JP2000152264A - Image processing apparatus and image processing method - Google Patents

Abstract

(57) Abstract: The present invention relates to an image processing apparatus and an image processing method, for example, applied to an image processing apparatus such as a television receiver, a video tape recorder, a television camera, and a printer. An output signal of a solid-state imaging device or the like is to be processed, and gradation can be corrected by effectively avoiding a partial decrease in contrast. SOLUTION: A feature amount xmax (i, j) indicating a feature in a predetermined range in the vicinity of each pixel is detected, and the feature amount xmax (i, j) is detected.
A region to which input image data belongs is determined based on (i, j), and a correction coefficient g is determined based on the determination result r (i, j).
(I, j) is generated to correct the pixel value x (i, j).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to an image processing apparatus and an image processing method, and can be applied to, for example, an image processing apparatus such as a television receiver, a video tape recorder, a television camera, and a printer. The present invention detects a feature amount indicating a feature in a predetermined range near each pixel, and generates a correction coefficient based on a determination result of an area to which image data belongs based on the feature amount to correct a pixel value.
For example, for an output signal of a single-plate solid-state imaging device or the like, a gradation can be corrected by effectively avoiding a partial deterioration in contrast.

[0002]

2. Description of the Related Art Conventionally, in an image processing apparatus such as a television camera, the gradation of image data obtained through image input means such as an image pickup means is corrected and output.

FIG. 18 is a characteristic curve diagram showing input / output characteristics of a signal processing circuit applied to the gradation correction processing.
This type of signal processing circuit reduces the gain when the input level 1 rises above a predetermined reference level lk. As a result, this type of signal processing circuit suppresses the signal level when the input level is higher than the reference level lk and outputs the signal. In this case,
The gradation is corrected at the expense of the contrast of a portion having a high signal level.

In the characteristic curve diagram shown in FIG. 18, the horizontal axis represents a pixel value 1 which is an input level of image data,
The vertical axis represents the pixel value T (l), which is the output level of the image data, and Lmax represents the maximum level that each pixel of the input / output image can take. In the following, a function indicating an input / output relationship as shown in the characteristic curve diagram is referred to as a level conversion function.

FIG. 19 is a characteristic curve diagram showing the input / output characteristics of the same type of signal processing circuit. The signal processing circuit using the level conversion function reduces the gain when the input level 1 is equal to or lower than the first reference level ls and when the input level l is equal to or higher than the second reference level lb. As a result, this signal processing circuit corrects the gradation at the expense of the contrast between the low and high signal levels.

On the other hand, in image processing using a computer, gradation is corrected by, for example, histogram equalization.

This histogram equalization is:
This is a method of adaptively changing the level conversion function according to the frequency distribution of the pixel values of the input image, and is a method of correcting the gray levels by reducing the gray levels of a portion having a low frequency distribution of the pixel values.

That is, as shown in FIG. 20, in the histogram equalization process, a frequency distribution H which is a total of the number of pixels based on the pixel value 1 of the input image is used.
Based on (l), a cumulative frequency distribution C (l) is detected by the following equation.

[0009]

(Equation 1)

In the process of histogram equalization, the cumulative frequency distribution C
By normalizing (l) by the following equation, a level conversion function T (l) is defined.
According to (l), the signal level of the input image is corrected. Here, Fmax is the cumulative frequency distribution C
(L) is the final value, and Lmax is the maximum value of the input / output level.

[0011]

(Equation 2)

[0012] Such a process of correcting the gradation may be performed even when the image data is transmitted through a transmission line, displayed on a display device, or stored in a storage device.
For example, for the purpose of suppressing the dynamic range and the like, the program is appropriately executed as needed.

[0013]

In the tone correction processing according to the above-described conventional method, the entire tone is corrected at the expense of the contrast of any part. This is because, in any of the methods, in order to avoid generation of an unnatural image, level conversion is performed using an input / output function having a monotonically increasing property.

Therefore, in the case of the conventional method, there is a problem that the contrast is partially reduced in the processed image.

The present invention has been made in consideration of the above points, and corrects gradation by effectively avoiding a partial decrease in contrast for an output signal of, for example, a single-chip solid-state image sensor. It is intended to propose an image processing apparatus and an image processing method that can perform the processing.

