JP2001008043A - Signal processor and signal processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a signal processor that can execute both conversion processing and error propagation processing of a color signal at a high speed with a small scale circuit configuration. SOLUTION: This signal processor 11 is provided with a signal correction circuit 13 that corrects an input signal C in eight bits per pixel sequentially received for every line by using a quantization error of peripheral pixels caused by quantization processing so far, a quantization circuit 24 that converts an input signal P after the circuit by the circuit 13 into an output signal Z in four bits per pixel, an error calculation circuit 15 that calculates a quantization error E by the quantization circuit 14, and an error feedback circuit 16 that feeds back the quantization error E to the signal correction circuit 13. Furthermore, a conversion table used for the quantization processing by the quantization circuit 14 and for the error calculation processing by the error calculation circuit 15 is integrated in a memory (a memory accessible from the quantization circuit 14).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a signal processing device and a signal processing method for converting signal characteristics of an image signal at high speed.

[0002]

2. Description of the Related Art In order to represent a color image, color signals of a plurality of colors are required. RGB (red-green-blue) signals, CMYK (cyan, magenta, yellow, black) signals and the like are typical examples. For example, in a CRT (cathode ray tube) color display, an RGB signal is often used as a color signal. In color printing of a printer, a CMYK signal is often used as a color signal.

When color reproduction is performed by a device using different color signals, such as printing a color display image on a CRT from a printer, color signal conversion processing is required. In color image processing, since one pixel is represented by a plurality of color signals, a color is represented as coordinates in a color space using each color component as a coordinate axis. Therefore, the conversion of the color signal is
This corresponds to mapping a coordinate point on a certain color space on a different color space. Such mapping on a different color space is often performed using a conversion table that associates coordinates on an input color space with coordinates on an output color space.

However, a memory for storing a conversion table for converting RGB signals (each color signal is 8-bit data) into CMY signals (3 bytes) requires a capacity of about 224 × 3 = about 50 megabytes. Therefore, as a technique for reducing the memory capacity, for example, US Pat. No. 4,783,722 has been proposed. In this technique, the coordinates of representative grid points obtained by coarsely quantizing the color space and interpolation coefficients for interpolating between the representative grid points are stored in a conversion table, and the input between the representative grid points is converted by interpolation. Is what you do. By preparing a conversion table and interpolation means for each color component, high-speed and high-precision color signal conversion processing is realized while reducing the memory capacity. Note that the conversion table creation method and data structure used in this technology are described in ICC (International
There is a standard called an ICC profile defined by the Color Consortium.

[0005] The image output apparatus has various restrictions on the number of expressed gradations per pixel. For example, there are a printer capable of expressing only two levels and a display capable of expressing 256 levels. In such an image output device, a gradation conversion method such as an error diffusion method or a dither method is used to express the same number of gradations as the input signal. For example, an ink jet printer or the like realizes smoother multi-level gradation reproduction by employing an error diffusion method that increases the number of reproduction gradations by arranging a plurality of pixels within a small area. Such a general error diffusion method is described in detail in “Digital Signal Processing of Images” (published by Nikkan Kogyo Shimbun).

[0006]

However, in order to output a display image on a CRT from the above-described conventional inkjet printer, it is necessary to perform color signal conversion processing and gradation conversion processing for each pixel. Therefore, a heavy load is applied to the signal processing device, and the processing speed of the signal processing device is reduced. Such a tendency becomes more remarkable as the image quality is improved by high-density and high gradation recording.

Further, since a signal processing device is required for each of the color signal conversion process and the error diffusion process,
There is also a problem that the configuration of the system becomes complicated.

Therefore, the first object of the present invention is to provide a signal processing apparatus capable of executing both the color signal conversion processing and the gradation conversion processing at a high speed with a small-scale circuit configuration.
To the purpose. It is a second object of the present invention to provide a signal processing method capable of executing both color signal conversion processing and gradation conversion processing at high speed.

[0009]

According to the present invention, there is provided a memory for storing a conversion table in which two types of signals represented as coordinates of mutually different coordinate systems are associated with each pixel, and an input signal of each pixel. A quantization circuit for sequentially converting a signal corresponding to the input signal in the conversion table, and an error calculation for sequentially calculating an error of the signal conversion each time the quantization circuit performs signal conversion for one pixel. And a signal correction circuit for diffusing the error calculated by the error calculation circuit into an input signal before signal conversion by the signal conversion circuit.

[0010] As described above, the quantization means converts the overlapping processing between the color conversion processing and the error diffusion processing, which were conventionally performed completely separately by separate devices, during the error diffusion processing repetition processing. Since the processing is performed based on the table, the circuit scale can be reduced and the processing speed can be increased. That is, during the repetitive processing of the error diffusion circuit, the quantization circuit converts the correction signal into an output signal having a different property from the correction signal (an output signal represented as coordinates in a coordinate system different from the correction signal).
As a result, the conventional color conversion processing is incorporated into the error diffusion processing to reduce the circuit scale and speed up the processing.

According to the present invention, in the quantization step, the input signal of each pixel is converted into the input signal based on a conversion table in which two types of signals represented as coordinates of mutually different coordinate systems are associated with each pixel. In the error calculation step, every time signal conversion is performed for one pixel, an error of the signal conversion is sequentially calculated, and in the signal correction step, the signal by the quantization step is converted. The error calculated in the error calculating step is diffused to the input signal before conversion.

According to such a method, it goes without saying that the processing can be speeded up as in the case of the signal processing device.

[0013]

Embodiments of the present invention will be described below with reference to the accompanying drawings. First, the basic configuration of the signal processing device according to the present embodiment will be described by taking, as an example, signal processing in units of pixels for a monochrome still image.

As shown in FIG. 25, the present signal processing device 11
Can be mounted on the information processing control device 51 to which an image input device such as a scanner and an image output device such as a printer are connected. The connection method is shown in FIG.
The configuration shown in any of (2) may be used. Further, the configurations of the input device and the output device are not limited.

