JP2000115791A - Method for calculating color value in color image recorder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain high image quality by interpolating a color value of each pixel based on a color value of a same kind detected from a plurality of pixels placed on each occasion in a pixel surrounding area through low pass filtering to correct color slurring. SOLUTION: Original color values R0, G0, B0d of a pixel detected by the image recorder 9 are given to color channel interpolation stages 11R, 11G, 11B acting like a low pass filter via a color channel changeover device 10, from which color values RTP, GTP, BTP that are low-pass-filtered are produced and fed to a differential forming stage 12, where differences DR, DG, DB with respect to the original color values R0, G0, B0 are formed. Then difference DR, DG, DB that are processed by high pass filtering and include only contour lines of image information are given to color channel adder stages 13R, 13G, 13B, where they are selectively added to the color values RTP, GTP, BTP to obtain color values R, G, B whose sharpness is compensated. The color values R, G, B are tentatively stored in a color value memory 14 for succeeding processing.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to the field of electronic duplication technology, and more particularly to a method for calculating color values in a color image recording device.

[0002]

2. Description of the Related Art A color image to be reproduced is projected onto a light-sensitive surface of a color image recording apparatus, for example, a digital color camera, using an objective lens. A color image recorder consists in principle of a CCD sensor array, which has a large number of adjacently arranged color sensors for "red", "green" or "blue", respectively. It has a sensor element. The sensor element converts the corresponding color components "red", "green" and "blue" of the pixels of the color image projected on the sensor element into corresponding color values R, G or B. Since the individual sensor elements of the CCD sensor element are sensitive to only one color component, each individual sensor element forms only one of the three color values R, G or B of the pixel. Is done. As a result, an obstructive color stripe (F
arbsaeume).

In order to correct a color shift, a missing color value for each sensor element is calculated from color values detected by surrounding sensor elements. For example, only the color value B is detected by the sensor element, and the missing color values R and G are
It is calculated from the color values R and G detected by the surrounding sensor elements.

A method of calculating a color value in a digital color camera is described in a publication "Kodak DCS420 DigitalCamera System Progr."
ammer's Reference Manual ", in which the color value of a sensor element is calculated as the average of all the color values of horizontally or vertically adjacent sensor elements. May not be completely deterred.

[0005]

SUMMARY OF THE INVENTION It is an object of the present invention to improve a method of calculating color values in a color image recording apparatus, to substantially correct an obstructive color shift, and to obtain a high image quality.

[0006]

According to the present invention, there is provided a method for calculating a color value in a color image recording apparatus provided with a sensor array having a large number of sensor elements arranged adjacent to each other. Projecting onto a sensor array, each sensor element being assigned a pixel of the color image, wherein each sensor element detects only one of the three color components as the color of the assigned pixel. In the color value calculation method of calculating at least the color value of the color component not detected for each pixel from the same kind of color value detected by another sensor element, the color value for each pixel is The problem is solved by a color value calculation method characterized in that low-pass filtering is used to interpolate the detected same color values of a plurality of pixels located each time in the peripheral area of the pixel.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below in detail with reference to FIGS.

FIG. 1 is a diagram showing a basic structure of a color image recording apparatus provided with a two-dimensional CCD sensor array 1. The CCD sensor array 1 includes a large number of sensor elements 2 having different color sensitivities arranged in rows and columns adjacent to each other, and the color components of the pixels projected on the CCD color array 1 are provided. Convert "red", "green" or "blue" to the corresponding color values R, G or B.

In the illustrated embodiment, the CCD sensor array 1 has 1023 × 1535 sensor elements 2. The color sensitivity of each of the sensor elements 2 is “R” for red, “G” for green, and “B” for blue.
Are indicated by. It can be seen that a plurality of sensor elements 2 having different color sensitivities are staggered. The sensor element block 3 having 2 × 2 sensor elements “G”, “R”, “B” and “G” is repeated in the row and column directions. A large number of CCD sensor elements 1 have sensor elements 2 having sensitivity to green. This is because the luminance of the color image is additionally detected for the color component “green”.

The color values R, G detected by the sensor element 2
Alternatively, B is first temporarily stored in a color value memory for addressing individual pixels of the recorded color image for subsequent processing.

Each of the sensor elements 2 has three color components "red" and "green" of a pixel according to its color sensitivity.
Or, only one of "blue" is detected, thus forming only one color value of the pixel. But the color of the pixel is
It is characterized by three color values R, G and B, resulting in a disturbing color shift in the recorded color image.

In order to improve the image quality, a color shift which is an obstacle is corrected. This is done by interpolating at least the missing color values R, G or B for the individual pixels of the color image. Here, this interpolation is performed from a value obtained by actually measuring the same color value of a pixel in a peripheral region surrounding the pixel by a measurement technique, and low-pass filtering is used.

FIG. 2 partially shows, in rows and columns, the correspondence between the color values R, G or B and the pixels 4 of the color image in rows and columns. The rows and columns follow the arrangement of sensor elements 2 with different color sensitivity in the CCD sensor array 1. The correspondence between the color values and the pixels corresponds to the correspondence between the color values in the color value memory and the memory locations. The position of each pixel 4, which is partially indicated by 10 × 10 pixels 4, is determined by the row number and the column number.