[0016]

According to the present invention, there is provided an image processing apparatus or an image processing method for sequentially detecting a feature amount indicating a feature in a predetermined range in the vicinity of each pixel. Based on the determination result, a region to which the image data belongs is determined, and a correction coefficient is generated based on the determination result to correct the pixel value of the image data.

If the area to which the image data belongs is determined and a correction coefficient is generated based on the determination result to correct the pixel value of the image data, the pixel value is corrected by the same coefficient in the same area and Holds the magnitude relation of the pixel values, and the magnitude relation of the pixel values can be reversed between the pixels belonging to different regions, thereby avoiding partial contrast deterioration and correcting the entire gradation. It becomes possible.
At this time, a feature amount indicating a feature in a predetermined range near each pixel is sequentially detected, and if an area to which image data belongs is determined based on the feature amount, for example, the output signal of a single-chip solid-state imaging device is directly processed and processed. In this case, the gradation can be corrected without losing the color information superimposed on the luminance signal.

[0018]

Embodiments of the present invention will be described below in detail with reference to the drawings.

(1) First Embodiment (1-1) Configuration of First Embodiment FIG. 1 is a block diagram showing a television camera according to a first embodiment of the present invention. In the television camera 1, a CCD solid-state imaging device (CCD) 2 outputs an imaging result by driving a timing generator (TG) 3.

As shown in an enlarged front view of the image pickup surface in FIG. 2, this CCD solid-state image pickup device 2 is a so-called single-plate type solid-state image pickup device. Be placed. That is, the CCD solid-state imaging device 2
Is that yellow (Ye) and cyan (Cy) color filters are repeated in pixel units to form odd lines, whereas magenta (Mg) and green (G) color filters are repeated in pixel units. Even lines are formed.

As a result, in the CCD solid-state imaging device 2, as shown in FIG. 3, the amplitude-modulated color signals are sequentially converted into luminance signals by time division from the correlated double sampling circuit associated with this type of imaging device. The imaging result is output so as to be superimposed.

To output such an imaging result, C
The CD solid-state imaging device 2 obtains an imaging result in a cycle of 1/60 [second] according to a charge accumulation time set by a user, and outputs the imaging result as an imaging result VN by normal exposure.
Further, in the vertical blanking period of the imaging result VN obtained by the normal exposure, the CCD solid-state imaging device 2 obtains an imaging result with a charge accumulation time shorter than the charge accumulation time by the normal exposure. Is output as the imaging result VS.

As a result, as shown in FIG. 4, in the CCD solid-state imaging device 2, the imaging result VN (FIG. 4 (A)) obtained by normal exposure in which the output level is saturated when the amount of incident light exceeds a predetermined value. The imaging result V of the short-time exposure in which the output level is not saturated due to the shorter charge accumulation time
S (FIG. 4B) is output as a set.

The memory 4N inputs the imaging result VN by the normal exposure via a correlated double sampling circuit, a defect correction circuit, an analog-to-digital conversion circuit, etc., and temporarily holds the imaging result VN by the normal exposure. Output.

Similarly, the memory 4S receives the imaging result VS of this short-time exposure via a correlated double sampling circuit, a defect correction circuit, an analog-to-digital conversion circuit, etc. VS is temporarily held and output.

The addition circuit 5 adds the imaging result VN obtained by the normal exposure held in the memory 4N and the imaging result VS held by the short-time exposure held in the memory 4S to provide a wide dynamic range and a sufficient dynamic range. The imaging result VT based on the pixel value is output, and the level correction circuit 6 outputs the imaging result VS by the short-time exposure output from the memory 4S so that practically sufficient linearity can be secured in the imaging result VT obtained by the adding circuit 5. Is corrected and output.

As a result, the television camera 1 is configured to generate an imaging result VT (FIG. 4C) with a much larger dynamic range than before.

The gradation correcting circuit 8 corrects the pixel value of the imaging result VT to correct and output the gradation of the imaging result VT. In the television camera 1,
A subsequent signal processing circuit (not shown) executes various signal processing required for the television camera and outputs the imaging result to an external device or the like. At this time, the pixel value of the imaging result is changed to correspond to the external device. By uniformly suppressing, the dynamic range of the imaging result is suppressed and output.

