JP3662949B2 - Signal interpolation method - Google Patents

Description

[0001]
[Industrial application fields]
The present invention relates to a signal interpolation method for interpolating a lack of a green signal due to a filter arrangement when photographing using a filter in which green filters are arranged in a checkered (mosaic) arrangement as a color filter of a two-dimensional image sensor. .
[0002]
[Prior art]
2. Description of the Related Art Conventionally, imaging devices such as a color camera using a two-dimensional image sensor such as a CCD or an imaging tube, and a camera-integrated VTR are often red (hereinafter referred to as R), green in consideration of economy and miniaturization. Instead of the so-called three-plate and three-tube type in which sensors are provided for each of the three primary colors (hereinafter referred to as G) and blue (hereinafter referred to as B), the sensor is configured as a single plate, single-tube type, or two-plate and two-tube type.
[0003]
In this case, a color filter in which G filters called G mosaic and Bayer array are arranged in a checkered pattern is provided on the light receiving surface of the sensor, and the imaging output of the sensor via this color filter is the G of a pixel that does not correspond to the G filter. The signal is missing and the luminance component is missing.
[0004]
In particular, when enhancing the contour of the image of the imaging output, improving the sharpness, etc., it is necessary to interpolate the missing G signal by some method to generate a luminance component.
Next, this type of conventional G signal interpolation method will be described with reference to the conventional configuration of FIG. 10 in which two-dimensional contour enhancement is performed by digital processing.
[0005]
First, the single-plate type two-dimensional image sensor 1 is provided with, for example, a color filter 2 having the structure shown in FIG. 2 of one embodiment of the present invention on its light receiving surface. The pixels are arranged in a checkered pattern, and the R filter 2r and the B filter 2b are alternately arranged for each horizontal line in the remaining pixels where the G filter 2g is not provided to form a so-called checkered pattern.
[0006]
The pixel signals of the filters 2r, 2g, and 2b of the imaging output of the sensor 1, that is, the R signal, the G signal, and the B signal (video signal) are subjected to preprocessing such as AGC and clamping by the preprocessing circuits 3r, 3g, and 3b. After being applied, the signals are converted into digital signals by the A / D converters 4r, 4g, and 4b.
Further, the output signals of the converters 4r, 4g, and 4b are supplied to the column circuits of the two line memories 5r, 5g, and 5b for delaying, for example, one horizontal scanning period (1H), and the time required for the processing of the subsequent circuit, respectively. Delayed.
[0007]
Then, the output signals of the converters 4r, 4g, 4b and the delayed signals of the line memories 5r, 5g, 5b are taken into the color processing circuit 6, and this processing circuit 6 is used as a combination of the pixel arrangement of the sensor 1 and the color filter 2. Signals along the lines are separated, combined, and filtered, and the G signals from which the pixels of the filters 2r and 2b are missing are interpolated by the interpolation processing in the vertical direction or horizontal direction described below.
[0008]
That is, when the pixel of the filter 2r or 2b and the interpolated pixel G _x as shown in FIG. 11, conventionally, a pixel G ₀ in the vertical direction of two neighboring this pixel G _x, G ₁ or horizontal both sides Rino noted pixel G _2, G _3, G signals of pixels G _x is interpolated by one-dimensional interpolation operation simple average or the like of the signals of pixels G _0, G ₁ or G _2, G _3.
[0009]
Further, R, G, B three primary color signals for each pixel are reproduced by the processing of the color processing circuit 6, and the three primary color signals are subjected to photoelectric conversion characteristic correction by the gamma correction circuit 7 and then applied to the matrix circuit 8. Supplied.
The circuit 8 matrix-processes the three primary color signals to form luminance signals (hereinafter referred to as Y signals) and RY and BY color difference signals.
