JP2002118744A - Image signal processing unit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a signal processing system and a signal processing unit configuration to obtain high image quality, in spite of a small scale circuit configuration through the signal processing applied to an image signal received by using a photoelectric conversion element, such as a CCD or a contact image sensor. SOLUTION: The data width of a referenced signal per pixel is selected to be d+f+j<=8-bits, where d is the number of bits of a white reference signal of a shading correction means, f is the number of bits of adjacent pixel signals of an edge enhancement means, and j is the number of bits of error component for an error spread processing means. Or the data width of the referenced signal per pixel is selected to be d+f+j'<=8-bits, where j' is the number of bits of the j-bits above shared by pixels (N pixels) in the error spread processing means divided by N (j'=j/N). Thus, the signal data width to be referenced in the respective signal processing processes can be reduced, and the accuracy of the signal processing can be fully satisfied, for example, memory sections can be mounted easily on one and the same LSI, and a required memory has only to store signals consisting of 8-bits per pixel, so that the image signal processing unit can utilize a low-cost memory configuration.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention is applicable to an image signal processing apparatus such as a scanner for inputting an image signal.

[0002]

2. Description of the Related Art In an image input device such as a facsimile,
Image signal processing is performed in which an image signal input using a photoelectric conversion element such as a CCD (charge coupled device) or a contact sensor is converted into a signal suitable for purposes such as storage and transmission. An example of this signal processing will be described below.

(1) Shading correction An image signal input using a photoelectric conversion element such as a CCD (charge coupled device) or a contact sensor includes:
Since distortion components such as unevenness of illumination and characteristic variations of each photoelectric conversion element are included, the distortion is corrected, and shading correction for reproducing a uniform image signal is performed.

(2) Edge enhancement (MTF correction) In order to correct MTF characteristics of an optical system or a sensor and to improve edge characteristics, edge enhancement processing is performed.

(3) Pseudo halftone In order to reproduce a halftone image in a monochrome printer, a signal processing method called an error diffusion process is performed.

In addition to these, gamma processing for converting the level of an input signal, binarization processing for outputting a signal such as a document image, a halftone image portion such as a photograph, and a monochrome image such as a document image are performed.
Signal processing and the like for separating a value image portion into regions have been proposed.

These signal processing methods are generally described, for example, in "Digital Signal Processing of Images" published by Nikkan Kogyo Shimbun.

[0008]

In realizing the above signal processing, a memory for storing a signal to be referred to in each signal processing is required.

Conventionally, each signal processing means and the memory were individually combined. However, as a whole device, there are a plurality of memory interfaces, the circuit scale is increased, and unused portions are generated in the memory. And so on.

It is an object of the present invention to reduce the memory capacity for realizing the above-described signal processing, to have a simple configuration for inputting and outputting data to and from the memory, and to obtain a high-quality image.

[0011]

According to the present invention, an image signal (a bit / pixel) is converted to a white reference signal (b).
In the shading correction means for obtaining an output signal (c bit / pixel) by performing correction using bit / pixel), each data width is set as a = c, a> b, and a white reference signal (b bit / pixel) Sets each data width to b> d using a means for compressing and expanding the signal and a means for storing the compressed white reference signal (d bit / pixel), and sets a signal (e bit / Pixel)
Is corrected by using signals (f bit / pixel) of a plurality of adjacent pixels to obtain an output signal (g bit / pixel), and the respective data widths are set to e ≧ f, and attention is paid. An error for obtaining a binarized output signal (k bit / pixel) based on a predetermined calculation procedure using a pixel signal (i bit / pixel) and an error component (j bit / pixel) before and after the binarization processing. In the diffusion processing means, each data width is set as i> j, k = 1, and a memory of 8 bits per pixel is distributed for the above means, and d + f + j = 8.

Alternatively, an error component shared by a plurality of pixels (N pixels) is set, and the data width per pixel is set to j ′ = j
/ N, d + f + j ′ = 8.

[0013]

By using the above means, it is clear in the shading correction means that the signal amplitude range of the reference signal including the shading distortion is narrower than the signal amplitude range of the image signal to be corrected. If
By setting each data width as a = c, a> b, it is possible to satisfy the calculation accuracy.

The white reference signal (b bit / pixel) is compressed and expanded by using a means for compressing and expanding the signal and a means for accumulating the compressed white reference signal (d bit / pixel) so that each data width is b> d. By setting, the memory capacity can be reduced.