The characteristic amount filter 9 detects and outputs a characteristic amount for each pixel value x (i, j) from the imaging result VT obtained in this way. Here, the feature amount indicates a feature of a predetermined range centered on a pixel based on the pixel value x (i, j) for each pixel value x (i, j) of the sequentially input imaging result VT. . In this embodiment, a two-dimensional maximum value filter is applied to the feature amount filter 9, and each pixel value x (i,
j), the maximum value of the pixel value is detected in a predetermined range near the pixel based on the pixel value x (i, j), and the maximum value xma
x (i, j) is output as a feature value at the pixel value x (i, j).

That is, the feature value filter 9 processes the pixel values x (i, j) sequentially inputted by the following arithmetic processing, thereby sequentially outputting the feature values xmax (i, j). As shown in FIG. 5, in this embodiment, in the imaging results VT input in the raster scanning order, the horizontal direction is indicated by a suffix with a code i, and the vertical direction is indicated by a suffix with a code j. In the following equation, max is a function for calculating the maximum value of x that satisfies a predetermined condition. Here, the condition is defined as a range of M × N pixels centered on a pixel having a pixel value x (i, j). It is.

[0031]

(Equation 3)

The area judgment filter 10 judges the area to which the input image data belongs based on the feature amount xmax (i, j) detected in this way, and outputs the judgment result. At this time, the area determination filter 10 determines the feature amount xmax
The average value of (i, j) is detected, thereby determining which average luminance level the input image data belongs to, and outputting this average value as an identification signal.

That is, the area judging filter 10 is a two-dimensional low-pass filter, and a low-frequency component r expressed by the following equation with respect to the feature quantity xmax (i, j) sequentially input in the raster scanning order. (I, j) is detected, and this low frequency component r (i, j) is output as an identification signal for each area.

[0034]

(Equation 4)

Note that N and M in the equation (4) are constants representing the size of the neighboring area for calculating the average value, and have no relation to N and M shown in the equation (3). As a result, the region determination filter 10 removes a structure smaller than the imaging result VT based on the feature amount xmax (i, j) and extracts a region having relatively flat pixel values. Since the area determination filter 10 aims at such processing, it is desirable that the band is relatively narrow.

The coefficient calculation circuit 11 calculates the low frequency component r
According to the signal level of (i, j), for example, a contrast correction coefficient g is calculated by a coefficient calculation function G as shown in FIG.
(I, j) is generated. Here, the coefficient calculation function G is
For example, the level conversion function T (l) described above with reference to FIG.
Is a function obtained by performing an arithmetic operation on the following equation.

[0037]

(Equation 5)

Thus, the coefficient calculation circuit 11 generates and outputs a contrast correction coefficient g (i, j) by the following arithmetic processing, and the signal level of the low frequency component r (i, j) is changed to a predetermined reference level. For a region equal to or smaller than lk, the contrast correction coefficient g (i,
j), and in the region above the reference level lk, the contrast correction coefficient g (i, j) is set so that the value gradually approaches the value gmin according to the signal level of the low frequency component r (i, j). Is output.

[0039]

(Equation 6)

The multiplication circuit 12 calculates the contrast correction coefficient g (i, j) generated in this way and the pixel value x
By multiplying by (i, j), the signal level of the imaging result VT is corrected and output by the contrast correction coefficient g (i, j).

(1-2) Operation of First Embodiment In the above configuration, in the television camera 1 (FIG. 1), amplitude modulation is performed by a color filter (FIG. 2) disposed on the imaging surface (FIG. 2). An imaging result in which the color signals are sequentially superimposed on the luminance signal by time division is output from the CDD solid-state imaging device 2 (FIG. 3).

Further, in the television camera 1, an imaging result VN (FIG. 4A) by the normal exposure based on the charge accumulation time set by the user and an imaging result VS (FIG. 4 (A)) obtained by the short exposure using the short charge accumulation time. B)) are output alternately, and the imaging results VN and VS are stored in the memory 4 respectively.
N and 4S. In television camera 1,
The two imaging results VN and VS are output to the level correction circuit 6,
The image pickup result VT is synthesized by the adder circuit 5 and has a significantly larger dynamic range as compared with the related art.
(FIG. 4C) is generated.