[0010]
Further, the G signals {G-0}, {G0}, {G + 0} after interpolation processing shifted by 1H, for example, are supplied from the color processing circuit 6 to the contour emphasizing circuit 9, and the emphasizing circuit 9 is supplied by the vertical aperture creating circuit 10. The signals {G-0}, {G0}, {G + 0} having luminance correlation are differentiated by a desired order to form a vertical aperture signal S _v , and an appropriate order for the signal {G0} by the horizontal aperture creating circuit 11 The horizontal aperture signal _Sh is formed by applying the differential filter.
[0011]
Both aperture signals S _v and _Sh are supplied to and added to the adder circuit 12, and an output signal of the adder circuit 12 is supplied to the adder circuit 13 as an aperture signal for two-dimensional contour enhancement. The aperture signal is added to the Y signal of the matrix circuit 8 with an appropriate size to enhance the contour of the Y signal.
[0012]
[Problems to be solved by the invention]
In the case of the conventional signal interpolation method, only one-dimensional signal level correlation is considered, and one-dimensional interpolation is performed by a simple average of signals of pixels adjacent to each other in the horizontal direction or vertical direction of the interpolation target pixel. Depending on the brightness pattern of shooting, there is a problem that proper interpolation cannot be performed as described below.
[0013]
That is, when the signal of the interpolation target pixel G _x in FIG. 11 is interpolated by the simple average of the signals of the pixels G ₂ and G ₃ adjacent to each other in the horizontal direction, the images shown in (a) to (d) of FIG. Based on four typical patterns of brightness position changes, the signal levels of the pixels G ₂ and G ₃ are as shown in (e) to (h) of FIG.
[0014]
Incidentally, in FIG. 12 (a) ~ (d) show the difference in brightness with diagonal lines, is (a) shows a uniform brightness of the flat pattern, (b) the left portion comprising a pixel G ₂ The step pattern is darker than the other portions, (c) shows an impulse pattern whose central portion is brighter than the left and right portions including the pixels G ₂ and G ₃ , and (d) is the left portion including the pixel G ₂ and the pixels G _A slope pattern that becomes brighter in the order of the central portion including ₀ and G ₁ and the right portion including the pixel G ₃ is shown.
[0015]
12 (e) to 12 (h) show the levels of the pixels G ₂ and G ₃ based on the patterns (a) to (d) and the simple average of both levels, where the horizontal axis is the position and the vertical axis is the signal level. actual interpolation level of the pixel G _x obtained from, shows a proper level of the pixel G _x, ▲ mark real interpolation level of the pixel G _x, ● mark is proper level of the pixel G _x.
[0016]
As is apparent from FIGS. 12E to 12H, when the brightness of the step pattern or impulse pattern changes, the actual interpolation level greatly deviates from the appropriate level, and proper interpolation cannot be performed. .
[0017]
In this case, the step pattern of FIG. 12 (f), (g) , when subjected to edge enhancement using the signal of the interpolated pixel G _x of the impulse pattern ▲ mark, the enhancement after the signal if the step pattern A false edge, moire, etc. occur in the reproduced image based on it, and if it is an impulse pattern, no emphasis is applied and the frequency characteristics deteriorate.
[0018]
12 (i) to 12 (l) show the optimum interpolation patterns for the brightness patterns (a) to (d) in FIG. 12, with the horizontal axis as the position and the vertical axis as the level. The Δ mark is the pixel G _x interpolated to an appropriate level.
Further, when the signal of the interpolation target pixel G _x is interpolated by the simple average of the signals of the pixels G ₀ and G ₁ adjacent to each other in the vertical direction, the case of the horizontal direction as the image brightness changes. As with, a situation occurs in which proper interpolation cannot be performed.
[0019]
In addition, instead of the simple average of two adjacent pixels, it may be possible to interpolate using, for example, the average of a certain range of pixels in the horizontal direction by filtering or the like. Therefore, proper interpolation cannot be performed.
An object of the present invention is to dramatically improve interpolation accuracy and perform proper interpolation when interpolating missing G signals necessary for contour enhancement or the like.