As shown in FIG. 1, the signal processing device 11 is a monochrome signal (input signal C) of 8 bits (256 levels) per pixel which is sequentially input for each line from an input device (not shown). ), And converts the input signal P (hereinafter, referred to as a correction signal P) corrected by the signal correction circuit 13 into a monochrome signal (output signal Z) of 4 bits (16 levels) per pixel. Error calculation circuit 1 for calculating quantization error E by quantization circuit 14 and quantization section 14
5, an error feedback circuit 16 for feeding back the quantization error E calculated by the error calculation circuit 15 to the signal correction circuit 13. Details of these units 13, 14, 15, 17 are as follows.

The signal correction circuit 13 uses the following equation (1)
The input signal C sequentially input from the input device is corrected,
The correction signal P is sequentially output to the quantization circuit 14.

P = C + Σ (Ei · Fi) (1) where C is the signal level of the pixel of interest to be processed,
Specifically, it is the density value of the pixel of interest, and Ei is the signal level (initial value 0) of the error signal for the reference pixel read from the internal line memory of the error feedback circuit 16 described later.
Where Fi is a weight coefficient determined according to the positional relationship between the target pixel and the reference pixel on the image. The reference image referred to here is a plurality of peripheral pixels Y0 to Y having a predetermined positional relationship with the target pixel X on the image as shown in FIG.
4 Note that, in relation to the image quality setting of the output image, it is desirable that the number of reference pixels and the positional relationship between the reference pixels and the target pixel can be adjusted in units of screens together with the value of the weighting factor Fi. Such a signal correction circuit 1
Reference numeral 3 denotes a weighting circuit for calculating a product of the error signals E0 to E4 of the reference pixels Y0 to Y4 and the weighting coefficients F0 to F4, and an adder for adding the output signal of the weighting circuit and the input signal C. Can be.

In the built-in memory of the quantization circuit 14, information representing the gamma characteristic of the printer, specifically, a non-linear gamma characteristic representing the correspondence between the input signal C and the output signal Z as shown in FIG. Curves are stored. The possible signal levels of the input signal C and the output signal Z are not limited. Since it is difficult to replace the non-linear characteristic with an arithmetic expression, it is necessary to create this conversion table while experimentally confirming the tone reproduction characteristics of the printer for the input signal C. Alternatively, the color correction method (Color Syn
c or ICM), ICC (Interna
device-specific conversion data (ICC profile) set based on the data format defined by the National Color Consortium.

In the case of a monochrome image represented by 8 bits,
Since there are only 256 types of input signals, all such characteristic curves can be realized as a table. However, in the case of a color image, for example, 8-bit RGB
Since the combination of signals is used, the input signals are of the 256th power type, and there are many problems in terms of cost, device scale, and the like when the device is configured as a table. An object of the present invention is to perform high-speed conversion processing of such a multidimensional signal based on an arbitrary characteristic curve.

The gamma characteristic curve shown here is step-shaped because the output signal Z has 16 levels and the input signal has 256 levels. The gamma characteristic curve becomes smoother as the number of divisions increases. For example, in the case of a gamma characteristic curve of a printer, it is often sufficient if each coordinate axis is divided into about 8 to 16 steps. .

The quantization circuit 14 sequentially gamma-converts the correction signal P from 0 to 255 into an output signal Z from 0 to 16 using such a conversion table, and outputs this to the output device as an output signal Z. Output sequentially. For example, when the signal level of the input signal C of the target pixel is 86, the input signal C has a quantization step width of 80 to 9 as shown in FIG.
6 is converted to an output signal Z of signal level 2 corresponding to 6.
As described above, the gamma correction according to the present embodiment is a process of converting the input signal P into the N representative levels set in the conversion table, and hence is hereinafter also referred to as quantization.

The quantizing circuit 14 includes a signal correcting circuit 13
A memory address is generated by cutting out the upper 4 bits (when the signal level is divided into 16) of the correction signal P from the memory, and accesses a memory or the like that stores the correspondence between the input and output signals of the above-mentioned conversion table (see FIG. 3) I do.

According to such a quantization circuit 14, for example, any correction signal P belonging to a quantization step width of 80 to 90 is converted into an output signal Z of the same signal level 2. Therefore, this quantization step width 80 to
The difference between the minimum input signal (= 80) belonging to 90 and the correction signal (= 86) of the target pixel can be considered as the quantization error E (see FIG. 4). Therefore, the error calculation circuit 15
Calculates the quantization error E due to the quantization by the following equation (2) every time the quantization is performed by the quantization circuit 14, and outputs the quantization error E to the error feedback circuit 16.

E = P−f (Z) (2) where f (Z) is an intersection signal of the quantization step width to which the input signal to the quantization circuit 14 belongs.

In the conversion table of the quantization circuit 14, the correction signal P is used as a memory address and the intersection signal f together with the output signal Z.
If (Z) is read, the memory output f
(Z) and only the subtractor that calculates the difference between the correction signal P and f
(Z) A calculation circuit may be configured. In this way, the conversion table can be shared by the quantization circuit 14 and the error calculation circuit 15, so that the circuit scale can be reduced and the output signal Z and the intersection signal f (Z) can be calculated simultaneously. Processing can be speeded up. The data stored in the memory may be compressed to reduce the data amount. The error feedback circuit 16 writes the quantization error into a plurality of line memories, reads the quantization error of the reference pixel from the error memory, and feeds it back to the signal correction circuit 13 as an error signal. More specifically, the error feedback circuit 16 includes a line memory for a plurality of lines and a delay circuit having a flip-flop circuit that performs a delay in pixel units. If the control timing is obtained from the counter circuit synchronized with the scan input operation of the image signal, this configuration allows the error signals E0 to E4 of the reference pixels Y0 to Y4 having a predetermined positional relationship (see FIG. 2) with the target pixel X to be obtained. Can be taken out.

According to the signal processing apparatus 11, the processing which has been overlapped between the color conversion processing and the error diffusion processing, which has been performed completely separately by the conventional apparatuses, is now repeated during the error diffusion processing. Since the quantization circuit is configured to execute based on the conversion table, the circuit scale can be reduced and the processing speed can be increased. That is, in the signal processing device 11, during the repetitive processing of the error diffusion processing, the quantization circuit
By converting the correction signal into an output signal having a different property (an output signal represented as a coordinate in a coordinate system different from that of the error signal and the correction signal), the conventional color conversion processing is performed by an error diffusion processing. To reduce the circuit scale and speed up processing.