The pixel block 5 corresponding to the sensor element block 3 composed of 2 × 2 pixels 4 is repeated in the row and column directions. The pixel block 5 has a first pixel at a position (0/0) having an even row number and an even column number, and a second pixel at a position (0/1) having an even row number and an odd column number.
Pixel, a position (1) having an odd row number and an even column number.
/ 0), and a fourth pixel at a position (1/1) with an odd row number and an odd column number. The four pixels correspond to the color values G ₀₀ , R ₀₁ , B ₁₀ and G ₁₁ actually detected by the sensor element 2.

In order to interpolate the color value R, G or B of the current pixel 4 from the same kind of color value detected by the measurement technique, an interpolation window 6 is first defined around the current center pixel 4. The interpolation window 6 includes a plurality of pixels 4 for each color component.
Is equal to the number of detected color values of the same color component to be used to interpolate the corresponding color values of the current pixel 4.

Furthermore, in the interpolation window 6, for each of the three colors and for each position of the four pixels 4 in the pixel block 5, a coefficient matrix with a weighting factor K is calculated. Here, each coefficient matrix for each pixel 4 includes, within the interpolation window 6, a weighting coefficient K that determines a characteristic. Here each pixel 4
Weighting factor K takes into account the respective distance between this pixel and the central pixel whose color values are to be interpolated. The calculation of the weighting factor K is performed by a two-dimensional weighting function, preferably by a Bessel function. Thus, in the illustrated example, twelve coefficient matrices are formed, and these coefficient matrices are callably stored in the coefficient memory.

FIG. 3 shows a color value R, G or B and a pixel 4
And the interpolation window 6 correspond to the position of the color value G ₅₅ (5 /
5) is partially shown for the center pixel. In the illustrated example, the interpolation window 6 has 9 × 9 pixels 4. In the pixel block 5 center pixel color value G ₅₅ is at the position (5/5), further pixel position of the color values _{G 44, R 45, B 54} (4/4), (4/5) and (5/4).

FIG. 4a shows a coefficient matrix for the color component "red" with a 9.times.9 coefficient field. A weighting coefficient K _RXX calculated for interpolating the color value R is assigned to this coefficient field.

FIG. 4b shows a coefficient matrix for the color component "green" with a 9.times.9 coefficient field. A weighting coefficient K _GXX calculated for interpolating the color value R is assigned to this coefficient field.

FIG. 4c shows a coefficient matrix for the color component "blue" with a 9.times.9 coefficient field. Assigned to this coefficient field is a weighting coefficient K _BXX calculated to interpolate the color value R.

Shown are the colors "red",
Three coefficient matrices for "green" and "blue", and locations with odd row numbers and odd column numbers (5 /
5) Only pixel 4. The coefficient matrix for each position of the other three pixels 4 in the pixel block 5 is not shown.

In each of the coefficient matrices, the weighting coefficients assigned to the pixels whose corresponding color values R, G and B have already been detected in the measuring technique are set to zero.

To interpolate a color value R, G or B for pixel 4, one of the twelve previously calculated coefficient matrices is selected. This selection is made depending on the color and position of the corresponding pixel 4 in each pixel block 5. Here, the four position types A, B, C and D can be distinguished depending on whether the pixel 4 has an even or odd row number or an even or odd column number.

[0024]

[Table 1]

The interpolation of the color values is carried out by multiplying the same color value in the interpolation window 6 by the assigned weighting factor K and adding the color values thus weighted.

R _XX = Σ (K _RXX × R _XX ) G _XX = Σ (K _GXX × G _XX ) B _XX = Σ (K _BXX × B _XX ) The position (5/5) in the example shown in FIG. , The color value R ₅₅ , G ₅₅ or B ₅₅ for a pixel of position type D is calculated as follows:

R ₅₅ = K _DR45 × R ₄₅ + K _DR65 × R ₆₅ + ... G ₅₅ = K _DG55 × R ₅₅ + K _DG44 × G ₄₄ + ... B ₅₅ = K _DB54 × R ₅₄ + K _DR56 × R ₅₆ + ... Color image In order to interpolate all of the color values R, G, or B, the interpolation window 6 is moved from the pixel 4 to the pixel 4
Then, it shifts over the entire image and performs an interpolation calculation accordingly at the individual positions.

Interpolation of color values by low-pass filtering results in a loss of image sharpness based on filter characteristics. Such effects can advantageously be compensated for by unsharp masking in the form of high-pass filtering.

FIG. 5 shows low-pass filtered color values.
Unsharp mask to reproduce the sharpness of the image
1 shows a corresponding device for carrying. Image recording device
The original color value of the pixel, detected technically by 9
R_O, G_OAnd B_OThrough the color channel switch 10
Three colors for "red", "green" and "blue"
Channel three color channel interpolation stages 11R, 11G,
Reaches 11B. Three interpolation stages 11R, 11G, 11B
Acts like a low-pass filter,
The filtered color value R_TP, G_TPAnd B_TPThe shape
To achieve. Original color value R _O, G_OAnd B_OLow pass fill
Color value R that has been tapped_TP, G_TPAnd B_TPIs the difference stage 1
2 and the original color value R_O, G_O
And B_OAnd low-pass filtered color values
R_TP, G_TPAnd B_TPIs formed.