[0020]
[Means for Solving the Problems]
In order to achieve the above object, in the signal interpolation method of the present invention, pixels that do not correspond to the G filter of the imaging output of the two-dimensional image sensor in which the G filter is arranged in a checkered pattern in units of pixels and the color filter is configured. When interpolating the missing G signal with the imaging output,
A predetermined two-dimensional interpolation unit area centered on the interpolation target pixel lacking the G signal is set in advance;
[0021]
The level change of the G signal in the horizontal direction and the vertical direction around the interpolation target pixel in the interpolation unit area is classified into a plurality of two-dimensional change patterns, and an optimal interpolation operation of the interpolation target pixel of each pattern is set in advance.
For each pixel to be interpolated in the imaging output, a pixel in the interpolation unit area centered on this pixel is cut out,
[0022]
From the signal level of each pixel of the G filter in the extracted interpolation unit area, the change in the G signal level in the horizontal and vertical directions centering on the interpolation target pixel is determined to identify the corresponding two-dimensional change pattern,
The G signal of the interpolation target pixel is interpolated and generated by the optimum interpolation operation corresponding to the specified two-dimensional change pattern.
[0023]
[Action]
In the case of the signal interpolation method of the present invention configured as described above, a two-dimensional interpolation unit region is set in advance, and the level change of the two-dimensional G signal centered on the interpolation target pixel in this region is patterned. Classification is performed, and an optimum interpolation operation for each pattern is set in consideration of a two-dimensional signal level correlation.
[0024]
Then, for each actual interpolation target pixel of the imaging output of the image sensor, a pixel in the interpolation unit area centered on this pixel is cut out, and the interpolation target pixel is centered from the actually obtained signal in that area. A two-dimensional change pattern of the G signal level in the horizontal and vertical directions is specified.
[0025]
Further, based on the specification of this pattern, the G signal of the interpolation target pixel is interpolated by the optimum interpolation operation corresponding to the specified pattern.
At this time, since the optimum interpolation operation is selected in consideration of the two-dimensional level change of the G signal, interpolation with extremely high accuracy can be performed.
[0026]
【Example】
One embodiment will be described with reference to FIGS.
In this embodiment, for example, an output signal of the A / D converter 4g in FIG.
[0027]
In this case, the image sensor 1 includes the color filter 2 shown in FIG. 2, and the output signal of the converter 4g is lost when the R and B filters 2r and 2b of the color filter 2 are used.
Then, the horizontal direction around the pixel G _x in FIG. 11, the change pattern in the vertical direction of the G signal level prediction in FIG. 12 (a) ~ brightness pattern similar brightness and the (d) Assuming various patterns, as shown in FIG. 3, one type of flat pattern, two types of slope patterns, four types of step patterns, and two types of impulse patterns are obtained for each direction.
[0028]
₃ indicates the pixels G ₀ and G ₁ or the pixels G ₂ and G ₃ in FIG. 11, and the horizontal pattern indicates the signal level, and the horizontal axis indicates the signal level. As for the pattern in the vertical direction, the horizontal axis and the vertical axis are the signal level and position.
Next, on the strong very correlated with the signal of the pixel G _x In this example, the pixel G _x using signals 2 pixels G _0, G ₁ and the left, two pixels in the right G _2, G ₃ below Is interpolated.
[0029]
As the interpolation operation, the following operations can be considered.
(I) Vertical average This is an interpolation operation in which the average of the signal levels of the pixels G ₀ and G _{1 is} the signal level of the pixel G _x .
(Ii) Left / Right Average This is an interpolation operation in which the average of the signal levels of the pixels G ₂ and G _{3 is used as} the signal level of the pixel G _x .
[0030]
(Iii) Lower value hold This is an interpolation operation in which the signal level of the pixel G _{1 is set to} the signal level of the pixel G _x .
(Iv) Upper value hold This is an interpolation operation in which the signal level of the pixel G _{0 is set to} the signal level of the pixel G _x .