The input signal of the edge enhancing means is subjected to signal conversion by gamma conversion. In the gamma conversion, there is a signal area in which a plurality of steps of an input signal are converted into an output signal of one step, or a signal area in which one step of an input signal is converted into an output signal of a plurality of steps. For this reason, in many of the signal regions, the calculation accuracy can be satisfied by setting the data width to be e> f.

In the error diffusion processing means, error components of a plurality of adjacent pixels (N pixels) are calculated by signal processing such as averaging, and commonly used by the N pixels. As described above, since the signal fluctuation is suppressed by the averaging, each data width can be set as i> j and k = 1.

By realizing image signal processing with the above configuration, the data width of a signal to be referred to in each signal processing is reduced, and the accuracy of the signal is sufficiently satisfied in each signal processing, and an image signal of high image quality is obtained. Obtainable.

By realizing image signal processing with the above configuration, memories used in each signal processing are integrated, data input / output timing of the memories is shared, and data input / output at one time is stored in each memory. By dividing and using the signal processing, it is possible to simplify the configuration of the device for realizing the image signal processing and obtain a high quality image signal.

Further, by sharing a signal related to an error component to be referred to in the error diffusion process by two pixels, the required memory capacity can be reduced.

By reducing the memory capacity required for image signal processing, a memory unit can be easily incorporated into a circuit for realizing image signal processing.
When the circuit is configured by the SI, it is easy to mount the memory unit on the same LSI.

Also, by obtaining the above-described operation, the memory capacity required for all image signal processing can be reduced to 8 bits per pixel. This is advantageous in that a standard, that is, inexpensive, memory can be used because a widely used standard memory configuration performs signal input / output in units of 8 bits.

[0022]

FIG. 1 shows a specific example of the present invention.

The shading correction means converts the image signal into C
Input from a photoelectric conversion element such as a CD or a contact sensor,
A signal for which shading correction has been completed is output while inputting / outputting a signal to / from a memory storing reference signals for shading correction.

The edge emphasizing means receives the signal for which shading correction has been completed, performs a filtering process based on a preset coefficient while outputting and inputting a signal to and from a memory storing adjacent image signals, and outputs a signal. .

The error diffusion means inputs a signal after signal processing such as shading correction and edge enhancement when a pseudo halftone image is selected, and stores the signal in a memory storing error components to be distributed to adjacent pixels. The error diffusion process is performed while performing the input / output of, and the signal is output.

In addition, there are gamma processing for converting density characteristics of an image signal, and a binarizing circuit for judging black and white of a black and white binary image. According to the present invention, these shading correction, edge enhancement, error diffusion, gamma processing,
It does not depend on the details of a specific signal processing procedure such as value conversion, but may be any signal processing based on the principle.

The above-described shading correction, edge enhancement,
To execute the error diffusion, signal accumulation means for at least one line is required for each signal processing.

In the following, when the image signal to be input after A / D conversion of the sensor output signal is 6 bits / pixel, the width of the data used in each signal processing is limited to the precision required for the calculation. An example will be described in which the signal delay unit is configured as a memory for one line of 8 bits per pixel.

In the shading correction means, as shown in FIG. 2, means for compressing and expanding the white reference signal are provided, and one pixel
The white reference signal is stored in the form of a bit (d = 1). When reproducing the waveform of the white reference signal, there is no signal deterioration due to compression and expansion, which is a condition necessary for accurately executing shading correction. Here, as shown in FIG. 3, if the distortion of the white reference signal is within 50% of the amplitude range based on white, the data of the white reference signal reproduced from the input image signal (a = 6) The width b may be narrow (a> b). For example, if the former is 6 bits, the latter may be 5 bits. As a result of shading correction, a correction signal having a 6-bit width (c = 6) can be obtained.

The edge emphasizing means executes a filtering process on the image signal after the gamma conversion with reference to the image signal of the previous line. Here, as shown in FIG. 4, in the gamma conversion processing, the output signal can be expressed with a data width f that is equal to or less than the data width e of the input signal in consideration of the setting of the conversion curve (e> f). For example, if the former is 6 bits, the latter may be 5 bits, so that the reference signal can be data having a width of 5 bits or less. As shown in FIG. 5, a memory for one line is prepared, and processing using a 2 × 3 or 2 × 5 coefficient matrix is executed with reference to the 5-bit data of the previous line.