A method of using these numerical data will be described with reference to FIG. Here, the lattice block through which the characteristic curve passes is shown in an enlarged manner. The characteristic curve corresponding to the input signal C is passing through the block 1, and the output signal Z corresponding to the bottom of the block 1 is the conversion result. Here, the coordinates of the intersection of the input signal C and the characteristic curve are not located at the bottom of the block 1. In order to calculate this deviation as an error signal E in the input signal coordinate system, a signal f (Z) at the intersection of the characteristic curve and the base of the block is prepared. Then, the error signal E = the input signal C-f (Z). Here, in the same block, since the input signal C may be larger or smaller than f (Z), the error signal E has a sign of ±.

The error signal E is reflected in the subsequent pixel signal processing, thereby realizing the preservation of the input signal level of the entire screen. For example, a correction signal P obtained by adding the error signal E to the input signal C of the next pixel
= C + E. By performing such correction processing, in the description of the present invention, the input signal C and the correction signal P can be treated as the same signal processing target. In this way, signal conversion based on an arbitrary characteristic curve is realized as an average signal level of a plurality of pixels.

Further, since the quantization error generated in the quantization is weighted and assigned to neighboring pixels to be signal-converted from now on, the quantization error as a whole is suppressed. More excellent gradation reproduction can be realized quickly with a smaller circuit configuration than in the case of performing. That is, even if a coarse conversion table suitable for a color printer or the like in which the number of reproduced gradations per pixel is limited to at most about 16 levels is used, interpolation processing requiring a long processing time and a large-scale arithmetic circuit is required. , It is possible to obtain an excellent gradation reproduction asymptotic to a gamma characteristic curve as a set of pixels. For example, as shown in FIG. 5, when the input signal (= 86) is continuously input from the input device, the input signal (= 86) of the second pixel is replaced by the input signal (= 86) of the first pixel. = 86), and a quantization error E1 (= 6) caused by the quantization of the second pixel is added to the input signal (= 86) of the third pixel. Quantization error E2 caused by
(= 12) is added. The correction signal of the third pixel (=
98) is converted into an output signal Z (= 3) of a quantization step one level higher than the second pixel, and the quantization error E3 is 98−96 = 2. As described above, by sequentially reflecting the quantization error generated in the quantization on the quantization of the input signal of the next pixel, as a result, two types of signals Z = 2 3 are mixed at 5: 3. With this ratio of 5: 3, two types of output signals Z =
Since the input signal C (= 86) is obtained by averaging the intersection signals C = 80 and 96 corresponding to 2 and 3, the gradation of the input signal C (= 86) is reproduced by a set of eight pixels. It is clear that.

The calculation processing of the quantization error described in the present embodiment is an example, and the calculation processing of the quantization error by the quantization circuit of the present signal processing apparatus is not necessarily limited to this. For example, if the reference pixels are limited to adjacent pixels on the same line, the line memory of the error feedback circuit 16 can be omitted.

In the present invention, as shown in FIG. 6, N times of signal processing are repeatedly performed on the same pixel, and an average value of the processing results of N times of loop processing can be used as an output signal.
The averaging circuit 20 in FIG.
, And after adding N times, an averaged output signal is output. The N times loop processing for the same pixel is
It can be replaced with signal processing for N pixels at the same signal level. FIG. 7 shows an operation of performing N = 4 repetitive calculations on two input pixels.
The operation content of each loop is the same as the processing shown in FIG. The error signal generated at the time of the N calculations is passed to the next pixel calculation, so that the signal level of the entire image can be preserved.
The output signal converges as shown in FIG. 8 by performing the averaged output by regarding the signal processing shown in FIG. 5 as N-times loop processing of a single pixel. Here, even if the output signal is not sufficiently converged, the error can be almost canceled as the whole image by reflecting the error to the next pixel.
Therefore, as shown in FIG. 7, even if the calculation is performed using the predetermined number of repetitions N = 4, the signal level of the entire image is preserved. By such N-time loop processing and averaging output, for example, the number of gradations of the input signal and the output signal can be set to the same 8 bits. The output signal can be used as print data by performing signal processing such as a conventional dither method or halftone method.

When the quantization number of the output signal, which is different from the quantization number of the error diffusion process, is set, the set number of quantization processes may be performed on the averaged output signal. FIG. 9 shows an example in which the averaging circuit includes an adding circuit and a dividing circuit. The averaging circuit 20 can be configured by a circuit for adding N times and dividing the addition result by N.
For example, if a power of 2 such as N = 4 is set, the division circuit can be replaced by a bit shift, and thus can be basically configured only by an addition circuit. The reset signal is a signal for erasing the operation result of the loop N times, and resets each time the target pixel of the signal processing is updated. The operation of these circuits can be controlled by a sequence circuit. As shown in FIG. 9, an output signal is obtained by connecting an output quantization circuit 21 to the output stage of the averaging circuit 20 described above. The error generated by the output quantization circuit 21 can be used as a feedback signal together with the quantization error of the error diffusion processing. By using such a signal processing means, an arbitrary output gradation number (bit number) can be set. In this manner, color signal conversion and gradation conversion can be performed with an input signal of 8 bits and an output signal of an arbitrary number of gradations. The present invention can be used without depending on the characteristics of the input device and the output device.

