JP3476605B2 - Image processing device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus,
The present invention relates to improvement of the quality of a read image of a document. 2. Description of the Related Art In an image processing apparatus, an original is read.
Read the white reference plate prior to
Thus, the output of the image sensor and the like are corrected. example
For example, a manuscript read in Japanese Unexamined Patent Publication No.
In the capture device, when the light source is turned off and on, the image sensor
Scans a white original, darkness correction data and lightness correction data
Data and store the stored dark correction data and light correction data.
A / D conversion of the image signal using the difference
I have. [0003] However, the above problems
An image sensor is used in the manuscript reading device when the light source is turned off and on.
Scans a white original to obtain dark and light correction data.
Output waveform characteristics and darkness
Time output waveform characteristics are not stored, so the image sensor
The image processing method is fixed regardless of the output waveform characteristics of
According to the image quality read by the image sensor, for example, M
Weighting for TF (Modulation Transfer Function) correction
Image processing such as the coefficient of the image, the threshold for binarization or the threshold for edge detection
Image processing cannot be performed by changing the parameters of
won. Further, the output waveform characteristics at the time of light and the output waveform at the time of darkness
Since the shape characteristics are not stored, the output waveform characteristics during light and the dark
Sensor image products based on the time output waveform characteristics
Monitor for image quality degradation and respond to read image quality degradation
A warning could not be displayed to the operator. The present invention has been made to solve such a disadvantage.
The output waveform characteristics at the time of light and the output waveform at the time of darkness
The purpose is to maintain good image quality based on the characteristics.
You. [0006] Image processing according to the present invention
The device is an image sensor when the white reference plate is exposed and scanned.
The maximum and minimum values of the output signal are used as the output waveform characteristics during light.
Turn off the light source and turn on the white light
Maximum value of the image sensor output signal when scanning the reference plate
And dark output to store the minimum value as dark output waveform characteristics
Waveform characteristic storage means for storing the stored bright output waveform characteristics and darkness
S / N ratio to calculate and store S / N ratio from time output waveform characteristics
Storage means, and a plurality of threshold values for which the stored S / N ratio is predetermined
After binarization of the read image according to the comparison result
Divide signal pixels into blocks consisting of adjacent pixels
And isolated blocks are detected by pattern matching.
And image processing means for performing correction. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention relates to a scanner or a camera.
In an image processing device such as a x-ray machine, the read image
It keeps quality in good condition. The image reading apparatus includes, for example, a light source, a white reference plate,
Condensing lens, mirror, image sensor, amplifier, peak
Hold unit, A / D converter, characteristic detection unit, characteristic storage unit,
Shading correction unit, line buffer, digital image
It has a processing unit, a system control unit, and an operation display unit. When an instruction to read a document is given, the reading of the document is performed.
Before taking the document, make sure that the light source is not
Expose the white reference plate. The light applied to the white reference plate is a mirror
After the optical path is changed by the
Incident on the image sensor. The image sensor is the incident optical signal
The signal is converted to an electric signal and amplified by an amplifier. Peak Hall
The load section detects the upper and lower limits of the amplified signal,
The / D converter changes the range of change of the signal amplified by the amplifier.
The range of conversion is the amplified signal from the image sensor.
Is quantized. The characteristic detection unit converts the quantized signal into an image
The maximum and minimum values of the output signal of the sensor
Detected as a characteristic, and the characteristic storage unit detects the detected bright output waveform.
Store the characteristics. Next, the A / D converter determines that the original is in the original reading position.
Image sensor with the light source turned off
Quantize the output signal of the The characteristic detection unit
The maximum and minimum values of the output signal of the image sensor from the signal
Detected as dark output waveform characteristics and detected by characteristic storage unit
Stores dark output waveform characteristics and read by image sensor
Be able to predict the image quality of the original. Where the characteristic
The detector detects the light output waveform characteristics and dark output waveform characteristics.
Set the maximum and minimum values of the image sensor output signal respectively.