[0031]
(V) Left value hold which an interpolation operation of a signal level of the pixel G ₂ and the signal level of the pixel G _x.
(Vi) Hold right value This is an interpolation operation in which the signal level of the pixel G _{3 is set to} the signal level of the pixel Gx.
[0032]
Then, as a result of examining the operation of properly interpolating the missing signal of the center pixel G _x for a total of 81 types of two-dimensional change patterns in the horizontal direction (9 types) × vertical direction (9 types) in FIG. As shown in the figure, although the types and number of interpolation operations differ depending on the pattern, it has been found that each pattern can be grouped for each appropriate interpolation operation.
FIG. 3 shows an optimum interpolation operation for each pattern using an arrow symbol.
[0033]
In addition, according to FIG. 3, the conventional simple average interpolation (vertical simple average interpolation) alone only properly interpolates 1/3 of the entire pattern, and uniform simple average interpolation is less effective. I understand that.
Then, the optimum interpolation operation for each pattern in FIG. 3 is grouped by narrowing down to one, and for example, the optimum interpolation operation for each pattern shown in FIG. 4 is set in advance.
[0034]
In FIG. 4, “RECOMMENDATION” indicates an interpolation operation when the horizontal direction and the vertical direction are both impulse patterns. If both directions are the impulse patterns, as is clear from FIG. It can not recover the signal of the pixel G _x at any interpolation operation (vi), because they can not be optimal interpolation means setting an appropriate value in advance.
[0035]
Next, if the vertical and horizontal patterns are determined from simple signal level comparisons of the two pixels G ₀ and G ₁ , G ₂ and G ₃ , as is apparent from FIGS. 3 and 4, for example, Cannot discriminate between flat pattern and impulse pattern, slope pattern and step pattern.
[0036]
Therefore, in this embodiment, a range of a total of 25 pixels having a strong correlation around the correction target pixel in the horizontal direction (5 pixels) × vertical direction (5 pixels) corresponding to FIG. The level of the G signal of the interpolation target pixel is estimated from the signal of the pixel of the filter 2r or 2b having the same color as the interpolation target pixel and the signal of the pixel of the G filter 2g as described below.
[0037]
That is, as is clear from FIG. 2, there are nine pixels of the filter 2r or 2b having the same color as the interpolation target pixel and twelve pixels of the G filter 2g in the interpolation unit region.
Then, the interpolation unit region considering that signal correlation is strong, for example, the interpolation target pixel G _x when the pixel of the R filter 2r, can estimate the G signal of the pixel G _x from Equation 1 below.
[0038]
[Expression 1]
S _GX = (S _GaV / S _RaV ) ・ S _RX
[0039]
However, S _GaV = ΣS _Gi / 12, S _RaV = ΣS _Ri / 9, S _Gi indicates the signal level of the pixel of the G filter 2g, S _Ri indicates the signal level of the pixel of the R filter 2r, and i is a variable. is there.
Further, S _RX indicates the actual signal level based on the R filter 2r pixel G _x.
[0040]
Then, the horizontal and vertical patterns are discriminated from the relationship between the estimated signal level S _GX of the pixel G _{x and} the signal levels of the two pixels G ₀ and G ₁ or G ₂ and G ₃ obtained from the equation (1).
For example, in the horizontal direction, the signal levels of the pixels G _{2 and} G ₃ have a difference according to the pattern as shown in FIGS. 5A and 5B, and if the difference is Δ, the slope pattern Since the difference Δ becomes large in the case of a step pattern, a threshold value E is set, and when the difference Δ is equal to or greater than E as shown in FIG.
[0041]
Further, when it is determined that the pattern is a slope pattern or a step pattern, as shown in FIG. 6, the estimated signal level of the pixel G _x marked with ● obtained from the equation 1 is the average of the signal levels of the pixels G _{2 and} G _3. If the level is within the range of ± Δ / 4 from the value, the slope pattern is assumed, and if the estimated signal level of the pixel G _x is outside the range of ± Δ / 4 as shown in FIG. Suppose that it is a pattern.