In this imaging result VT, each pixel value x (i,
The feature value in j) is detected. That is, the imaging result V
T is a maximum value x of a pixel value in a predetermined range near each pixel.
max (i, j) is detected as a feature amount in the corresponding pixel x (i, j).

The imaging result VT is determined by the following area determination filter 1
0, the feature amount xmax detected in this way
The region to which the input image data belongs is determined based on (i, j), and the determination result is output. More specifically,
The characteristic amount xmax (i, j) is obtained by the area determination filter 10.
The low frequency component r (i, j), which is the average value of
As a result, a fine structure in the image is removed, and a region having a relatively flat pixel value is extracted. Also, this low frequency component r
(I, j) is output as an identification signal for each area.

At this time, in the imaging result VT obtained by the color filter of the complementary color one pine format, the average of adjacent pixel values represents luminance information (FIG. 3), and the low frequency component r (i, j)
In, since the maximum value is detected by the characteristic amount filter 9 and the characteristic amount xmax (i, j) is detected, the characteristic value changes according to the amplitude of the color signal superimposed on the luminance signal. .

In the imaging result VT, a coefficient correction circuit 11 generates a contrast correction coefficient g (i, j) in accordance with the signal level of the low frequency component r (i, j). The pixel value x (i, i) of the imaging result VT is obtained in the multiplication circuit 12 by (i, j).
j) is corrected.

As a result, in the imaging result VT, in a region where the signal level of the low frequency component r (i, j) is equal, the pixel value is corrected by the same gain,
In a region where the signal level of the low frequency component r (i, j) is different, the pixel values can be made close to each other in accordance with the setting of the level conversion function T (l). It can also be reversed. As a result, the contrast in each region can be naturally increased with respect to the entire gradation, and the whole gradation can be corrected while effectively preventing a partial decrease in contrast.

That is, as shown in FIG. 7, the characteristic amount xmax (i, j), which is the luminance level of the image pickup result, pulsates at a frequency equal to or higher than the cutoff frequency of the area judgment filter 10 which is a low-pass filter, and furthermore, the characteristic amount xmax
The case where the DC level of (i, j) rises rapidly (FIG. 7B), and the change of the low frequency component r (i, j) accompanying the sudden change of the DC level is a coefficient calculation function. In the case where the inflection point of G (l) is straddled (FIG. 7A), the contrast is suppressed in a portion having a large luminance level by the conventional level conversion function described above with reference to FIG. 15 (FIG. 7). (C)).

However, according to this embodiment, before and after the signal level of the low frequency component r (i, j) sharply rises, the gain according to the signal level of the low frequency component r (i, j) is obtained. Pixel value x of imaging result VT
By correcting (i, j), in a portion having a small luminance level, the imaging result VT is obtained by the gain gmax based on the average value level 12 of the peak value 13 and the bottom value 11.
Pixel value x (i, j) is corrected, whereby the same level of contrast as in the conventional method can be obtained in the low-level region (FIG. 7D).

On the other hand, on the high level side, the average level 15 of the peak value 16 and the bottom value 14 is similarly calculated.
The pixel value x (i, i,
j) is corrected. At this time, the peak value 16 and the bottom value 14 are corrected by a uniform gain, so that the contrast between the peak value 16 and the bottom value 14 is amplified by the gain g5. become.

Thus, in the gradation correction circuit 8 according to this embodiment, although the gradation when viewed as a whole does not change significantly, the microscopic pulsation does not correspond to the input image. As a result, the pulsation due to the VT can be increased.

Similarly, as shown in FIG. 8, there is a case where the DC level rapidly rises due to the pulsation of the feature quantity xmax (i, j) which is the luminance level of the imaging result.
When the large change in the feature amount xmax (i, j) is biased toward a higher level than the inflection point of the coefficient calculation function G (l) (FIG. 8B), the conventional level conversion function described above with reference to FIG. In some cases, the contrast is suppressed at all pixel values x (i, j) (FIG. 8C).