Because it is adopted, the output waveform characteristics at the time of light and the output waveform at the time of darkness
It does not require a large amount of storage capacity to store shape characteristics.
Costly. The digital image processing section stores the stored
Output waveform characteristics and dark output waveform characteristics or S / N ratio
As an isolated pixel due to the deterioration of the read image based on
In the isolated pixel detection pattern where the pixels to be detected are continuous
If you expect that detection will not be possible,
By detecting and correcting standing, images due to background contamination etc.
The deterioration of quality may be further prevented. In addition,
Blocks are divided, for example, into four
Divide the element so that it becomes one block, and
To detect isolation. FIG. 1 shows an image processing apparatus according to an embodiment of the present invention.
It is a block diagram. As shown in the figure, the image processing device is a system
Control unit 1, characteristic detection unit 2, characteristic storage unit 3, target
Lath 4, light source 5, white reference plate 6, mirror 7, imaging lens
8, image sensor 9, amplifier 10, peak hold unit
11, A / D converter 12, shading correction unit 13,
Digital image processing unit 14, line buffer 15, and operation
It has a display unit 16. The system control unit 1 includes a main control unit 101 and S /
N ratio calculating unit 102, parameter changing unit 103, warning output
Unit 104 and a read control unit 105;
(Not shown) based on the system program stored in the
It controls the entire processing device. The main control unit 101 is a system
The control section 1 is entirely controlled. The S / N ratio calculation unit 102 has a characteristic
Bright output waveform characteristics and dark output waveforms stored in storage unit 3
S / N ratio = 20 × log ((output waveform characteristics at light
) / (Maximum value of dark output waveform characteristics))
To obtain the S / N ratio and store it in the characteristic storage unit 3. Parame
The data change unit 103 outputs the bright output waveform stored in the characteristic storage unit 3.
Characteristics in accordance with the characteristics and dark output waveform characteristics or S / N ratio.
Digital image processing section.
According to the output waveform characteristics at light and the output waveform characteristics at dark.
For example, MTF correction coefficient, isolated pixel detection pattern, binarization
Image processing parameters for the scanned document, such as thresholds and edge detection thresholds
Change parameters. Here, the MTF correction coefficient is optical
To correct the deterioration of image contrast caused by passing through the system
Is a weighting coefficient for the surrounding pixels. Binarization threshold and
Is a threshold value for converting a multi-level image signal into a black-and-white binary signal.
is there. The edge detection threshold is the pixel of interest and its surroundings
When the density difference from the pixel is greater than or equal to
The edge of the original in the process of replacing with white or black pixels
This is a threshold used for the determination. The warning output unit 104 stores characteristics
Light output waveform characteristics and dark output waveform characteristics stored in section 3
Can maintain a certain level of image quality even when corrected based on
When there is no such message, a warning is displayed on the operation display unit 16. For example,
When the S / N ratio falls below 20db, the warning output unit 104
A warning is displayed on the operation display unit 16. The read control unit 105
Light source when reading images or reading output waveform characteristics
5 emits light. Note that MTF correction and isolated pixel detection
Will be described later. The characteristic detecting section 2 has a maximum value detecting section 201 and a minimum value detecting section 201.
A value detection unit 202 is provided. Maximum value detection unit 201 is based on white
Output signal of the image sensor 9 when the plate 6 is exposed and scanned.
Of the image sensor 9 when the light source 5 is turned off.
Detect the maximum value of the force signal. Similarly, the minimum value detection unit 202
Is the image sensor 9 when the white reference plate 6 is exposed and scanned.
The minimum value of the output signal and the image sensor when the light source 5 is turned off
The minimum value of the output signal of the sensor 9 is detected. The characteristic storage unit 3 is clear
The time waveform characteristic storage unit 301 and the dark waveform characteristic storage unit 302
Prepare. The bright waveform characteristic storage unit 301 stores the maximum value detection unit 20.
Exposure of white reference plate 6 detected by 1 and minimum value detection unit 202
Maximum value of the output signal of the image sensor 9 when scanning is performed,
And the minimum value is stored as 8bit each as the output waveform characteristic at the time of light.
