10 is reduced. When KY is within the range of ± 10/2, Y is used as it is without correction. Subsequently, in the correction of M, it is determined whether (K−M) + (K−Y) / 2 ≦ 10/2 is satisfied (S202). If not satisfied, M = M + 10 is increased, and if satisfied. Is (K−M) + (K
−Y) / 2 ≦ −10 / 2 is determined (S20)
3) If not satisfied, reduce as M = M-10. When (K−M) + (K−Y) / 2 is within the range of ± 10/2, M is used without correction. Furthermore, in the correction of C, (K−M) + (K−Y) / 2 + (K−M)
/ 2 ≦ 10/2 is satisfied (S204), and when not satisfied, C = C + 10 is increased, and when satisfied, (KM) + (KY) / 2 + (K- M)
It is determined whether or not 1/2 ≦ −10 / 2 is satisfied (S205),
If not satisfied, reduce as C = C-10. (K
-M) + (KY) / 2 + (KM) / 2 is ± 10/2
If it is within the range, the C is used as it is without correction.

Here, when correction is performed using each color position shown in FIG. 3 as an example, Y has a KY value of -3, and -10/2.
Since ≦ K−Y ≦ 10/2 is satisfied, it is within the range where no correction is made. Therefore, Y is not corrected. M has a value of (K−M) + (K−Y) / 2 of −6.5 (= −5−).
Since 1.5), (K−M) + (K−Y) ≦ −1
Since 0/2 is satisfied, it is within the range where no correction is made. Therefore, M is not corrected. Furthermore, C is (K-C) +
Since the value of (K−Y) / 2 + (K−M) / 2 is 0,
(K−C) + (K−Y) / 2 + (K−M) / 2 ≦ −10
/ 2 is not satisfied, so it is reduced to C = C-10. That is, 3-10 = -7, and C corrects K by reducing the value 7. FIG. 3B shows the relative error of each color after correction. The relative error of each color is 8 in FIG. 7B, whereas in this example, the K-C registration error is 7, which is small.

Further, when the order of the colors to be corrected is changed and the correction is performed, the relative error between the respective colors is further reduced as shown in FIG. In this example, the KY registration error is 6
Is. Therefore, if all combinations of the correction order are performed by this algorithm, (YMC, YCM, MYC, MC
(Y, CMY, CYM) It is possible to obtain the setting with the smallest relative error of each color.

(Embodiment 2) The embodiment 1 makes it possible to obtain the setting with the smallest relative error of each color, but in the case of color, there is a visual problem that cannot be expressed only by numerical values. For example, it is a well-known fact that Y has a characteristic in which misregistration is relatively unnoticeable. In addition, it is difficult to perform the optimum correction due to the characteristics of the output device and individual differences in vision. Therefore, in order to optimize these, the present invention experimentally finds the relationship between the color information, the size of the deviation, and the badness / goodness of vision at the time of correction, and determines the optimum correction using a lookup table or the like.

FIG. 6 shows the misregistration amount nY for each color with respect to K,
This is an example in which the optimum correction in nM and nC was obtained by experiment. There are 27 combinations of three types of correction methods for any YMC registration deviation value (below the minimum correction unit: mathematically correct above the minimum correction unit). nY is Y
Of the registration deviation amount with respect to K, nM with respect to the registration deviation amount with respect to K of n, nC with respect to the registration deviation amount with respect to K of C, Y is corrected by +1 minimum correction unit, and M is corrected by -1 minimum correction unit. C indicates that the unregistered one has the least noticeable misregistration. Then, among these, the best combination of the correction results is experimentally determined as the optimum value. If this combination is enormous and the number of man-hours for experimentation and the look-up table become large, the above-mentioned relation is used for the typical data by using the neural network technology.
For example, 1000 types are used as teacher data, and the relational expression is made into a function to determine a correction value.

[0026]

As described above, according to the present invention, the best registration correction can be performed with a limited minimum correction unit, and not only the improvement of color image quality can be expected, but also the minimum correction unit can be increased and the cost, This can greatly contribute to the ease of design.

[Brief description of drawings]

FIG. 1 is a system configuration diagram of an image forming apparatus to which a registration error correction method of the present invention is applied.

FIG. 2 is a diagram showing an example of a measurement pattern.

FIG. 3 is a diagram showing an example of correction by adding the corrected color information before the color to be corrected.

FIG. 4 is an algorithm for correcting by adding the corrected color information before the color to be corrected.

FIG. 5 is a diagram showing an example of performing correction by changing the order of colors to be corrected in the correction example of FIG.

FIG. 6 is a diagram showing a YMC correction relationship that is the most inconspicuous when correction is made with respect to an arbitrary YMC registration deviation value.

FIG. 7 is an explanatory diagram of conventional registration shift correction.

FIG. 8 is a diagram showing an example of a conventional registration shift correction algorithm.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... CCD (image detection means), 2 ... Light source, 3 ... Image forming device, 4 ... Interface board, 5 ... Registration misregistration correction board, 6 ... Control board, 7 ... Transfer belt (transfer member), 8 ... On transfer belt Record of 10 ... Assay body, 1
1 ... Memory / image processing board

Claims (3)

[Claims] 1. A multicolor image recording apparatus having one or a plurality of image forming units and a registration detection and correction mechanism, in which a limited minimum correction unit is used and correction order is performed from position information of each color. The registration error correction method is characterized in that the correction is performed so as to minimize the relative deviation of each color by taking into account the information previously corrected. 2. The registration error correction method according to claim 1, wherein when the respective colors are corrected, it is considered that the deviation between the respective colors is minimized by utilizing the positional information of the pattern and the human visual sense. 3. The registration error correction method according to claim 2, wherein a neural network is used as a means for forming an input / output function that associates pattern position information with human vision.