[0042]
However, when extremely large the difference between the estimated signal level and delta pixel G _x as shown in FIG. 8 (3/4) delta or more, and an impulse pattern.
Further, when the difference Δ is smaller than the threshold value E as shown in FIG. 5B, it is determined that the pattern is a flat pattern or an impulse pattern.
[0043]
In this case, the second threshold value E ′ is set as shown in FIG. 9, and the average of the estimated signal level of the pixel G _{x and} the signal levels of the pixels G _{2 and} G ₃ is set as shown in FIG. If the difference from the value is equal to or smaller than the threshold value E ′, the pattern is a flat pattern. If the difference is larger than the threshold value E ′ as shown in FIG.
[0044]
6 to 9, the horizontal axis represents the position, and the vertical axis represents the signal level.
Then, threshold values similar to the threshold values E and E ′ are set in the vertical direction, and the pattern is discriminated in the same manner as in the horizontal direction.
[0045]
Furthermore, a two-dimensional change pattern is specified by discriminating changes in the horizontal and vertical G signal levels centered on the pixel G _x from the horizontal and vertical pattern discrimination results, and corresponding to the specified pattern. 4 interpolation operation is performed.
Then, in order to perform the above-described processes digitally and perform contour enhancement similar to FIG. 10, the output signal of the A / D converter 4g of FIG. 10 is supplied to the input terminal 14 as shown in FIG. Four vertical delay units 15 to 18 formed by a 1H line memory, and five horizontal delay units 20 to 24 formed by a cascade circuit of four delay elements 19a to 19d such as a pixel memory Thus, the output signal of the converter 4g is delayed two-dimensionally to cut out the interpolation unit area.
[0046]
At this time, the output signals of the delay elements 19a, 19b and 19c of the delay unit 22 become the signals of the pixels G ₃ , G _x and G ₂ of FIG. 11, and the output signals of the delay elements 19b of the delay units 21 and 23 are the same as FIG. Pixels G ₀ and G ₁ .
Then, a 5-bit signal shifted by one pixel of the delay units 20 to 24, that is, a signal of each pixel in the interpolation unit region is supplied to the average value circuits 25 and 26, and the average value circuit 25 receives the signal level S _GaV (= ΣS _Gi / 12) is calculated, and the average value circuit 26 calculates the signal level S _Rav (= ΣS _Ri / 9).
[0047]
Furthermore, the divider 27, the multiplier 28 calculates the Equation 1 using the output signal and the output signal of the delay element 19b of the delay device 22 of the averaging circuit 25 (signal of the pixel G _x), the pixel An estimated signal level S _Gx of G _x is calculated, and a signal of this level S _Gx is supplied to the horizontal and vertical pattern determination circuits 29 and 29 ′.
The determination circuit 29 calculates the output signals (the signals of the pixels G ₃ and G ₂ ) of the delay elements 19 a and 19 c of the delay unit 22 using the subtracter 30 and the average value circuit 31, and the subtracter 30 and the absolute value circuit 32 The absolute value of the difference Δ is obtained and supplied to the comparator 33 and the constant multipliers 34 and 35 of 1/4 and 3/4.
[0048]
The average value circuit 31, subtractor 36 and the absolute value circuit 37 calculates the absolute value of the difference ε between the average of the signal level of the estimated signal level and the pixel G _2, G ₃ pixels G _x, the result of the calculation It supplies to the comparators 38 and 39.
Further, the comparator 33 compares the signal of the absolute value of the difference Δ of the absolute value circuit 32 with the signal of the threshold value E of the terminal 41, and “1” when the absolute value of the difference Δ is equal to or greater than the threshold value E. A determination signal is formed and supplied to the logic gates 42 to 46.
[0049]
Further, the comparators 38 and 39 compare the absolute value signals of Δ / 4, (3/4) · Δ of the multipliers 34 and 35 with the absolute value signal of the difference ε of the absolute value circuit 37, and the difference ε When the absolute value of is equal to or smaller than the absolute value of Δ / 4, when the absolute value of the difference ε is equal to or greater than (3/4) · Δ, a determination signal that becomes “1” is formed. The signal is supplied to the logic gates 42 and 43, and the discrimination signal of the comparator 39 is supplied to the logic gate 44.
[0050]
Further, the comparator 40 compares the signal of the absolute value of the difference ε of the absolute value circuit 37 with the signal of the threshold value E ′ of the terminal 47. When the absolute value of the difference ε is equal to or greater than the threshold value E ′, A discrimination signal which becomes 1 ″ is formed, and this signal is supplied to the logic gates 45 and 46.
The logic gate 42 outputs “1” when the slope pattern shown in FIG. 6 is determined, and the logic gate 43 outputs “1” when the step pattern shown in FIG. 7 is determined.
[0051]
Further, the logic gate 48 for calculating the logical sum of the output signals of the logic gates 44 and 45 has an output signal of “1” when the impulse pattern shown in FIG. 8 and FIG. When the flat pattern of (a) is determined, the output signal becomes “1”.
Then, the output signals of the logic gates 42, 43, 46, and 48 are supplied to the interpolation circuit 49 as the horizontal pattern discrimination result.
[0052]
The determination circuit 29 ′ is configured in the same manner as the determination circuit 29, and is based on the output signal of the delay element 19c of the delay units 21 and 23 (the signals of the pixels G ₀ and G ₁ ) and the signal S _{Gx of the} multiplier 28, A vertical pattern discrimination result signal is supplied to the interpolation circuit 49.
[0053]
Then, the interpolation circuit 49 discriminates the change in the G signal level in the horizontal direction and the vertical direction around the pixel G _x from the discrimination results of the discrimination circuits 29 and 29 ′, and specifies the corresponding two-dimensional change pattern. The interpolation operation of FIG. 3 giving priority to the horizontal resolution corresponding to the pattern is alternatively selected, and the signals of the pixels G _{0 to} G ₃ and the estimated signal S _Gx are used to calculate the average or the upper, lower or left, The G signal lacking the pixel G _x is digitally interpolated and generated by an optimal interpolation operation such as hold down, left, or right.
[0054]
Furthermore, switching the switch 51 by the switching signal terminal 50, the pixel G _x is R, B filter 2r, a delay element 19b of the delay device 22 when the G filter 2g selects the output signal of the interpolation circuit 49 when 2b Output signal is supplied, and the G signal of each pixel of the imaging output is supplied to the nth-order differentiation circuit 52. The differentiation circuit 52 uses the same two-dimensional contour enhancement aperture as the output signal of the addition circuit 12 of FIG. Form a signal.
[0055]
At this time, the lack of the G signal is interpolated with higher accuracy than before by the interpolation operation selected in consideration of the two-dimensional level change of this signal, and a large deviation from the appropriate interpolation level as in the past is prevented. Therefore, false edges, moire, etc. of the reproduced image emphasized by the aperture signal of the differentiation circuit 52 are remarkably suppressed, the deterioration of the frequency characteristics due to the disappearance of the edge is greatly improved, and the reproduced image quality is dramatically improved.
[0056]
The size of the interpolation unit area and the types and number of horizontal and vertical patterns are not limited to those in the embodiment.
[0057]
Of course, the present invention can be applied not only to the imaging output of a single-plate or single-tube image sensor but also to the imaging output of a two-plate or two-tube image sensor.
The color filter of the image sensor only needs to have a G filter arranged in a checkered pattern in units of pixels. For example, it may be a two-color filter of a G filter and an R filter.
[0058]
Further, a method for determining a change in the G signal level in the horizontal direction and the vertical direction is not limited to the embodiment.
Of course, the present invention can be applied not only to edge enhancement but also to G signal interpolation for various video processing such as improvement of sharpness or Y signal based on this signal.