I do. The dark waveform characteristic storage unit 302 stores an image when the light source 5 is turned off.
When the maximum value and the minimum value of the output signal of the
Store as output waveform characteristics. The target glass 4 guides the conveyance of the original 17.
To prevent dust from entering the optical system. Light source 5 is Hara
When reading a manuscript or reading the bright output waveform characteristics,
The white reference plate 6 is illuminated. Mirror 7 is original 17 or white base
The reflected light from the quasi-plate 6 is guided by a predetermined optical path length. Imaging lens
The mirror 8 converts the original image guided by the mirror 7 at a predetermined reduction ratio.
Form an image. Image sensor 9 forms an image with imaging lens 8
The obtained optical information is photoelectrically converted into an electric signal. Amplifier 10
The analog signal from the image sensor 9 is increased at a predetermined magnification.
Width. The peak hold unit 11 is amplified by the amplifier 10
The detected maximum value of one line of the image signal is detected and held. A /
D converter 12 is the maximum value held by peak hold unit 11
The image signal is quantized based on. Shading correction
The unit 13 is an image data quantized by the A / D converter 12
The shading distortion caused by the characteristics of the optical system.
to correct. The digital image processing unit 14
Various types of image signals from which distortion has been removed by the corrector 13
Image processing. Line buffer 15 is Shadin
Various types referred to by the image correction unit 13 and the digital image processing 14.
Store the data. Diagram of the operation of the image processing apparatus having the above configuration
This will be described with reference to the flowchart of FIG. The operator inputs an instruction to read the document 17
Then (step S1), prior to reading the document 17,
The system control unit 1 stores in the light-time waveform characteristic storage unit 301.
Since the maximum value is cleared to zero and the minimum value is represented by 8 bits
After setting 0FFH, which is the maximum number that can be
2) Turn on the light source 5 (step S3). From light source 5
Is transmitted through the target glass 4 and reaches the white reference plate 6.
And a part of the irradiation light is reflected according to the reflectance of the white reference plate 6.
Shoot. The reflected light is divided into two types: specularly reflected light and irregularly reflected light.
Among them, part of the irregularly reflected light is transmitted through mirror-7.
To form an image on the image sensor 9 by the imaging lens 8.
You. A / D converter 12 outputs image sensor 9 for each pixel
, And the characteristic detecting unit 2 sequentially converts the quantized image data
The maximum value and the minimum value are detected by comparison (step S4),
It is stored in the light waveform characteristic storage unit 301 (step S
5). Here, the characteristic detecting unit 2 is, for example, one line in the main scanning direction.
When there are 2048 pixels in the area, the characteristic detection unit 2
The maximum value of the child pixel data that has been detected up to that point
And the minimum value, the pixel data
Greater than the maximum value detected so far or less than the minimum value
The maximum value or the minimum value is replaced for all pixels
Min and max values by repeating 2048 times which is the minute
Is detected. The reflected light of the white reference plate 6 when illuminating the white reference plate 6
The maximum value and the minimum value are detected, and the light waveform characteristic storage unit 301 is detected.
After that, the system control unit 1 stores the dark waveform characteristic storage unit.
The maximum value stored in 302 is cleared to zero and the minimum value is 8 bits
Set 0FFH, which is the maximum number that can be represented (step
Step S6), the light source 5 is turned off (step S7). In the dark
Just as when the waveform characteristics were stored, the image
The output of the sensor 9 is quantized by the A / D converter 12 for each pixel,
The characteristic detecting unit 2 sequentially compares the quantized pixel data, and
The maximum value and the minimum value are detected (step S8), and the dark waveform
It is stored in the sex storage unit 302 (step S9). like this
In addition, as the output waveform characteristics during light,
Takes the maximum and minimum values of the output of the di-sensor 9 and outputs it in the dark
Output of image sensor 9 when light source 5 is turned off as waveform characteristics
Since the maximum value and the minimum value of
It is not necessary to store all the outputs of the
The amount can be reduced. The maximum and minimum values
Instead of taking the standard deviation or the mean.
good. Parameter changing section 10 of system control section 1
3 is a light output waveform characteristic stored in the characteristic storage unit 3 and a dark output characteristic.
Based on output waveform characteristics, for example, MTF correction coefficient, isolated pixel
Read detection patterns, binarization thresholds, edge detection thresholds, etc.
The image processing parameters of the original document (step S
10). After that, start reading the image and
The image data read by the sensor 9 to the shading correction unit 1
3 and processed by the digital image processing unit 14 (step
Step S11). Note that the parameter changing unit 103 outputs the light
Based on S / N ratio instead of waveform characteristics and dark output waveform characteristics
May change the image processing parameters of the scanned original.
No. Next, the read image data is digitally converted.
As an example of image processing when processing is performed by the image processing unit 14, M
MTF correction which is a weighting factor when performing TF correction
The coefficient will be described. MTF correction of digital image data is performed by reading
A predetermined area including the target pixel of the captured image, for example, 3 pixels × 3
Cut out the pixel, weight the surrounding pixels, and
Generally. 3 pixel x 3 pixel area
(Hereinafter referred to as “matrix register”)
, And E as a target pixel, A, B, C, D, F,
G, H, and I are defined as peripheral pixels. For example, as shown in FIG.
As described above, the target pixel E is weighted by the coefficient 3, and the target pixel E
Weight the peripheral pixels B, D, F, and I of the
Then, the pixel density E1 of the target pixel E after the density correction is E1 =
3 × E0− (B0 + D0 + F0 + H0) / 2. This
Here, B0, D0, E0, F0, and H0 are multi-value density data.
It is. First, a correction operation at a uniform density will be considered.
Then, the constituent pixel data of the matrix register are all the same
Since it is data, the operation result is E1 = 3 × E0−4 × E
0/2 = E0, and the density correction of the target pixel E is not performed.
No. Next, the density of the peripheral pixel is, for example,
1/2, that is, the level of the pixel of interest is
Considering the case of being higher than the prime by E0 / 2, E1 = 3 ×
E0− (4 × E0) / (2 × 2) = 2 × E0.
Density correction is performed by 0. That is, the density E of the peripheral pixels
The density is increased by 3 × E0 / 2 compared to 0/2,
The contrast will be enhanced. Conversely, the level of the peripheral pixel is 3/2 of the pixel of interest.
, E1 = 3 × E0−3 × E0 = 0, and −
The density correction of E0 is performed. That is, the density of peripheral pixels is 3 ×
The density is lower by 3 × E0 / 2 than E0 / 2,
The side contrast will be enhanced. like this
Contrast correction is changed by weighting coefficient of each pixel
be able to. FIGS. 5 to 7 show examples of setting weighting coefficients. Figure
The weighting coefficient of 5 is E1 = 4 × E0− (2 × B0 + D0
+ F0 + 2 × H0) / 2, which is smaller than the case shown in FIG.
In all cases, the contrast correction effect in the sub-scanning direction is strong. In FIG.
The weighting factor is E1 = 4 × E0− (B0 + 2 × D0 + 2
× F0 + H0) / 2, which is smaller than the case shown in FIG.
Strong contrast correction effect in the main scanning direction. 7 weights
The weighting coefficient is E1 = 5 × E0− (B0 + D0 + F0 + H
0), and in the main scanning direction and in comparison with the case shown in FIG.
Strong contrast correction effect in the sub-scanning direction. FIG. 7 shows the contrast correction function.
Is the largest, and the reproducibility of the character image is good. However
To reduce the amount of illumination of the original due to deterioration of the light source over time.
Pay attention as the reading quality of the scanner unit deteriorates
Strong correction works on noise components superimposed on pixel E
However, a so-called black spot occurs in an area that should be a white background.