The difference values D _R , D _G, and D _B including only the contour line of the image information by the high-pass filtering are ₃ = D _R = R _O -R T _T D _G = G _O -G T _T D _B = B _O- B _TP The low-pass-filled color values R _TP , G _{TP with the} selectable component M in the adder stages 13R, 13G, 13B of the two color channels
And B _TP to obtain the color values R, G and B whose sharpness has been compensated.

R = R _TP + M _R (R _O −R _TP ) G = G _TP + M _G (G _O −G _TP ) B = B _TP + M _B (B _O −B _TP ) R, G and B are then temporarily stored in the color value memory 14 for subsequent processing.

[Brief description of the drawings]

FIG. 1 is a diagram of a CCD sensor array of a color image recording apparatus.

FIG. 2 is a diagram partially showing a correspondence between a sensor element of a CCD sensor array and a color value;

FIG. 3 is a diagram with an interpolation window, partially showing a correspondence between a sensor element of a CCD sensor array and a color value;

FIG. 4 is a diagram of a coefficient matrix for “red”, “green”, and “blue”.

FIG. 5 is a diagram of an apparatus for unsharp masking.

[Explanation of symbols]

 Reference Signs List 1 CCD sensor array 2 Sensor element 3, 5 Sensor element block 4 Pixel 6 Interpolation window 9 Image recording device 10 Color channel switch 11 Interpolation stage 12 Difference formation stage 13 Addition stage 14 Color value memory

Continuation of front page (71) Applicant 390009232 Kurfuersten-Angel 52-60, Heidelberg, Federal Republic of Germany many

Claims (11)

[Claims] 1. A method for calculating a color value in a color image recording apparatus provided with a sensor array (1) having a large number of sensor elements (2) arranged adjacent to each other, comprising the steps of: And the pixels (4) of the color image are assigned to the individual sensor elements (2), and the three color components (red,
Green, blue) as the color value (R, G, B) of the assigned pixel (4), and at least the undetected color component for each pixel (4) The color values (R, G, B) of the other sensor element (2)
In the color value calculation method of the form of calculating from the same kind of color value (R, G, B) detected by (1), the color value (R, G, B) for each pixel (4) is converted to the pixel (4, A color value calculation method characterized by interpolating the detected same-type color values (R, G, B) of a plurality of pixels (4) located each time in the peripheral area of 4). 2. An interpolation window (6) around a central pixel (4).
The interpolation window (6) includes a plurality of pixels (4) for each color component (red, green, blue), and the number of the pixels (4) is the corresponding color value of the pixel (4). (R,
G, B) equal to the number of detected color values (R, G, B) of the same color component (red, green, blue) to be used to interpolate; To interpolate the color values (R, G, B) for each pixel (4) of
2. The method of claim 1, wherein shifting is performed across the color image from pixel to pixel. 3. An interpolation window (6) to which color values (R, G, B) of the same color component (red, green, blue) for each pixel (4) are assigned, further weighting the coefficient matrix. 3. The method according to claim 1, wherein the coefficient (K) is calculated. 4. A color value combination assigned to a pixel group (5).
Is repeated in the image. For each pixel (4) of the pixel group (5), the coefficient matrix is
Rix is calculated for each color component (red, green, blue) Claim 3
The described method. 5. A weighting coefficient (K) of a coefficient matrix.
A method according to claim 3 or 4, wherein is calculated according to a two-dimensional weighting function, preferably a Bessel function. 6. The method according to claim 1, wherein the calculated coefficient matrix is stored callably. 7. A coefficient matrix required for interpolation of a color value (R, G, B) of a pixel (4) is calculated for each color component (red,
7. The method according to claim 1, wherein the selection is made depending on the position of the pixel (4). 8. A pixel (4) of a color image is associated with a row and a column, the position of the pixel (4) in the color image is determined by a row number and a column number, and a coefficient relating to the position of the pixel (4) is determined. 8. The matrix selection according to claim 1, wherein the selection of the matrix depends on whether the position of the pixel (4) is determined by an odd or even row number and by an odd or even column number. A method according to any one of the preceding claims. 9. The calculation of the color values (R, G, B) of the color components (red, green, blue) of the pixel (4) is performed by the same kind of color values (R, G, B) in the interpolation window (6). 9.) The method according to claim 1, wherein the weighted color values are multiplied by the assigned weighting factor (K) of the selected coefficient matrix and the weighted color values are added. 10. The method according to claim 1, wherein the low-pass filtered color values are high-pass filled to improve the sharpness of the image. 11. Difference values (D _R , D _G , D _B ) between original color values (R _O , G _O , B _O ) and low-pass filtered color values (R _TP , G _TP , B _TP ). And the low-pass-filtered color values (R _TP , G _TP , B _TP ) with the selectable intensity of the difference values (D _R , D _G , D _B )
The method according to any one of claims 1 to 10, further comprising: