JPH03145271A - Sensitivity correction method for image sensor - Google Patents

Abstract

PURPOSE:To correct the dispersion in the sensitivity characteristic of each picture element with high accuracy over the entire region of an incident luminous quantity even when each picture element has a nonlinear sensitivity characteristic by correcting the sensitivity characteristic of each picture element even at an intermediate gray region. CONSTITUTION:Output signals of each picture element of an image sensor corresponding to a white reference density W, a black reference density B and gray reference densities G1, G2,..., Gn, that is, a white reference density data f(W), a black reference density data f(B) and gray reference density data f(G1), f(G2),..., f(Gn) are obtained. Among which of the reference density data f(B), f(G2),..., f(Gn), f(W) a picture signal level f(S) of each picture element resides is discriminated. When two reference density data having a picture signal level f(S) of each picture element inbetween is discriminated, the picture signal f(S) of each picture element is corrected based on the reference density data and the relation between a known reference density corresponding to each reference density data. Thus, the dispersion in the sensitivity characteristic is corrected with high accuracy.

Description

[Detailed Description of the Invention] <Industrial Application Field> The present invention is applicable to, for example, a CCD (charge couple).
The present invention relates to a method for correcting variations in sensitivity characteristics of each pixel of an image sensor configured with an image sensor such as an ed device.

<Prior Art> -a. There are variations in the sensitivity characteristics of each pixel constituting an image sensor, and even if uniform light is incident on all pixels, the output level of each pixel will not be uniform. Conventionally, in order to correct variations in sensitivity characteristics between pixels of such an image sensor, methods such as those described in, for example, Japanese Patent Application Laid-open No. 61-53868 have been proposed.

In this method, the sensitivity characteristic of any pixel N in the image sensor is expressed by the following formula rs (X) = aN
・x+bs ・・・・・・■However, aNl
bN is a constant, X is given by the amount of incident light, and it is based on the premise that since a N l b s differs for each pixel, variations in sensitivity characteristics occur between pixels. FIG. 5 shows pixels 1, 2, . . . expressed by such a -order function. ..., sensitivity characteristic f1 ft of N,
.

f(W) -f(B) where f(-> is the output reference density data of the image sensor corresponding to the predetermined white reference density W, f(B)
is the output reference density data of the image sensor corresponding to the black reference density B, f (S) is the output image data corresponding to the amount of incident light S when reading the original, and S' is the output after correction.

Figure 6 shows pixels 1.2..., N
The sensitivity characteristics f, , f, . . . fl, / are shown.

Such correction by standardizing sensitivity characteristics is realized by an apparatus as shown in FIG.

That is, before reading and scanning the original PS, the black reference density B is
A black reference plate PB that gives The black reference density data f (B) is stored in the black reference line memory 13B. Next, white base f1! The white reference plate PW that provides the density is similarly scanned and processed, and the white reference density data f(W) of each pixel at that time is stored in the white reference line memory 13W.

When the above preparations are completed, the original PS is scanned and one line of image data f (S) at that time is stored in the signal line memory 13S. And each line memory 13
B, 13W, 135, each data of the corresponding pixel is read out in order, and each output of the line memories 13W, 13B is sent to the subtracter 14a, and the line memories 13B, 13S are sent to the subtracter 14a.
Each output is input to the subtracter 14b.

From the subtractor 14a, the difference data f (W) f for each pixel
(B) is output, and this difference data is given to ROM table ]5. The ROM table 15 is a conversion table that converts input data into a reciprocal number. The reciprocal number 1/(
f(W) −f(B) ) is given to the multiplier 16 as one input. On the other hand, the subtracter 14b outputs difference data f(S)-f(B) for each pixel, and this difference data is given to the multiplier 16 as the other input. As a result, the multiplier 16 outputs a normalized and corrected signal expressed by the above equation (2).

<Problems to be Solved by the Invention> However, the conventional example described above has the following problems.

That is, the sensitivity characteristic of each pixel of the image sensor is not necessarily expressed by a linear function as described above, but may have a nonlinear characteristic as shown in FIG. 8, for example.

If such nonlinear variations in sensitivity characteristics are normalized using the conventional example, an error occurs in the intermediate region between the white reference density W and the black reference density B, as shown in FIG.

In the case of commonly used CCD image sensors,
Linearity of sensitivity characteristics is guaranteed up to about 1/100 of the saturated output, but below that, linearity is not guaranteed.
For example, as shown in FIG. 10, it is often nonlinear on the low incident light amount (black) side.

In prepress scanners that read images using a CCD image sensor, signals in a level area of 1/100 or less of the saturation output (area corresponding to the shadow part of the original) are also used as valid signals. Since the image signal is emphasized logarithmically, a deviation from the linearity of the sensitivity characteristic in the shadow region causes a significant deterioration in the image quality of the printed matter.