[0059]
【The invention's effect】
Since the present invention is configured as described above, the following effects can be obtained.
For each actual interpolation target pixel of the imaging output of the two-dimensional image sensor, a pixel in the interpolation unit region centered on this pixel is cut out, and the interpolation target pixel is centered from the actually obtained signal in that region. A two-dimensional change pattern of the green signal level in the horizontal and vertical directions is specified, and an optimal interpolation operation set in advance for each pattern is selected based on the specified pattern in consideration of the correlation of the two-dimensional signal level. Selected, and the missing green signal is interpolated by the selected interpolation operation.
[0060]
At this time, since the optimum interpolation operation is selected in consideration of the two-dimensional level change of the green signal, interpolation with extremely high accuracy can be performed, and the interpolation accuracy is significantly improved compared to the conventional method.
[Brief description of the drawings]
FIG. 1 is a circuit block diagram of an embodiment of a signal interpolation method of the present invention.
FIG. 2 is an explanatory diagram of a color filter according to an embodiment of the present invention.
FIG. 3 is an explanatory diagram of a combination of a two-dimensional signal level change and an interpolation operation.
4 is an explanatory diagram of the optimum interpolation operation of FIG. 1. FIG.
FIGS. 5A and 5B are first explanatory diagrams of pattern determination for level change in the horizontal direction in FIG. 1;
6 is a second explanatory diagram of a pattern determination for level change in the horizontal direction in FIG. 1; FIG.
FIG. 7 is a third explanatory diagram of the level determination pattern determination in the horizontal direction in FIG. 1;
FIG. 8 is a fourth explanatory diagram of pattern determination for level change in the horizontal direction in FIG. 1;
FIG. 9 is a fifth explanatory diagram of a pattern determination for level change in the horizontal direction in FIG. 1;
FIG. 10 is a circuit block diagram for explaining a conventional example.
11 is an explanatory diagram of a pixel output of the image sensor of FIG.
FIGS. 12A to 12L are explanatory diagrams of processing of a conventional example.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Two-dimensional image sensor 2 Color filter 15-18, 20-24 Delay device 25, 26 Average value circuit 27 Divider 28 Multiplier 29, 29 'Pattern determination circuit 49 Interpolation circuit

Claims (2)

Signal interpolation for interpolating green signals with missing pixels that do not correspond to the green filter of the imaging output of a two-dimensional image sensor in which green filters are arranged in a checkerboard pattern in units of pixels to form a color filter. In the method
A predetermined two-dimensional interpolation unit area centered on the interpolation target pixel from which the green signal is missing is set in advance;
The green signal level of the interpolation target pixel is estimated based on the level of the pixel signal corresponding to the green filter in the interpolation unit region and the level of the pixel signal corresponding to the filter of the same color as the interpolation target pixel in the interpolation unit region. And
The estimated green signal level of the interpolation target pixel and the horizontal and vertical green signal level changes centered on the interpolation target pixel are classified into a plurality of two-dimensional change patterns, and the interpolation target of each pattern Preset the optimal interpolation operation of the pixel,
For each pixel to be interpolated in the imaging output, the pixel in the interpolation unit area centered on the pixel is cut out,
The change in the green signal level in the horizontal direction and the vertical direction centering on the pixel to be interpolated is determined from the signal level of each pixel of the green filter in the extracted interpolation unit area, and the corresponding two-dimensional change pattern is specified. And
A signal interpolation method, comprising: generating an interpolated green signal of the interpolation target pixel by the optimum interpolation operation corresponding to the specified two-dimensional change pattern.   The green signal level of the interpolation target pixel is estimated by interpolating the ratio of the average level of the pixel signal corresponding to the green filter in the interpolation unit area and the average level of the pixel signal corresponding to the filter of the same color as the interpolation target pixel. 2. The signal interpolation method according to claim 1, wherein a value obtained by multiplying the signal level of the target pixel is set.

Images