Background dirt may occur. So read the manuscript
The reading quality of the image sensor 9 is monitored every
MTF correction according to the reading quality of the image sensor 9
By changing the coefficient, due to aging after shipping
Prevent degradation of image quality. Also, for example, an MTF correction
If changing the number does not prevent the image quality from deteriorating,
The warning output unit 104 of the system control unit 1
Outputs a warning from the
A warning may be required. This allows the maintainer
To easily determine the necessity of installation and improve the reliability of equipment.
Can be. The matrix size of the MTF correction is 3
It is not limited to pixels x 3 pixels.
The number can also be set arbitrarily for the data in the matrix.
Is possible. The MTF correction calculation is performed by the system controller 1
May be performed, for example, as shown in FIG.
May be performed by the MTF correction calculation unit configured as described above. Figure smell
The matrix register size to 3 pixels x 3 pixels,
The target pixel E, and peripheral pixels of the target pixel E are A, B, C, D,
Expressed as F, G, H, I. Also, the density of each pixel A to I
A0, B0, C0, D0, E0, F0, G0,
H0 and I0 are represented by 5 bits. The MTF correction operation unit includes a first subtractor 21 and a second
Subtractor 22, third subtractor 23, fourth subtractor 24, first adder
An arithmetic unit 25, a second adder 26, a first multiplexer 27,
Second multiplexer 28, third adder 29, inverter
30 and a fourth adder 31. The first subtractor 21 is B
The difference (B0-E0) between the density of the pixel and the density of the E pixel is calculated. No.
The two subtractor 22 calculates the difference between the densities of the H pixel and the E pixel (H0−E
0) is calculated. The third subtractor 23 calculates the density of the D pixel and the E pixel.
The degree difference (D0-E0) is calculated. The fourth subtractor 24 is
The difference between the densities of the D pixel and the E pixel (D0-E0) is calculated.
The first adder 25 is a function of the first subtractor 21 and the second subtractor 22.
Add the calculation results and calculate (B0 + H0-2 × E0)
You. The second adder 26 includes a third subtractor 23 and a fourth subtractor 24.
Is added to calculate (D0 + F0-2 × E0)
I do. The first multiplexer 27 has a selection signal MTFSL0.
When the signal is 1, the operation result of the first adder 25 is output, and MT
When the FSL0 signal is 0, the operation result of the first adder 25 is
((B0 + H0) / 2-E0) is output. No.
The dual multiplexer 28 outputs the MTFSL1 signal, which is a selection signal, to 1
At the time, the operation result of the second adder 26 is output, and MTFSL1 is set to 0.
The result of the operation of the second adder 26 was shifted by 1 bit
((D0 + F0) / 2-E0) is output. Third adder
29 is the output of the first multiplexer 27 and the second multiplexer.
The output of the kusa 28 is added. Inverter 30 adds third
The result of operation of the unit 29 is inverted in bit units. Fourth adder
31 indicates the signal inverted by the inverter 30 and the density of the E pixel.
to add. When the MTF correction calculation unit is configured as described above,
Depending on the setting of the MTFSL0 and MTFSL1 signals
As shown in FIG. 9, the density E0 of the pixel E is calculated as E1 = 3 ×
E0− (B0 + D0 + F0 + H0) / 2, E1 = 4 × E
0− (2 × B0 + D0 + F0 + 2 × H0) / 2, E1 =
4 × E0− (B0 + 2 × D0 + 2 × F0 + H0) / 2,
E1 = 5 × E0− (B0 + D0 + F0 + H0) / 2 = 4
Can be corrected as follows. When the MTF correction is performed by the MTF correction calculation unit,
The operation in the case of FIG.
I will tell. As described in the flowchart of FIG.
Brightness output waveform characteristics and darkness output waveform characteristics stored in the characteristic storage unit 3
Is stored (step S21), the system control unit 1
S / N ratio calculating means 102 of S / N ratio = 20 × log
((Minimum value of light output waveform characteristics) / (dark output waveform characteristics
) Is calculated to determine the S / N ratio (step S
22). The main control unit 101 of the system control unit 1 has an S / N ratio
The S / N ratio calculated by the calculation means 102 is compared with a predetermined threshold value.
For example, if the S / N ratio is set to a threshold value of 20db,
If it is low (step S23), the selection signal MTFSL0
The signal and the MTFSL1 signal are set to 0, and the pixel E of interest is corrected.
As a formula for calculating the density E1, E1 = 3 × E0− (B0 + D
0 + F0 + H0) / 2 is selected (step S24).
A warning is displayed via the notification output unit 104 (step S2).
5). Thereafter, the original is read by the image sensor 9 (scan).
Steps S26 and S27), read from the image sensor 9
After the shading correction unit 13 corrects the
The density of the pixel of interest is E1 = 3 × E0−
(B0 + D0 + F0 + H0) / 2
Correct. As described above, when the S / N ratio is low, MTF compensation is performed.
The correction strength is reduced to prevent image quality deterioration.
Can be Next, the read image data is digitally converted.
Isolated pixel detection processing when processing is performed by the image processing unit 14;
The isolated block detection processing will be described. The isolated pixel detection pattern corresponds to 2 of the read image.
Isolated pixels of the image signal after binarization are detected by pattern matching.
It is used when issuing. For example, 3 pixels x 3 pixels
And the matrix configuration data is shown in FIG.
(A) and an isolated pixel detection pattern as shown in FIG.
Pixel of interest at the center of the matrix when matching
Is regarded as an isolated point, and as shown in FIG. 11B or FIG.
Invert the polarity as shown. For example, for facsimile machines
Is to compress the binarized image signal with a predetermined algorithm
And modulates the image signal and sends it out over the telephone line.
The rate greatly affects the transmission time, but white information or black information
The compression ratio increases as the continuity increases. So, like this
Transmission time by eliminating isolated points of data after binarization
Can be reduced. It is indicated by a black circle in the figure.
Pixels are black pixels, and pixels indicated by white circles are white pixels.
You. The reading quality of the image sensor 9 is
When it is reduced, isolated pixels are generated and the image quality is also deteriorated. There
As the reading quality deteriorates, the isolated pixel detection
Change the turn to increase the ability to detect isolated pixel blocks
This prevents significant image quality degradation due to background contamination.
Can be stopped. The isolated image shown in FIG.
Isolation by rotating the element detection pattern around the pixel of interest
A pixel detection pattern may be used. FIG. 11 (a)
In FIG. 12A, an isolated pixel detection in which a black pixel is isolated is performed.
Output pattern, but white pixels are isolated
Patterns may also be used. Further, the reading quality of the image sensor 9 is improved.
If the S / N ratio decreases, the background becomes dirty or white pixels
Frequency increases, and the ground
Skin stains or white pixels are generated. Background dirt
Or, if white pixels continue, for example, as shown in FIG.
In the isolated pixel detection pattern shown in FIG.
Or a blank pixel cannot be detected. So, for example
For example, a matrix of 6 pixels × 6 pixels is formed, and FIG.
And invert four adjacent pixels as shown in FIG.
Such an isolated pixel detection pattern may be used. What
Note that an isolated image obtained by rotating the isolated pixel detection pattern in FIG.
An element detection pattern may be used. Further, a matrix of 6 pixels × 6 pixels is formed.
4 pixels adjacent to each other as one block
Alternatively, an isolated pixel may be detected on a per-clock basis. This place
The operation in this case will be described with reference to the flowchart of FIG.
You. As described in the flowchart of FIG.
Brightness output waveform characteristics and darkness output waveform characteristics stored in the characteristic storage unit 3
Is stored (step S31), the system control unit 1
S / N ratio calculating means 102 of S / N ratio = 20 × log
((Minimum value of light output waveform characteristics) / (dark output waveform characteristics
) Is calculated to determine the S / N ratio (step S
32). Thereafter, the original is read by the image sensor 9.