The present invention has been made in view of the above circumstances, and provides an image sensor sensitivity correction method that can accurately correct variations in sensitivity characteristics of each pixel of an image sensor over the entire range of incident light amount. The purpose is to

<Means and Effects for Solving the Problems> The method of the present invention will be specifically explained below with reference to FIGS. 1 and 2.

For example, each pixel of a one-dimensional image sensor consisting of N pixels has a nonlinear sensitivity characteristic f1 as shown in FIG. fz,
..., fM. The horizontal axis in FIG. 1 is the amount of incident light (that is, the density) S, and the U axis is the output density data f of each pixel.

In order to correct such nonlinear sensitivity characteristics, first,
Predetermined white reference density W, black reference density B, and at least one gray reference density (in the example of FIG. 1, C,
, C, . . . , G, l) of each pixel of the image sensor, that is, white reference density data f(W)
, black reference density data f(B), and each gray reference density data f(Gl) r(cz)..., f(G,)
seek. Although the sensitivity characteristics of each pixel are non-linear, they each increase monotonically, so each of the above reference density data has the following magnitude relationship.

f(B) < f(Gl) <f(Gri<・< f(G
,,) < f (W) Next, the image signal level of each pixel f (S) when the original is read by the image sensor
is the at least three reference density data corresponding to each pixel (in the example of FIG. 1, r(B), f(G
l), r(Gz)f(G, ), f(W)
) between which reference concentration data is present.

Then, when the two reference density data sandwiching the image signal level f (S) of each pixel are known, each pixel is determined based on the relationship between these reference density data and the known reference density corresponding to each reference density data. The image signal "(S) is trimmed.

Specifically, the standardized corrected image signal S' is obtained by using the following correction calculation formula.

(1) f (B) ≦ f　(S) Ku When f(Gl), (2) r(c+)≦ f　(S) < When f(G2), (3) f　(Gz)≦ “(S) Ku When r(cz), (n) f(G, l )≦ f　(S) < f(G,) When, (n+1) f(G, l) ≦ f　(S) ≦ f (W) When, Reference concentration B.

1 G.

G.

W is known So, Above formula (1) ~ (n+1) In (G.

-B)/ (W-B) (Gz -G.

)/(W B), ... What is It is given as a preset constant.

FIG. 2 shows the sensitivity characteristics f, , f, . . . (, I) of each pixel standardized by the correction process as described above.

<Example> Hereinafter, an example of the present invention will be described in detail based on the drawings.

The apparatus according to this embodiment corrects variations in the sensitivity characteristics of, for example, a CCD image sensor, and in particular, two gray standards are provided between a black reference density and a white reference density as reference densities for sensitivity correction. Concentration is set.

This will be explained below with reference to FIG. FIG. 3 is a block diagram showing a schematic configuration of a sensitivity correction device for a CCD image sensor according to an embodiment.

The analog output signal of the CCD image sensor 11 consisting of N pixels is converted into a digital signal by the A/D converter 12, and then sent to five line memories 13 (13B).

130,. 1302.13W, 13S). Each line memory is CC
Each has a capacity sufficient to store one line of digital data from the D image sensor 11. Of these, the line memory 13B stores black reference density data f (
B), the line memory 13Cz stores the first gray standard density data f(Gt) corresponding to the first gray standard Y$ density G1, and the line memory 13G stores the second gray standard density data f(Gt) corresponding to the second gray standard value density G2. The line memory 13W is for storing the density data f (Gz), the line memory 13W is for storing the white reference density data f (W) corresponding to the white reference density W, and the line memory 13S is for storing the original image data f (S). be.

Line memory 13 selected by selector 17a
The reference density data for one line in B, 13c;l, and 13G2 is given as a negative input to subtracters 14a and 14b, respectively. Further, the reference density data for one line in the line memories 13G1, 13Gt and 13W, selected by the selector 17b, is given to the subtractor 14a as normal data. One line of image data in the line memory 13S is given as a positive input to the subtracter 14b.

The difference data output from the subtracter 14a is given to the ROM table 15. The ROM table 15 performs reciprocal conversion on this differential data, and the converted output is given to the multiplier 16 as one input. On the other hand, the difference data of the subtracter 14b is given to the multiplier 16 as the other input.

The output signal of multiplier 16 is given to ROM table 1. The ROM table 18 performs correction conversion processing as will be described later on the input signal, and generates a corrected image signal S'.
Output.

The comparator 19 receives the image data f(S) of each pixel given from the line memory 13S and the other line memory 13B.
, 13G+ , 13c;t , reference density data r(u) corresponding to each pixel given from 13W,
r(Gt), r(G, ), and f(W) to determine which base Y$ density data the image data exists between, and according to the result, the selector 17a , 17b and the conversion process of the ROM table 18.