In the error diffusion processing, 2 in a plurality of pixels
Since the error components before and after the binarization process are weighted and added and sequentially fed back, the data width j of each error component need not be the same as the data width i of the image signal of the pixel of interest (i> j). ), And the overall image density is preserved. Further, the error diffusion means averages the error components of adjacent N pixels, and stores the error components in the memory for N pixels, so that the memory capacity required for each pixel can be reduced. For this reason, as shown in FIG. 6, after averaging the error components at two adjacent pixels (N = 2) to obtain data having a 4-bit width (j = 4), 2 bits per pixel (j ′ = j) / N).
The error component stored in the memory is referred to as shown in FIG.

In this way, a signal to be referred to in shading correction, edge enhancement, and error diffusion signal processing is set to 8 bits per pixel. Then, as shown in FIG.
By using means for allocating data to each signal processing means, it is possible to use a memory for inputting and outputting 8-bit data in one access. Generally, a memory configuration in units of 8 bits is widely used, so that a wide variety of general-purpose and inexpensive memories can be used.
The apparatus can be simplified and the price can be reduced. LS
The image processing circuit and the memory circuit can be easily incorporated into the I chip.

When the line memory is configured as an external circuit of the image processing means, the memory for inputting / outputting data in units of 8 bits is configured according to the number of pixels per line, and the memory capacity is increased. Can be set.
For example, when reading a B4 size document at 8 pixels per mm, one line is composed of 2048 pixels, so it is sufficient to use a memory of 2048 bytes.

The memory can provide further effects by preparing not only the respective signal processing means but also data input / output means from a control device (such as a CPU) for executing signal processing based on a program. For example, to set a white reference signal for shading correction,
After inputting an image signal obtained by reading an object as a white reference without performing signal processing of shading correction, edge enhancement, and error diffusion, and executing a signal procedure based on predetermined software using a CPU, By writing the calculated reference signal in the memory, shading correction can be performed using the white reference signal calculated by an arbitrary algorithm.

Using an image signal stored in a line memory, an image signal of an area composed of a plurality of pixels is used.
A determination of signal properties can be made. The area determination means for this purpose can naturally use a signal processing procedure fixedly set in advance, and also uses a learning means such as a statistical or neuro-computer which converts the input image signal into a statistical one. , Can also be set adaptively.

For example, a procedure for determining a signal property using a neuro computer will be described below.

(1) A circuit (which can be realized by software) is set using a learning process such as back propagation.

(2) An image signal of a minute area composed of a plurality of pixels is input to the circuit, and the screen is sequentially scanned while outputting the judgment result of the minute area.

(3) Using the result of the determination of the minute area, the area on the wider screen is divided, and image signal processing suitable for each area is executed.

Here, it is necessary to accumulate a plurality of pixel signals constituting a minute area in a line memory or the like. The line memory used in the signal processing means and the accumulated signals can be used. . For this reason, it is possible to realize signal processing for judging the signal properties in the area without having to prepare a new memory. It can be used for setting a coefficient matrix for edge enhancement, setting a diffusion coefficient of an error component in error diffusion, and further, for example, separating a binary image and a halftone image.

[0041]

According to the present invention, it is possible to reduce the data width of the signal referred to in each signal processing, sufficiently satisfy the accuracy of the signal processing, and obtain a high-quality image signal.

For example, when the circuit is configured by an LSI, it is easy to mount the memory unit on the same LSI.

The memory required for image signal processing can be reduced to 8 bits per pixel, so that an inexpensive memory configuration can be used.

As described above, according to the present invention, it is possible to realize a small-scale circuit configuration in consideration of the entire image processing of a read signal and to obtain high image quality.

[Brief description of the drawings]

FIG. 1 shows a specific example of the present invention;

FIG. 2 is a configuration diagram of a shading correction unit;

FIG. 3 is an explanatory diagram of a white reference waveform for shading correction,

FIG. 4 is an explanatory diagram of gamma conversion,

FIG. 5 is a configuration diagram of an edge enhancement unit;

FIG. 6 is a configuration diagram of an error diffusion unit,

FIG. 7 is an explanatory diagram of an error diffusion process;

FIG. 8 is an explanatory diagram of data input / output of a line memory.

[Explanation of symbols]

 100 ... Line memory.