The signal read from the image sensor 9
After shading correction by the shading correction unit 13 (step
Step S33), the digital image processing unit 14 calculates the S / N ratio
A predetermined first threshold value for the S / N ratio determined by stage 102, eg
S / N ratio obtained by comparing with 20db (step S34)
Is smaller than the first threshold, a matrix of 3 pixels x 3 pixels
And the isolated image shown in FIGS. 11 and 12, for example.
Detect isolated pixels using element detection pattern (step
S35). If an isolated pixel is detected, a digital image
The processing unit 14 inverts the polarity of the isolated pixel and outputs the detected isolated image.
Element is forcibly changed to a white pixel or a black pixel (step S
36). Next, the digital image processing unit 14 performs S / N
The S / N ratio obtained by the ratio calculation means 102 is set to a second threshold value which is predetermined.
The value is compared with, for example, 10 db (step S37). Asked
If the S / N ratio is smaller than the second threshold, the image sensor 9
Image reading quality has deteriorated.
The digital image processing unit 14 has 6 pixels ×
Construct a matrix of 6 pixels, 4 pixels adjacent to each other
Is divided into nine blocks as one block. Digital
The image processing unit 14 determines the number of white pixels and black pixels in each block.
Are compared, for example, if there are three or more black pixels in the block
If present, the block is a black pixel block, block
If there are three or more white pixels in the block,
White pixel block, two white and black pixels in the block
If there is a single element, judge as the same attribute as the previous block
Then, for example, as shown in FIG.
(Step S38). That
Thereafter, the digital image processing unit 14 performs the above-described 3 × 3 pixels.
The pattern is used in the same way as when the matrix of
To detect an isolated block (step S39).
As shown in the figure, all pixels in the block of interest are
Is changed to a black pixel (step S40). in this way,
For example, the first threshold is set to 20d as the threshold to be compared with the S / N ratio.
b, the second threshold value is set to 10db, and the number of isolated pixels depends on the S / N ratio.
And isolated block detection, isolated pixel detection
Processing to be performed, processing to detect isolated pixels and isolated blocks
Since three types of processing are performed, the reading of the image sensor 9 is performed.
If the captured image quality is good,
Read the image sensor 9 so as to reproduce the image
If the image quality is poor, remove background stains or white pixels.
It can be corrected. In the above embodiment, the characteristic detecting section 2
A large value detection unit 201 and a minimum value detection unit 202 are provided.
As shown in FIG.
Calculate the maximum and minimum values of the shape characteristics and dark output waveform characteristics
You may make it. When the maximum value is detected, the system control unit 1
For example, the held value of the 8-bit maximum / minimum value detection unit 203 is set to 00H.
Reset and maximum value to exclusive logic gate 204
/ Set the minimum value switching signal to H level. Then enter
The data is the maximum / minimum value detection unit 203 and the comparator 20
5 is input. Comparator 205 detects maximum and minimum values
When the input signal is larger than the held value of
Generates an inactive pulse signal. This pulse
Normally, glitches occur when input data is switched.
This is different from the switching of the input data as shown in FIG.
Sampling with a clock SCK that has a certain timing
After that, it is input to the latch of the maximum / minimum value detection unit 203. Sa
Input data that has occurred after sampling
And the maximum value is updated. When the minimum value is detected, the system control unit
1 sets the control signal to the exclusive logic gate 204 to L level.
Set. Thereby, the input and output of the exclusive logic gate 204 are output.
The polarity of the input data is reversed between the forces. That is, small
Data is converted to larger data,
As a result, the maximum value is detected using the
A small value will be detected. Of the maximum / minimum value detection unit 203
The polarity of the output data is inverted again by the exclusive logic gate 207,
Make the same polarity as the input data. With such a circuit configuration
2 circuits independent of maximum value detection and minimum value detection
It is not necessary to provide
Can be. In the above embodiment, the S / N ratio was calculated.
Unit 102 and parameter changing unit 103 as a system control unit
1, but may be provided outside the system control unit 1. As described above, the image processing of the present invention
The device is an image sensor when the white reference plate is exposed and scanned.