The operation of the apparatus of this embodiment will be specifically explained below.

First, original PS'4: In preparation before reading and scanning, in order to collect reference density data used for calculation processing of sensitivity correction, black reference plate PB, white reference plate pw, first gray reference plate PG1, first gray reference plate PG1, The density of each reference plate (that is, the amount of light incident on the CCD image sensor If) for reading the two-gray reference plate PG2 is measured in advance. 1st
The setting values of the reference densities of the second gray reference plates PC, PC, are not particularly limited, but are preferably set as follows.

As shown by the solid line in FIG. 4, the sensitivity characteristics of the CCD image sensor 11 generally tend to deviate nonlinearly to a greater extent on the black (B) side than on the white (W) side. therefore,
At least one gray reference density is preferably set between the midpoint N between the black reference density B and the white reference density W and the black base density B (region G in FIG. 4).

More preferably, the point at which the sensitivity characteristic of the CCD image sensor 11 deviates the most from the linear characteristic connecting the black reference density B and the white reference density W with a straight line (point 00 in FIG. 4) is measured in advance, and at least One gray standard is set at the 00 point.

Each reference plate PB read by the CCD image sensor 11
, PC, PG, , PW reference concentration data f(B)
, f(G,), r(Gt), f(W) are
The above-mentioned predetermined line memories 13B, 13G, 13G
t and 13W, respectively.

When the collection of the reference density data is completed, the image data f (S) of the original PS is read and stored in the line memory 13s. Then, data of corresponding pixels is read out in order from each line memory 13 and outputted to the comparator 19.

The comparator I9 receives the image data f (S) read from the line memory 13s and the other line memo 1 river 3B.

13C;, , 13G! , the reference density data r(B) of the corresponding pixel read from 13W,
r(c+).

rL) and f(W), f(B) < f
In the relationship (Gt)<r (Gt)<f (W), it is determined which reference density data the image data "(S) is between, and the selectors 17a and 1
7b and the ROM table 18 are controlled as follows.

(1) When f(B)≦f(S)<f(Gl), the selectors 17a and 17b are switched to the terminal a side, and the selector 17a receives the black reference density data f(B), and the selector 17b receives the black reference density data f(B). 1 gray standard density data r(c,) is selected. On the other hand, the ROM table 18 uses the operator rX (C,, -B) / (
W-B)".

(2) r(c+)≦f(S)<f(Gt)(7
), the selectors 17a and 17b are switched to the terminal side, and the selector 17a selects the first gray reference density data f.
(Gl) and the selector 17b selects the second gray reference density data r(Gx), respectively. On the other hand, the ROM table 18 operates the operator 1'×(G2
CI)/(WB)+ (G,-B)/(WB)
It is controlled to act on J.

(3) When f(Gt)≦f(S)<f(W), the selectors 17a and 17b are switched to the terminal C side, and the selector 17a selects the second gray reference density data r(c
z), and the selector 17b selects the white reference density data f (W).
Select each. On the other hand, the ROM table 18 is
For the input signal, the operator rX (W Gz ) /
(W B) + CGt B)/ (W B)
It is controlled to act on J.

In the following explanation, the image signal f (S) is defined as r (Gt)≦
Let us take as an example the case where F(S) < f(Gt).

In this case, as described above, the selector 17a selects the first gray reference density data r(c,), and the selector 17b selects the second gray reference density data r(c,), so the subtractor 14a selects the difference Data f(Gz)-f(G
, ) is output, and the subtracter 14b outputs the difference data fcs
> −f(Gl) is output.

RO given differential data f(Gz) f(Gl)
The M table 15 is the reciprocal 1/(r(cz)-f
(Gl)), which is supplied to the multiplier 16.

The multiplier 16 calculates 1/(f(Gg) f(Gl)) and the difference data f(S) − given from the subtracter 14b.
f(Gl) and its output (f(S) - f(G
l))/(f(Gri−f(G,)) is given to the ROM table 18.

The ROM table 18 has its input (f(S) −f(G
l)) / (fcGx) fcGl)),
Operator rX (Gg −Gl )/(W−B)+
As a result of acting (G, -B)/(W-B)J, RO
The M table 18 outputs a corrected image signal S' expressed by the following equation.

Similarly, if the image signal f (S) is f (B)≦f(S)
< f (G1), the corrected image signal f (S) expressed by the following formula is f (h)≦fcs)
< f(W), the corrected image f(W)-r(Gx) W-B expressed by the following formula
WB Note that the present invention can also be modified as follows.