However, also in this case, on the low level side and the high level side, the average value levels l2 and l5 are respectively provided.
Are corrected by the gains g2 and g5 corresponding to
Although the gradation when viewed as a whole does not change significantly, the microscopic pulsation is not affected by the imaging result V which is the input image.
The pulsation due to T can be expanded (FIG. 8).
(D)).

To correct the gradation of the imaging result VT in this manner, in this embodiment, the low frequency component r (i, j) of the feature amount xmax (i, j) detected by the feature amount filter 9 is used. By setting the gain by
Appropriate color reproduction is possible by correcting the gradation without losing color information.

That is, in such an imaging result VT, the average of the adjacent pixel values represents the luminance information (FIG. 3), and the difference between the adjacent pixel values represents the color information, so that the pixel value x (i, When the gradation is corrected by the average value level of j) (that is, when the feature amount filter 9 is omitted), the gradation is fixed by a constant gain regardless of the magnitude of the pixel value x (i, j) based on the color information. It will be corrected. Therefore, in this case, the pixel value y (i, j) whose gradation has been corrected may be saturated at a portion where the pixel value x (i, j) of the imaging result VT rises largely due to the high degree of color saturation. When the pixel value x (i, j) is saturated as described above, it is difficult to properly reproduce colors.

However, in this embodiment, since the maximum value of the pixel value x (i, j) is detected as the feature value xmax (i, j), the portion where there is a possibility of being saturated, Average luminance level r as the identification result
By increasing (i, j), the gain at the time of gradation correction can be reduced, and the color can be correctly reproduced by effectively avoiding the saturation of the pixel value x (i, j).

(1-3) Effect of First Embodiment According to the above configuration, the area to which the input image data belongs is determined, a correction coefficient is generated based on the determination result, and the correction coefficient is calculated in accordance with the correction coefficient. By correcting the imaging result, it is possible to make the pixel values close to each other as necessary between the pixels belonging to different areas while maintaining the magnitude relation of the pixel values by the same coefficient in the same area, and in an extreme case, Can be reversed. As a result, the gradation can be corrected by effectively avoiding partial deterioration of the contrast.

At this time, the maximum value xmax (i, i,
By detecting j) as a feature amount and obtaining a determination result based on this feature amount, it is possible to effectively avoid the saturation of the pixel value y (i, j) and correct the gradation. Can reproduce colors correctly.

Further, by dividing the imaging result by the low-pass filter into regions, it is possible to correct the entire gradation with a simple configuration while avoiding a partial decrease in contrast.

(2) Second Embodiment FIG. 9 is a block diagram showing a gradation correction circuit applied to a television camera according to a second embodiment of the present invention. This gradation correction circuit 18 is applied instead of the gradation correction circuit 8 described above with reference to FIG. In the tone correction circuit 18, the same components as those of the above-described tone correction circuit 8 are denoted by the corresponding reference numerals, and duplicate description will be omitted.

In the gradation correction circuit 18, the characteristic amount filter 19 detects and outputs a characteristic amount for each pixel value x (i, j) of the imaging result VT. Feature filter 19
Is a two-dimensional minimum value filter, and in detecting a feature amount, each pixel value x (i, i) of the imaging result VT sequentially input.
j), the minimum value xmin (i, j) of the pixel value in a predetermined range centered on the pixel by the pixel value x (i, j)
And outputs the minimum value xmin (i, j) as a feature value at the pixel value x (i, j).

That is, the feature value filter 19 processes the pixel values x (i, j) sequentially inputted by the calculation processing of the following equation, thereby sequentially outputting the feature values xmin (i, j). Here, min is a function for calculating the minimum value of x satisfying a predetermined condition.

[0063]

(Equation 7)

The coefficient calculation circuit 21 calculates the low frequency component r
According to the signal level of (i, j), a contrast correction coefficient g (i, i) is obtained by a coefficient calculation function G obtained by performing arithmetic processing on the level conversion function T (l) shown in FIG.
j). Here, this level conversion function T (l)
Is a characteristic set so as to suppress the signal level in a portion where the luminance level is lower than the predetermined reference level lk.

As described above, the correction coefficient g
When setting (i, j), simply set the pixel value x
When the gain is controlled by the low frequency component of (i, j),
Contrary to the first embodiment, low frequency components r (i, j)
In a portion where the signal level is small, the pixel value y (i, j) whose gradation has been corrected may be saturated and the color information may be lost. In such a portion, it is difficult to correctly reproduce the color.

However, in this embodiment, since the minimum value of the pixel value x (i, j) is detected as the feature value xmin (i, j), in the portion where there is a possibility of saturation, Average luminance level r as the identification result
By decreasing (i, j), the gain at the time of gradation correction can be increased, and saturation of the pixel value y (i, j) can be effectively avoided and color reproduction can be performed correctly.

According to the configuration shown in FIG. 9, by detecting the minimum value of a predetermined area as a feature amount and correcting the gradation,
Even in the case where the gradation is corrected so as to suppress the signal level in a portion where the luminance level is lower than the predetermined reference level lk, the same effect as in the first embodiment can be obtained.

(3) Third Embodiment FIG. 11 is a block diagram showing a gradation correction circuit applied to a television camera according to a third embodiment of the present invention. This gradation correction circuit 28 is applied instead of the gradation correction circuit 8 described above with reference to FIG. In the gradation correction circuit 28, the same components as those of the above-described gradation correction circuits 8 and 18 are denoted by the corresponding reference numerals, and the duplicate description will be omitted.

In the gradation correction circuit 28, the characteristic amount filter 29 detects and outputs a characteristic amount for each pixel value x (i, j) of the imaging result VT. That is, in the feature amount filter 29, the maximum value filter 29A has the same configuration as the feature amount filter 9 described in the first embodiment, and each pixel value x of the imaging result VT sequentially input.
For (i, j), the maximum value xmax (i, j) of the pixel value is detected and output in a predetermined range near the pixel based on the pixel value x (i, j).

The minimum value filter 29B has the same configuration as the feature amount filter 19 described in the second embodiment, and each pixel value x of the sequentially input imaging result VT.
For (i, j), a minimum value xmin (i, j) of the pixel value is detected and output in a predetermined range near the pixel based on the pixel value x (i, j).

The low-pass filter 29C is a two-dimensional low-pass filter. The average value xave of each pixel value x (i, j) of the sequentially input imaging result VT is obtained.
(I, j) is detected and output. In the low-pass filter 29, the constants M and N that define the size of the neighboring area expressed by the equation (4) are set to values smaller than that of the area determination filter 10. Is set so that the pass bandwidth is widened.

The selector 29D determines the average value xave based on a comparison result between the average value xave (i, j) output from the low-pass filter 29C and a predetermined reference value.
If (i, j) is larger than this reference level, the maximum value xmax (i, j,
j) is selected and output, while the average value xave (i,
When j) is smaller than the reference level, the minimum value xmin (i, j) output from the minimum value filter 29B is selectively output. As a result, the selector 29D sets the maximum value xm
ax (i, j) and the minimum value xmin (i, j) are combined to output a feature quantity xmm (i, j).

The coefficient calculation circuit 31 calculates the low frequency component r
According to the signal level of (i, j), the contrast correction coefficient g (i, j) is obtained by the coefficient calculation function G obtained by performing the arithmetic processing on the level conversion function T (l) shown in FIG.
Generate

In this case, when the contrast correction coefficient g (i, j) is set as described above and the gain is simply controlled by the low frequency component of the pixel value x (i, j), the first In addition, color information may be lost in a portion where the signal level of the low frequency component r (i, j) is large or small, as described above in the second embodiment, and it is difficult to reproduce color correctly in such a portion. become.

However, in this embodiment, the maximum value xmax (i, j) of the pixel value x (i, j) and the minimum value x
min (i, j) is the average value xa of the pixel values x (i, j)
ve (i, j) and the combined feature xmm
Generating a low frequency component r (i, j) from (i, j),
Since the gain is controlled by the low-frequency component r (i, j), the gain at the time of gradation correction can be increased or decreased in a portion where there is a possibility of saturation, and the pixel is accordingly reduced. Color can be correctly reproduced by effectively avoiding saturation of the value y (i, j).