The maximum and minimum values of the output signal are used as the output waveform characteristics during light.
Turn off the light source and turn on the white light
Maximum value of the image sensor output signal when scanning the reference plate
And dark output to store the minimum value as dark output waveform characteristics
Waveform characteristic storage means for storing the stored bright output waveform characteristics and darkness
S / N ratio to calculate and store S / N ratio from time output waveform characteristics
Storage means, and a plurality of threshold values for which the stored S / N ratio is predetermined
After binarization of the read image according to the comparison result
Divide signal pixels into blocks consisting of adjacent pixels
And isolated blocks are detected by pattern matching.
And image processing means for performing correction. Therefore, the image
Even if the sensor output has a lot of noise components
It is possible to prevent a decrease in image quality due to background contamination or the like.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a configuration diagram of an image processing apparatus according to the present invention. FIG. 2 is a flowchart illustrating a schematic operation of the image processing apparatus. FIG. 3 is a configuration diagram of a matrix register. FIG. 4 is an explanatory diagram illustrating an example of weighting for MTF correction. FIG. 5 is an explanatory diagram showing an example of weighting of MTF correction having a high correction strength in the sub-scanning direction. FIG. 6 is an explanatory diagram illustrating an example of weighting MTF correction with a high correction strength in the main scanning direction. FIG. 7 is an explanatory diagram showing a weighting example of MTF correction having a high correction strength. FIG. 8 is a configuration diagram of an MTF correction calculation unit. FIG. 9 is an explanatory diagram illustrating a relationship between a selection signal and an arithmetic expression for MTF correction. FIG. 10 is a flowchart illustrating an operation of an MTF correction calculation unit. FIG. 11 is a configuration diagram of an isolated pixel detection pattern that differs only in a pixel of interest. FIG. 12 is a configuration diagram of an isolated pixel detection pattern in which a target pixel protrudes and is different. FIG. 13 is a configuration diagram of an isolated pixel detection pattern in which four central pixels are different. FIG. 14 is a configuration diagram of an isolated pixel detection pattern in which four central pixels protrude and are different. FIG. 15 is a flowchart illustrating an operation of detecting an isolated block. FIG. 16 is a configuration diagram of a 6 × 6 pixel matrix register. FIG. 17 is a configuration diagram of a block. FIG. 18 is a configuration diagram of a 6 × 6 pixel matrix register after correction. FIG. 19 is a configuration diagram of a circuit that detects a maximum value and a minimum value. FIG. 20 is a timing chart when a maximum value is detected. FIG. 21 is a timing chart when a minimum value is detected. [Description of Signs] 1 System control unit 102 S / N ratio calculation unit 103 Parameter change unit 2 Characteristic detection unit 3 Characteristic storage unit 301 Bright waveform characteristic storage unit 302 Dark waveform characteristic storage unit 5 Light source 6 White reference plate 9 Image Sensor 10 Amplifier 12 A / D converter 13 Shading correction unit 14 Digital image processing unit

──────────────────────────────────────────────────続 き Continued on front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N 1/04 H04N 1/407

Claims (1)

(1) An image processing apparatus having a light source for exposing an original and an image sensor for converting an optical image signal by reflected light from the original into an electric image signal in pixel units, wherein a white reference plate is provided. A light output waveform characteristic storage means for storing the maximum value and the minimum value of the image sensor output signal when the image is exposed and scanned as a light output waveform characteristic, and the image sensor output when the light source is turned off and the white reference plate is scanned. Dark output waveform characteristic storage means for storing the maximum value and the minimum value of the signal as dark output waveform characteristics; and S based on the stored light output waveform characteristics and dark output waveform characteristics.
S / N ratio storage means for calculating and storing the / N ratio, comparing the stored S / N ratio with a plurality of predetermined thresholds, and determining the pixels of the binarized image signal of the read image in accordance with the comparison result. An image processing apparatus comprising: an image processing unit that divides a block into a plurality of adjacent pixels and detects and corrects an isolated block by pattern matching.