■ In the above embodiment, two gray reference densities G, , G It may be possible to improve the

■ To simplify the circuit configuration of the device, one gray reference density may be set. In this case, as mentioned above, the gray reference density is set to the midpoint between the black reference density and the white reference density, and the black reference density. It is preferable to set it between the white reference density and the ring reference density, and more preferably set it at a point where the sensitivity characteristics of the image sensor deviate the most from the straight line connecting the white reference density and the ring reference density.

■ In the above embodiment, in order to obtain the output signal (reference density data) of each pixel of the image sensor corresponding to each predetermined reference density, each density reference plate of white, black, and gray is used with the image sensor. However, for example, each density reference data may be obtained by using one white reference plate and changing the amount of light from the light source.

■ In the case of a CCD image sensor with logarithmic characteristics, after A/D converting the logarithmic output, convert it so that it has a linear relationship with the amount of incident light using a ROM table, etc., and then convert the sensitivity as in the above example. Correction may also be made. Alternatively, the sensitivity correction described above may be directly performed on the relationship between the logarithmic light amount and the CCD output without performing such linear conversion, so that a corrected logarithmic output can be obtained. .

(2) Some image sensors, for example, transfer and output the outputs of odd-numbered pixels and even-numbered pixels individually. In the case of such a separate output type image sensor, since it has two signal processing systems, the manner of deviation in sensitivity characteristics may differ between odd-numbered pixels and even-numbered pixels. In such a case, separate gray bases f! for the outputs extracted from odd-numbered pixels and the outputs extracted from even-numbered pixels, respectively. A? ! The sensitivity may be corrected by setting 4 degrees.

<Effects of the Invention> As is clear from the above explanation, according to the image sensor sensitivity correction method according to the present invention, the sensitivity characteristics of each pixel are
It not only corrects on the white side and black side, but also on the intermediate gray area, so even if each pixel has non-linear sensitivity characteristics, variations in the sensitivity characteristics of each pixel can be accurately corrected over the entire range of incident light amount. can do.

Also, the gray standard density is the mastermind! Jl - When the density is set between the midpoint between the density and the white reference density and the black reference density, or when the sensitivity characteristics of the image sensor deviate the most from the linear characteristics connecting the black and white reference density with a straight line. If the gray reference density is set to a certain value, relatively accurate correction can be performed using a small number of gray reference densities.

[Brief explanation of the drawing]

1 and 2 are explanatory diagrams of the method of the present invention, and the first
The figure is an explanatory diagram of the nonlinear sensitivity characteristics of an image sensor.
The figure is an explanatory diagram of the corrected sensitivity characteristics. FIG. 3 is a block diagram showing a schematic configuration of a sensitivity correction device for an image sensor according to an embodiment of the present invention, and FIG. 4 is an explanatory diagram of a preferred setting location of a gray reference density. Figures 5 to 10 relate to the explanation of conventional examples. Figure 5 is a diagram showing the sensitivity characteristics of the image sensor as a linear function, Figure 6 is an explanatory diagram of the corrected sensitivity characteristics, and Figure 7 is the conventional example. A block diagram showing the schematic configuration of the device, FIG. 8 is a diagram showing the nonlinear sensitivity characteristics of the image sensor, and FIG. 9 is an explanatory diagram of problems when the sensitivity characteristics shown in FIG. 8 are corrected using a conventional example. FIG. 1O is an explanatory diagram of nonlinear sensitivity characteristics of a general image sensor. 11... CCD image sensor 12... A/D converter 13B... mastermind (line memory 1 for storing data)
3G, . . . line memory 13G for storing first gray reference density data, . . . line memory 13W for storing second gray reference density data, . . . line memory 1 for storing white reference density data.
3S...Line memory 14 for storing image data of the original
a, 14b...Subtractor 15, IF3...ROM table 17a, 17b...Selector 19...Comparator

Claims (3)

[Claims] (1) The output signal of each pixel of the image sensor corresponding to predetermined white reference density, black reference density, and at least one gray reference density is converted into white reference density data, black reference density data, and gray reference density data. determining which reference density data among the at least three reference density data corresponding to each pixel the image signal level of each pixel when the original is read by the image sensor falls between; Sensitivity of an image sensor characterized in that the image signal of each pixel is corrected based on the relationship between two reference density data sandwiching the image signal level of each pixel and a known reference density corresponding to each reference density data. Correction method. (2) In the image sensor sensitivity correction method according to claim (1), at least one gray reference density is set between a midpoint between the black reference density and the white reference density and the black reference density. Image sensor sensitivity correction method. (3) In the image sensor sensitivity correction method according to claim (1), the at least one gray reference density is such that the sensitivity characteristic of the image sensor is the most from the linear characteristic connecting the black reference density and the white reference density with a straight line. A method for correcting the sensitivity of an image sensor that is set at a large deviation.