The functions of the signal processing device 11 described above can also be realized by software, as shown in FIG. That is, as described below, the processor can execute the same signal processing as that of the signal processing device 11 using the software and the conversion table stored in the memory. First, the processor initializes the error memory (S110). Thereafter, the processor performs the following S111 to S111 for all pixels on the screen, respectively.
The processing up to 116 is executed. When the input of the input signal C of one pixel is received (S111), the quantization error Ei for the reference pixel is read from the error memory, and the aforementioned equation (1) is obtained.
To correct the input signal C (S112). Then, the output signal Z and the intersection signal f (Z) associated with the correction signal P obtained by the correction are read from the conversion table (S113). Then, the output signal Z is output, and the quantization error E is calculated by the above equation (2) (S114). Then, the quantization error E calculated at this time is written in the error memory so as to be used for correcting the input signal of the pixel thereafter (S115). Next, when the loop process is executed N times for the input pixel, the number of repetitions is measured (S116). If the process has been completed N times, an averaged signal is output (S117).
Returning to step 2, the operation is repeatedly executed. Then, it is determined whether or not the target pixel is the last pixel on the screen. If the target pixel is the last pixel, the process ends (S118). Otherwise, the processes from S111 onward are repeated. Execute. As described above, even when the signal processing of the present signal processing apparatus is implemented by software, the overlapping processing between the color conversion processing and the error calculation processing, which have been performed completely separately in the past, is reduced to an error. Since it is executed based on the conversion table during the repetition processing of the diffusion processing, it goes without saying that the processing can be sped up.

In the normal signal processing, the signal processing according to the present embodiment needs to be terminated in the same way as the termination processing is performed on the left, right, upper, and lower edges of an image.
Any termination processing method may be adopted.

In the above, an example of applying the present invention to signal processing for a black-and-white still image has been described. However, the present invention can also be applied to signal processing for a color still image. Hereinafter, (1) an example of application to signal conversion processing from an RGB signal to a CMY signal will be described with reference to the drawings.
Application to signal conversion processing from GB signal to multi-color signal (CMYK signal), (3) Application to signal conversion processing from RGB signal to palette color, (4) Signal conversion processing from RGB signal to black and white signal And the like can be realized with the same device configuration. It should be noted that these application examples are merely examples, and the present invention does not prevent an input signal from being another color signal in the field of color optics such as LAB, LUV, and XYZ. . It does not limit the number of signal types of the input signal and the output signal.

(1) Example of application to signal conversion processing from RGB signals to CMY signals The colors of pixels formed by three types of color signals are represented by coordinates in a three-dimensional color space RGB as shown in FIG. Is done.
If the input signal is 8 bits, the coordinate point of this color space is 2
The number is 563. Since the relationship between the color signal level and the print density in the case where such a color signal represented as coordinates in the three-dimensional color space RGB is captured by a scanner and output from a color printer depends on the characteristics of both devices, Does not become linear, but exhibits three-dimensional nonlinear characteristics. In order to represent such characteristics, when a conversion table that associates the output signal Z with all coordinates in the three-dimensional color space RGB of FIG. 11 is created,
The capacity of the memory for storing this increases. Therefore, in the present embodiment, in order to reduce the memory capacity of the quantization circuit 14, the three-dimensional color space RGB is coarsely quantized to about 16 levels, and the three colors of the input signal corresponding to the coordinates of the representative lattice point are obtained. A conversion table in which a combination of components (R, G, B) and a corresponding combination of three color components (C, M, Y) of an output signal are created, and this conversion table is stored in a quantization circuit. 14 is stored in a memory that can be accessed. Then, an error diffusion process is performed so that the signal levels of the input signals (R, G, B) are preserved.

In this case, the signal processing device 11
As shown in the following, for each color component R, G, B of the input signal,
In each case, it is necessary to mount the signal correction circuit 13 described above. Unlike the case of the above-described signal processing of a black-and-white image, the quantization circuit 14 uses a memory address obtained by combining the three color components (R, G, B) of the correction signal from the signal correction circuit 13 with a built-in memory. The combination (C, M, Y) of the three color components of the output signal is read from the conversion table of the memory,
The error calculation circuit 15 calculates a quantization error for each color component, and the error feedback circuit 16 feeds back to the signal correction circuit 13 for each color component.

The conversion table as shown in FIG. 11 can be constituted by a semiconductor memory as shown in FIG. The memory can realize high-speed access by being mounted on the same chip as a circuit that executes signal processing, for example. Similarly, a memory for storing image data can be mounted on the same chip. As an input / output signal of the memory, as shown in FIG. 14, a combination of upper bits of the input signal is set as an address signal, and a conversion result is set as memory read data. Also, the intersection signal f (Z) shown in FIG. 3 is stored and stored together so that reading can be performed simultaneously.

According to the signal processing device 11 having such a configuration, even if a rough conversion table suitable for a color printer or the like in which the number of reproduced gradations per pixel is limited to at most about 16 levels is used, considerable processing is required. As in the case of the above-described signal conversion of a monochrome signal, a set of pixels is asymptotically close to a gamma characteristic curve, without performing interpolation between representative grid points requiring time and a large-scale arithmetic circuit. Tones can be reproduced. That is, it is possible to reduce the circuit scale and speed up the processing without reducing the tone reproducibility.

In this embodiment, when the conversion table is created, the three-dimensional color space RGB is roughly quantized equally, but it is not always necessary to do so. For example, human skin
If a region corresponding to a color conspicuous on the screen is subdivided more than other regions, tone reproducibility is further improved. That is, when creating the conversion table, the three-dimensional color space R is adjusted so that the visual error in the entire reproduced color is suppressed to a certain level or less.
It is desirable to divide the GB, and as a result, the division of the three-dimensional color space RGB will be uneven.

In order to further improve the gradation reproducibility, a separately prepared conversion table is registered in each quantization circuit 14 for the subdivision area in the three-dimensional color space RGB, 3D color space RG based on the correction signal
It is determined whether it is included in any of the subdivided regions on B or a plurality of subdivided regions. If the conversion table referred to by the quantization circuit 14 for each color is switched to a conversion table according to the determination result, the entire three-dimensional color space is not subdivided, that is, the size of the conversion table is significantly increased. The color reproducibility can be more excellent without increasing the color reproducibility. Specifically, the conversion table of the three color components may be configured by a single semiconductor memory, and the conversion table may be switched by setting the memory address.