────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] October 30, 2001 (2001.10.
30)

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] Claims

[Correction method] Change

[Correction contents]

[Claims]

1. A shading correction, edge enhancement, the image signal processing apparatus for executing the signal processing of the error diffusion,
A memory for accumulating signals to be referred to in each of the signal processing, and an area discriminating means for outputting a state value relating to image properties using image signals of a plurality of pixels accumulated in the memory; An image signal processing device that performs signal processing suitable for image characteristics by changing parameters used for enhancement and error diffusion processing.

2. An image signal processing apparatus for executing each signal processing of shading correction, edge enhancement, and error diffusion,
It has a memory for accumulating signals to be referred to in each of the signal processing, and inputs and accumulates image signals to a pre-configured neuro computer, and outputs a state value related to image properties, thereby achieving the edge emphasis and error. An image signal processing apparatus that performs signal processing suitable for image characteristics by changing a parameter used for diffusion processing.

3. An image signal processing apparatus for executing each signal processing of shading correction, edge enhancement, and error diffusion,
It has a memory for accumulating a signal to be referred to in each of the signal processing, and inputs an image signal to be used as a white reference without performing shading correction, edge enhancement, and error diffusion processing, and converts the image signal into predetermined software. An image signal processing apparatus comprising: writing a result of signal processing based on the white reference signal to the memory as a white reference signal for the shading correction; and performing shading correction of the image signal based on the white reference signal.

4. An image signal processing apparatus according to claim 1, wherein said memory comprises a line memory having a capacity for storing a signal of one line constituting an image.

5. The image signal processing apparatus according to claim 1, wherein said memory is configured with 8 bits per pixel as a unit.

 ──────────────────────────────────────────────────の Continuing on the front page (72) Inventor Shinichi Shinoda 4026 Kuji-cho, Hitachi City, Ibaraki Prefecture Within Hitachi, Ltd.Hitachi Research Laboratory (72) Inventor Yasuyuki Kojima 216 Totsuka-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture Hitachi, Ltd. F-term in Information and Communication Division (Reference) 5B057 BA02 CA02 CA08 CA12 CA16 CB02 CB08 CB12 CB16 CC01 CE03 CE13 CH08 CH11 5C077 LL02 LL19 MM03 MP01 NN11 PP03 PP06 PQ12 PQ22 PQ24 SS01 TT06

Claims (8)

[Claims] In a signal processing of an image signal input using a photoelectric conversion element, a white reference signal (b bit / pixel) for shading correction is provided.
Means for compressing / expanding to d bits per pixel (data width is b> d), using the signal of the pixel of interest (e bit / pixel) using the signal of multiple adjacent pixels (f bit / pixel) Means for edge emphasizing by using a signal of a pixel of interest for calculating a pseudo halftone image and an error component (j bit / pixel) of a pixel which has already been binarized to obtain a binarized output signal An image signal processing apparatus comprising: an error diffusion unit; and a line memory for accumulating signals by setting a data width of a signal per pixel to d + f + j = 8. 2. The error diffusion processing means according to claim 1, wherein an error component (j bit / pixel) of a pixel which has already been binarized is averaged over a plurality of pixels (N pixels). Means for distributing the converted error components to a data width of j '= j / N per pixel, and comprising a line memory for storing signals with the data width of a signal per pixel being d + f + j' = 8. An image signal processing device characterized by the above-mentioned. 3. A line memory for storing signals according to claim 1 or 2, wherein a signal input / output by said compression / expansion means, edge enhancement means, error diffusion means and a signal input / output by CPU are selected by a switching means. An image signal processing device characterized by being able to. 4. An image signal comprising a shading correction means, an edge emphasis means, an error diffusion means, and a line memory for storing a signal referred to by these means on one LSI chip. Processing equipment. 5. An image signal processing apparatus according to claim 1, wherein said line memory is readable and writable from an external device other than said image signal processing means. 6. An image signal to be used as a white reference is input without performing shading correction, edge enhancement, and error diffusion processing, and the result of signal processing of the signal based on predetermined software is used as a white reference for the shading correction. An image signal processing apparatus, wherein after writing to a signal storage means, shading correction of an image signal is executed based on a signal of the storage means. 7. A parameter for use in the edge emphasizing means and the error diffusion means, which is changed by area discriminating means for outputting a state value relating to image properties using the image signals of a plurality of pixels stored in the line memory. An image signal processing apparatus that performs signal processing suitable for image characteristics. 8. An input and accumulated image signal is input to a pre-configured neuro computer, and a state value relating to image properties is output, thereby changing parameters used in the edge enhancement means and error diffusion means. An image signal processing apparatus that performs signal processing suitable for image characteristics.