According to the configuration shown in FIG. 11, the pixel value x
The maximum value xmax (i, j) and the minimum value xmi of (i, j)
n (i, j) and the average value xave of pixel values x (i, j)
By switching at (i, j) to generate the feature amount xmm (i, j), the signal level is suppressed at a portion having a luminance level lower than a predetermined reference level ls and a portion having a luminance level higher than the reference level lb. Even when the gradation is corrected as described above, the same effect as in the first embodiment can be obtained.

(4) Fourth Embodiment In this embodiment, a weighting addition circuit is applied instead of selector 29D in the configuration shown in FIG.

That is, the weighting and adding circuit executes the following arithmetic processing from the average value xave (i, j) of the pixel values x (i, j) output from the low-pass filter 29C, and thereby calculates the weighting coefficient a. Generate. Here, THL and THH are constants for normalization.

[0079]

(Equation 8)

Further, the weighting and adding circuit executes the following arithmetic processing based on the weighting coefficient a generated in this way, thereby obtaining the maximum value xmax (i, j) and the minimum value xmax.
min (i, j) and the characteristic amount xmm (i, j)
Generate Accordingly, the weighted addition circuit obtains the pixel value x
A feature quantity xmm (i, j) is generated by a weighted average of a maximum value xmax (i, j) and a minimum value xmin (i, j) based on an average value xave (i, j) of (i, j). I do.

[0081]

(Equation 9)

According to the fourth embodiment, the pixel value x
Based on the average value xave (i, j) of (i, j),
Maximum value xmax (i, j) and minimum value xmin (i, j)
Can be smoothly combined to generate the feature amount xmm (i, j), whereby the same effect as in the third embodiment can be obtained by natural gradation correction.

(5) Fifth Embodiment FIG. 12 is a block diagram showing a gradation correction circuit applied to a television camera according to a fifth embodiment of the present invention. This gradation correction circuit 38 is applied instead of the gradation correction circuit 8 described above with reference to FIG. In the gradation correction circuit 38, the same components as those of the above-described gradation correction circuit 8 are denoted by the corresponding reference numerals, and duplicate description will be omitted.

Here, the quantization circuit 43 calculates the characteristic amount xmax
(I, j) is requantized and the number of bits is reduced and output. Note that, in this embodiment, the quantization circuit 43
The pixel value x (i, j) is subjected to a calculation process of the following equation by a preset quantization step Q, whereby the feature value xmax (i, j) is linearly quantized to perform the feature value x
Output maxq (i, j). Where int
(A) is a function for truncating the decimal part of a.

[0085]

(Equation 10)

The area judgment filter 40 is formed in the same manner as the area judgment filter 10 according to the first embodiment except that the number of bits is different.

The look-up table (LUT) 44 is
A coefficient calculation circuit is configured to output a correction coefficient g (i, j) using the low frequency component r (i, j) output from the area determination filter 40 as an address. Thus, the lookup table 44 stores a correction coefficient LUT expressed by the following equation.
(I) is stored at the i-th address.

[0088]

[Equation 11]

According to the configuration shown in FIG. 12, by quantizing the feature in advance and processing it, the same effect as in the first embodiment can be obtained with a much simpler configuration. In addition, by generating the correction coefficient by using the look-up table, the entire process can be simplified by that much, and the configuration of the region determination filter can be simplified by performing quantization in advance at this time. Can reduce the size of the lookup table.

(6) Sixth Embodiment FIG. 13 is a block diagram showing a gradation correction circuit applied to a television camera according to a sixth embodiment of the present invention. This gradation correction circuit 48 is applied in place of the gradation correction circuit 38 described above with reference to FIG. 12, and a lookup table 54 and an interpolation circuit 55 are arranged instead of the lookup table 44 of the gradation correction circuit 38. .

Here, the lookup table 54 has an address smaller than the number of levels that the output value r (i, j) of the area determination filter 40 can take, and the output value r (i, j)
, Two addresses addr0 (i, j), ad expressed by the following equation
dr1 (i, j) and correction coefficients g0 (i, j), g1
(I, j). Here, the lookup table 54 outputs the two addresses addr0 (i, j) by omitting the lower bits of the output value r (i, j) of the area determination filter 40, and outputs the address addr1 ( i, j), this address ad
By adding a logical 1 bit to the least significant bit of dr0 (i, j), these addresses addr0 (i, j),
addr1 (i, j) is generated and output. Here, Rmax is the output value x (i,
The maximum value that can be taken by j), R'max, is the maximum value that the address of the lookup table 54 can take.