By the way, each color component RGB of the input signal is
Each has its own variable factors. In order to suppress such a variation unique to each color, a gamma conversion circuit for performing gamma conversion based on a conversion table stored in a built-in memory is provided at a preceding stage (input side) or a subsequent stage (output side) of the signal processing device 11. It is desirable to provide for each. Alternatively, a gamma conversion circuit may be arranged in both the former stage and the latter stage of the signal processing device 11. The gamma conversion of the input signal (8 bits) of each color is performed by an 8-bit address signal.
Because it is realized by the conversion table of bit data width,
It is sufficient that the built-in memory of each gamma conversion circuit 12 has a capacity of 256 bytes.

Here, each color component of the input signal is gamma-converted by an individual gamma conversion circuit. However, when one gamma conversion circuit performs gamma conversion on all color input signals, the conversion is performed. The internal memory where the table is stored is 2
It is sufficient if there is a capacity of 56 bytes × 3 colors × 1 input = 768 bytes. In the case of signal conversion from an RGB signal to a CMYK signal, the configuration may be such that a memory for storing the conversion table has a capacity of 1024 bytes.

The arrangement of the dots formed by the error diffusion process is not regular, and it is not always possible for a color printer to record the dots of each color ink at the same position. Generally, in a low-density image with a small number of dots, dots are hardly overlapped, and in a high-density image with a large number of dots, dot overlap is difficult to avoid. Depending on how such dots overlap, the way the viewer sees the image may differ. In order to solve such a problem, the remaining output signals may be adjusted according to the first output signal at the same position. For example, when calculating the output signals in the order of CMYK, if the output signal of the color component C fluctuates higher, dots are printed with high-density C ink. The suppression signal for suppressing the signal level is subtracted, and the remaining color components M
The output signal of YK may be adjusted. By feeding back the error signal, a signal that fluctuates due to the suppression signal is added in another pixel, and the signal level of the entire image can be maintained. This makes it difficult for dots to overlap, thereby improving the color development of the ink.

(1-1) More Specific Example of Apparatus Configuration FIG. 15 shows an example of an apparatus configuration for realizing the present invention.

Using the multivalued data of the black-and-white image as an input signal C, signal conversion is performed based on the characteristic curve stored in the conversion table, and the result of the loop operation is averaged N (N is 1 or more) times to obtain an output signal. Z is obtained. It has already been described that a similar configuration can be applied to a color image.

The input signal C can be temporarily stored in the memory 33. The data of the pixel to be processed is read from the memory 33, and the error signal E
And a correction signal P is calculated and sent to the quantization device 14. The conversion table 19 is a memory device storing characteristic curve data and the like, and this memory address is created by the address generation circuit 30. The conversion table 19 can store a plurality of types of characteristic curves. The conversion table can be switched by inputting the determination result of the color area determination circuit 18 for determining the color type using the pixel data read from the memory 33 to the address generation circuit 30 and adjusting the memory address. The output of the conversion table 19 is an output signal Z,
This is an intersection signal f (Z) for calculating an error signal. The error calculating device 15 is a device for obtaining an error signal generated when outputting the output signal Z in the coordinate system of the input signal, and the error calculating circuit 71 uses the correction signal P and the intersection signal f (Z).
To calculate. The error propagation circuit 32 plays a role of an attenuator when outputting the calculated error signal E to the error feedback device 16. Normally, the error signal E is output without being attenuated. However, for example, when the pixel to be processed is located at an edge portion of the input image, the edge is preserved by attenuating the error signal E. For example, the operation of the attenuator can be an on / off switch. For the leading pixel of the edge change detected using the edge detecting circuit 31, the error propagation circuit 32 is turned off so as not to transmit the error signal E generated in the previous processing, thereby correcting the signal level of the leading pixel. By not doing so, the edge portion of the image can be preserved. The output signal Z is input to the averaging circuit 20, and outputs an averaged output signal.

The error feedback device 16 uses the error signal E calculated as described above to calculate a correction signal for correcting the target pixel. For this purpose, it comprises a register for storing an error signal for a pixel which has already been processed, and an error distribution circuit 13 for calculating a correction signal using the signal of the register. The register may be a memory for storing an error signal for a pixel on a line, but here, an example is shown in which the register is composed of two types of registers. One is a pre-pixel error register 36 for storing an error signal E calculated by using the error calculation circuit 15, and the other is a residual register 35 for storing a residual signal of a correction process. Signal distribution circuit 37
Transmits to the signal correction device 13 the result of weighting and adding the signals stored in both registers with appropriate coefficients.
Further, the signal distribution circuit 37 calculates the difference between the sum of the error signals stored in the two registers and the error signal transmitted to the signal correction device 13, and calculates the remainder of the error signal not used for correction. Is stored in the residual register 35. The need for the residual register 35 in this way is a result of not using the total value for correction by weighting appropriate coefficients as described above. Further, the signal distribution circuit 37 is incorporated with, for example, a random number generation circuit to variably set the coefficient value, thereby preventing a pattern called a texture from being generated. By combining the previous pixel error register 36 and the residual register 35 in this way, it is possible to feed back the generated error signal without leaving it. The configuration of these registers is not limited, and a configuration using a register that stores a value obtained by adding a previous pixel error and a residual, or a configuration using a line memory including a target pixel can also be used.

In this way, the output signal Z as a conversion result based on the characteristic curve can be output while the signal level of the input signal C is preserved.

Although the above description has been given of the case of a monochrome image, the same operation is performed for a color image expressed by a combination of a plurality of color signals. However, the conversion table is determined by the combination of the input color signals C (or the combination of the correction signals P). The conversion table stores conversion results that are similarly combined corresponding to these combination color signals, and this value is read out from the memory and output.

FIG. 16A shows the structure of the quantization device.
From the input correction signal P, an address generation circuit is used to generate an address signal for obtaining a conversion table. The conversion table is
It is composed of a memory device that stores the output signal Z and the f (Z) signal. The contents of the conversion table can be written in advance or rewritten externally. FIG. 16B shows a specific configuration example of the address generation circuit. R entered
The upper bits of the correction signals Pr, Pg, Pb relating to GB are extracted and output as the memory addresses of the conversion table.
Also, a plurality of types of conversion tables can be prepared and used while switching. For example, the correction signals Pr, Pg, Pb
An address signal for switching the conversion table by the area determination circuit is output by using the upper bits of. This area determination circuit can be constituted by a memory device in which inputs and outputs are associated in advance, and the contents can be rewritten from outside. In this manner, for a color type sensitive to a difference, such as a skin color, the operation can be performed so as to reduce the calculation error.

FIG. 17 shows a specific configuration of the error feedback device 16.

The input error signal E is temporarily stored in the previous pixel error register 36. Further, a residual register 35 is prepared. Then, coefficient values F0, F1 set by the coefficient circuit
Using (F0 + F1 = 1), the calculation results of the weighting circuit and the adding circuit for performing weighted addition of the contents of both registers are passed to the signal correction device 13. Here, the coefficient circuit may set the coefficient values F0 and F1 based on a random number generation circuit for the purpose of preventing the occurrence of texture due to error distribution. The difference between the error signal stored in both registers and the error signal transmitted to the signal correction device 13 is written to the residual register 35 through a residual circuit that calculates the residual signal.
In this manner, the input error signal E can be fed back for calculating the correction signal without being left. An advantage of this circuit is that a feedback loop for an error signal is formed without using a line memory. Conventionally, many methods use two
In order to realize an operation between a plurality of pixels that are dimensionally close to each other, a unit that stores a signal using a line memory has been used. In the present embodiment, since the signal operation is originally performed for an independent pixel, texture generation is prevented by setting a coefficient using a random number generation circuit. As a result, there is an advantage that the circuit configuration can be simply realized without the need for the line memory.

Using the device configuration as described above, (1) a signal is input at a constant cycle, and each circuit operates in accordance with this cycle. (2) The error signal E is fed back to the input signal C to generate the correction signal P. (3) An address signal is created from the correction signal P, and (4) an output signal Z and an intersection signal f (Z) are obtained by referring to a conversion table. (5) The transmission of the error signal E calculated from the intersection signal f (Z) and the correction signal P is turned on / off based on the edge detection result. (6) The error signal E is written into the previous pixel error register, (7) error distribution is performed using the content, and (8) the correction signal is calculated by adding the error signal to the input signal. (9) The output signal Z can be output at a timing according to the conversion table. These basic operation timings can be changed based on various effects. For example, if the operation of each circuit is fast, the number of steps can be reduced. Both can be realized by providing a scheduler for managing the overall operation timing. Also, the signal processing of the above procedure can be realized by software, or an operation in which software and hardware are mixed can be realized.

(2) From a RGB signal to a multicolor signal (CMYK)
Application example of signal conversion processing to RGB signals) CMY having more color components than RGB signals
When converting to a K signal, the signal processing device
As shown in (1), it is necessary to mount the above-described signal correction circuit 13 for each of the color components RGB of the input signal.
The error calculation processing and the error feedback processing do not change for any of the color components RGB of the input signal.
6 does not require one for each color. The quantization circuit 14 has the configuration shown in FIG.
1, the four-color signal CMYK is added to the coordinates (R, G, B) of each representative point on the three-dimensional space composed of the input signals.
Are generated, a black signal K associated with the combination of the correction signals P of the respective color components is generated. Then, an error diffusion process is performed so as to preserve the signal levels of the input signals (R, G, B). As a result, the converted signal (C, M, Y, K) is output.

Here, a correction signal of the input signal (R, G, B) is used in generating an address for reading the conversion table. Therefore, for the black signal K, the error calculation circuit 15 and the error feedback circuit 16 are unnecessary. That is, regardless of the number of colors of the output signal, the arithmetic processing based on the number of colors of the input signal may be performed. Basically, a similar configuration can be used to convert a three-color signal into a multicolor signal of five or more colors. Therefore, if this configuration is used, it is possible to cope with printing in which recording is performed with inks having a wide color reproduction range, such as inks having different densities, specially toned inks, and the like.
In any case, smoother gradation reproduction can be realized by writing the contents of the conversion table in advance.

In this manner, a signal processing apparatus suitable for image recording by an ink-jet printer or the like using multi-color ink is realized.

(3) Example of circuit configuration for high-speed processing The circuit configuration for realizing high-speed signal processing is shown below.

(A) Signal Processing for Compressed Data According to the present invention, color conversion and gradation conversion can be executed for each block in which a plurality of pixels are put together. For example, as shown in FIG. 19, it can be combined with a data compression method in which the number of pixels included in the block is smaller than that of the approximate color signal. In the example shown in the figure, the color type of the internal pixels is represented by C1 = (R1,
G1, B1) and C2 = (R2, G2, B2) can reduce the data amount. For example,
In a color printer or the like, since the pixel density is high, even if the color signals are approximated in block units, the image quality is hardly affected. In a color printer or the like, it is necessary to convert a color signal according to the properties of ink used for printing. Further, there are restrictions on the gradation levels that can be printed in pixel units due to the characteristics of the printer. For example, in an ink jet printer, cyan, magenta,
Using yellow, the gradation level of the ink dot is ON /
In many cases, there are two gradations of OFF. In the present invention, the signal conversion adapted to the characteristics of the output device is performed by performing color conversion and gradation conversion on the approximate color signal in block units as described above. Then, by performing expansion processing using the converted signal, a signal conversion result in pixel units can be obtained. By utilizing the present invention, high-speed signal processing can be performed.

An apparatus configuration as shown in FIG. 20 is shown. The basic configuration is the same as that described above, and high speed can be realized by inputting an approximate color signal for each block and executing color conversion and gradation conversion. The compression processing and the decompression processing can be executed by another circuit.

(B) Parallel processing When the inside of a block is represented by a plurality of approximate color signals as described in (a), signal processing is performed by temporally parallelizing the plurality of approximate color signals. Speed can be realized.

For example, as shown in FIG. 21, two systems of circuits can be prepared to execute signal processing. here,
Since the contents of the conversion table (memory) 14 can be used in common for both operations, setting the shift of the access time allows the two signal processing circuits to use the common, thereby reducing the circuit scale. Can be.

Further, six signal processing circuits are prepared.
If the configuration is such that two types of RGB color signals are processed and executed simultaneously, all the two types of approximate color signals C1 and C2 in the block described in (a) can be processed in parallel, which is effective for speeding up. There is.

(C) Reduction of Conversion Table Access Counts In image data, color signals of adjacent pixels may be the same. For the same input signal, the conversion result is basically the same. In this way, by observing the input signal before conversion, for example, when the input pixel is at the same level as the previous pixel, the conversion result of the previously generated previous pixel is output, thereby performing signal processing on the input pixel. Can be deleted.

In order to execute the above processing, as shown in FIG. 22, the already processed color signal is temporarily stored in a latch, and when the input color signal has the same value as the color signal of the previous pixel, Reads the conversion result from the latch and outputs it. The configuration of the latch is arbitrary, and can be configured similarly to a so-called cache memory.

Alternatively, conversion results corresponding to frequently used specific color signals such as white, black, red, blue, and green can be prepared in a memory in advance.

(4) Pixel Skip Conversion Image data is often represented by combining one pixel with a color signal such as RGB. However, for example, a liquid crystal display device or the like is configured by a combination of an optical switch using a liquid crystal element and a color filter. That is, one pixel can display only one color. For example, as shown in FIG.
In a liquid crystal display device having a GB color filter,
Each color signal is displayed as a skip pixel corresponding to the arrangement of the color filters. On the other hand, the format of the color signal used in a television or the like includes three color signals (for example, RGB) in one pixel so as not to restrict the configuration of the display device. In order to display such a television signal on the liquid crystal display device, the color signal is used while jumping in accordance with the color arrangement of the pixels. There are the following two ideas for using color signals at intervals according to the display device configuration.

(A) The skipped k pixels are virtual pixels restricted to the same signal level as the adjacent pixels.
(K + 1) including virtual pixels due to visual integration effect
It is recognized that the pixels have the same color signal.

(B) It is assumed that the output level of the skipped k pixels is restricted to zero. Due to the visual integration effect, it is recognized as having a color signal including the zero-constrained virtual pixel.

The signal processing of (a) and (b) is (k +
1) pixels having color signals are (k + 1) pixels,
Or one. That is, since the problem can be reduced to the problem of the magnification of the output signal, the signal processing procedure can be treated as the same.

By performing color signal conversion and gradation conversion in accordance with the characteristics of the display device, it is possible to reduce the processing time, reduce the circuit scale, and improve the image quality. In the present invention, as shown in FIG. 24, the number of gray scales can be increased by averaging the output signals for k pixels at the jump position not used for display.

In the above configuration example, N is set for a single pixel.
Although the number of gradations has been increased by performing the loop processing, in the present embodiment, k operations on the interlaced pixels achieve the same effect as the loop processing. For the skipped pixel, the error is calculated and the feedback is performed, assuming that the output signal of the previous pixel is maintained.

Further, by combining the loop processing of N times for a single pixel and the processing of k interlaced pixels, the average output of (N + k) repetition operations for the color signal used for the display pixel is used. In addition, the conversion error can be quickly converged.

As a result, by performing signal conversion and gradation conversion according to the color arrangement of the display device, the load of signal processing can be reduced.

Further, the color signal corresponding to the skipped pixel is
If there is a large change in the signal level, it can be determined that the edge of the image is located at the pixel position. In the case where such a signal change appears significantly, the feedback of the error component is of little significance, so that a switch circuit can be provided to turn on and off the feedback signal.

(6) Means for Creating and Using Conversion Table The color calibration of the conversion table will be described. However, here,
The case where the signal processing according to the present invention is realized by software will be described as an example. For example, as shown in FIG.
A case will be described in which the color image data read from 0 is output by the printer 52. Color correction device 58 of the device
Is based on the results of the special inhabitants of the input / output devices in advance.
Prepare a color conversion table. However, the reproduction characteristics of the color signals created in advance may fluctuate due to some factors. For example, in a color laser printer,
Changes in the charging characteristics of the photoconductor due to temperature and humidity, changes in the charging characteristics of the toner, and the like are factors that cause variations in the reproduction characteristics of color signals.

Factors such as these (eg, temperature and humidity)
Can be measured directly, and based on the measurement result, fluctuations in the reproduction characteristics of the color signal can be suppressed.However, when the reproduction characteristics of the color signal change over time, the cause of the fluctuation and the causal relationship are not always clear. If it cannot be said, it is necessary to measure the print result and suppress the fluctuation of the reproduction characteristics of the color signal based on the measurement result. Further, for example, it is necessary to correct the characteristic variation for each device based on the inspection result at the time of factory shipment.

To suppress these fluctuations, it is necessary to input a reproduced color of a display device or a printed matter to an image input device, and then create a conversion table based on the input data. Thus, in the present embodiment, a calibration mode that operates by combining a personal computer and a printer is prepared in advance, and the user can create a conversion table by activating the calibration mode. Specifically, as shown in FIG. 26, when the calibration mode is activated (S100), the personal computer temporarily sets existing data in a conversion table as one of initializations (S101).

Then, the print data of the calibration color patch prepared in advance is output from the printer (S102), and the output of the printer is output by a measurement-specific sensor such as a spectrophotometer or a general-purpose input device (scanner, digital camera, etc.) 227. The measurement is performed (S103). The measurement result is compared with the calibration patch data to determine whether or not the desired color reproduction characteristic and the error are small (S104). If the error is large, an operation for updating the conversion table necessary to realize the desired color reproduction characteristics is executed (S105). Then, the personal computer stores the conversion table in a non-volatile memory such as a flash memory as a conversion table to be used during execution of the software in which the above-described signal processing is defined. Is output (S106). In some cases, the above procedure repeats a plurality of loops until the error converges.

Alternatively, a plurality of types of conversion tables are created in advance, stored in a memory, and one of the conversion tables is selected and used based on a feedback signal of a characteristic signal from an output device such as a printer. It is also possible to realize signal processing according to the characteristics of the output device. Here, as the characteristic signal of the output device, a signal of the temperature of the device, the environmental humidity, the environmental illumination light, the paper used, or the like can be used. For the temperature, humidity, and the like, a signal from a sensor incorporated in the device can be fed back. By performing signal correction by the signal processing described in the present invention based on such characteristics of the output device, color reproducibility can be improved. As shown in FIG. 27, a memory 63 accessible from the CPU 60,
The processing of the present invention is executed while arranging image data, data of a color conversion table, and the like, and inputting and outputting memory data as the processing is executed. The processing can be executed by storing these data in the same memory device and inputting and outputting the data while shifting the access time. The signal processing may be configured by a dedicated circuit, and it goes without saying that the signal processing can be executed by software of the CPU 60. By mounting the signal processing circuit and the memory element on the same LSI chip, it is possible to shorten the memory access time and increase the input / output data width, thereby improving the signal processing execution speed. .

[0081]

According to the present invention, it is possible to provide a signal processing method and a signal processing apparatus capable of executing both the color signal conversion processing and the gradation conversion processing at a high speed with a small-scale circuit configuration.

[Brief description of the drawings]

FIG. 1 is a schematic circuit configuration diagram of a signal processing device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a pixel arrangement example and a positional relationship between a target pixel and a reference pixel according to an embodiment of the present invention.

FIG. 3 is a diagram exemplifying the contents of a conversion table according to an embodiment of the present invention.

FIG. 4 is a diagram for explaining an operation of the present invention by enlarging and displaying a part of a conversion table.

FIG. 5 is a diagram for explaining the operation of the error diffusion process using a numerical example.

FIG. 6 is a schematic circuit configuration diagram of a signal processing device according to an embodiment of the present invention, illustrating an example of the configuration of an averaging circuit.

FIG. 7 is a diagram for explaining the operation of the circuit configuration of the present invention using numerical examples.

FIG. 8 is a diagram for explaining the operation of the circuit configuration of the present invention using a numerical example.

FIG. 9 is a schematic circuit configuration diagram of a signal processing device according to an embodiment of the present invention, showing an example of the configuration of an averaging circuit and a quantization circuit.

FIG. 10 is a diagram illustrating a flowchart of a signal processing method according to an embodiment of the present invention.

FIG. 11A is a diagram showing a three-dimensional color space;
(B) is a diagram for explaining the data structure of the conversion table.

FIG. 12 is a schematic circuit configuration diagram of a signal processing device according to an embodiment of the present invention.

FIG. 13 is a diagram showing a three-dimensional color space and a memory configuration for storing a conversion table.

FIG. 14 is a diagram illustrating a configuration example of input / output signals of a memory that stores a conversion table.

FIG. 15 is a schematic circuit configuration diagram of a signal processing device according to an embodiment of the present invention.

FIG. 16 is a diagram illustrating an example of a circuit configuration of a quantization circuit.

FIG. 17 is a diagram illustrating an example of a circuit configuration of an error calculation circuit.

FIG. 18 is a schematic circuit configuration diagram of a signal processing device according to an embodiment of the present invention.

FIG. 19 is an explanatory diagram of a processing procedure of the embodiment of the present invention in combination with a color image compression method.

FIG. 20 is a schematic circuit configuration diagram of a signal processing device according to an embodiment of the present invention, which is combined with a color image compression method.

FIG. 21 is a schematic circuit configuration diagram of a signal processing device according to an embodiment of the present invention.

FIG. 22 is a schematic circuit configuration diagram of a signal processing device according to an embodiment of the present invention.

FIG. 23 is a diagram for explaining the correspondence between pixels of a color image and color signals.

FIG. 24 is a diagram showing the operation timing of the present invention corresponding to the display device characteristics of a color image.

FIG. 25 is a schematic configuration diagram of an image processing system according to an embodiment of the present invention.

FIG. 26 is a view illustrating a flowchart of a signal processing method according to an embodiment of the present invention.

FIG. 27 is a schematic configuration diagram of an image processing system according to an embodiment of the present invention.

[Explanation of symbols]

11 signal processing circuit, 13 signal correction circuit, 14 quantization circuit, 15 error calculation circuit, 16 error feedback circuit, 1
8: color region determination circuit, 20: averaging circuit, 21: output quantization circuit.

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5B057 CA01 CA02 CA08 CA12 CA16 CB01 CB02 CB08 CB12 CB16 CE11 CE17 CE18 CH01 CH07 CH11 CH18 DB02 DB05 DB06 DB09 DC25 5C077 LL17 LL18 MP01 MP08 NN11 NP01 PP15 PP32 PP08 PQ23 5C079 HB01 HB03 HB12 LA12 LA31 LB02 LC09 MA04 MA11 NA09 NA11

Claims (6)

[Claims] 1. A memory storing a conversion table in which two types of signals represented as coordinates in mutually different coordinate systems are associated with each pixel, and an input signal of each pixel is stored in the conversion table. A quantization circuit that sequentially converts the signal into a signal associated with: an error calculation circuit that sequentially calculates an error of the signal conversion each time the quantization circuit performs signal conversion for one pixel; A signal correction circuit for spreading the calculated error to an input signal before signal conversion by the signal conversion circuit. 2. The signal processing device according to claim 1, further comprising: a determination circuit for determining a characteristic of the input signal; wherein the memory stores a conversion table for each characteristic of the input signal; A signal processing device that performs the signal conversion based on a conversion table according to a determination result of the determination circuit. 3. The signal processing device according to claim 1, wherein the conversion table associates the input signal with an output signal having more color components than the input signal. A signal processing device, wherein the quantization circuit converts the input signal into an output signal whose number is larger than the number of the input signals. 4. The signal processing device according to claim 1, wherein said conversion table is a means for rewriting contents based on a feedback signal from a device to which an output signal is output, or a plurality of conversion tables. A signal processing device having means for selecting any one of the following. 5. The signal processing device according to claim 1, wherein N repetitive operations are performed on one input signal, and a result of averaging the N output signals is output. Signal processing device. 6. A quantization step in which an input signal of each pixel is associated with the input signal based on a conversion table in which two types of signals represented as coordinates in mutually different coordinate systems are associated with each pixel. In the error calculation step, each time signal conversion is performed for one pixel, an error of the signal conversion is sequentially calculated. In the signal correction step, the input before the signal conversion by the quantization step is performed. A signal processing method for diffusing an error calculated in the error calculation step into a signal.