[0092]

(Equation 12)

[0093] The interpolation circuit 55
Address addr0 (i, j), ad input from
dr1 (i, j), correction coefficient g0 (i, j), g1
Using (i, j), an interpolation operation is performed by the following equation.
The result of the interpolation is output as a correction coefficient g (i, j).

[0094]

(Equation 13)

According to the configuration shown in FIG. 13, by generating a correction coefficient by performing an interpolation operation, it is possible to generate a correction coefficient whose value changes smoothly using a small-scale lookup table. The gradation can be corrected with minute accuracy.

(7) Seventh Embodiment FIG. 14 is a block diagram showing a gradation correction circuit applied to a television camera according to a sixth embodiment of the present invention. This gradation correction circuit 58 is applied instead of the gradation correction circuit 8 described above with reference to FIG. In the gradation correction circuit 58, the same components as those of the above-described gradation correction circuit 8 are denoted by the corresponding reference numerals, and the duplicate description will be omitted.

In the gradation correction circuit 58, an area determination filter 60 determines identification signals r0 (i, j) and r1 (i, i) obtained by determining areas to which input image data belongs at different resolutions.
j), r2 (i, j),..., rN-1 (i, j), and low-pass filter sections 60A that output identification signals r0 (i, j), r1 (i, j) with different resolutions. r
2 (i, j),..., RN-1 (i, j),
And a signal synthesizing unit 60B that generates an identification signal r (i, j) which is one synthesized signal.

The low-pass filter section 60A includes low-pass filters (LPF) F0 and FP having different pass bandwidths.
, FN-1, and low-pass filters (LPF) F0, F1, F2,..., FN-1
, A feature amount xmax (i, j) is input to the corresponding low-frequency component, and identification signals r0 (i, j), r1 (i, j), r
2 (i, j),..., RN−1 (i, j).

The signal synthesizing section 60B includes a multiplying circuit M
0, M1, M2,..., MN-1, the identification signal r
0 (i, j), r1 (i, j), r2 (i, j), ...
.., RN-1 (i, j) are weighted and then added
6 to obtain an identification signal r which is a composite signal of 1
(I, j) is generated and output. At this time, the weighting coefficients w0, w1, w2,..., WN−1 in the multiplication circuits M0, M1, M2,..., MN−1 are set in advance so as to satisfy the following relational expression. You.

[0100]

[Equation 14]

Thus, in this embodiment, by setting the weighting coefficients w0, w1, w2,..., WN-1, the contour in the imaging result VT is not abnormally emphasized.

That is, as shown in FIG.
When (i, j) changes rapidly (FIG. 15
(A)), in the low frequency component r (i, j), the signal level changes as if the sudden change in the pixel value was mitigated. Low frequency component r based on this pixel value x (i, j)
In the case where the change of (i, j) is biased to the higher level side than the inflection point of the characteristic described above with reference to FIG. 6, the correction coefficient g () is simply obtained by the output signal of the low-pass filter as in the first embodiment. i, j), the pixel value x
Immediately before (i, j) suddenly changes, the pixel value is amplified by the extra gain, and immediately after the pixel value x (i, j) rapidly changes, the pixel value is amplified by a small gain. Output value y with abnormally enhanced contour
(I, j) (FIG. 15B) is obtained.

In this case, the pixel value of such an outline can be corrected with a substantially uniform gain to reduce the emphasis of an abnormal outline.

Thus, in this embodiment, by generating a correction coefficient from a plurality of low-frequency components, an effect similar to that of the first embodiment can be obtained by effectively avoiding the emphasis of an abnormal contour. It has been made possible.

According to the configuration shown in FIG. 14, by generating a correction coefficient from a plurality of low-frequency components, the effect similar to that of the first embodiment can be obtained by effectively avoiding abnormal contour enhancement. be able to.

(8) Eighth Embodiment FIG. 16 is a block diagram showing a gradation correction circuit applied to a television camera according to an eighth embodiment of the present invention. This gradation correction circuit 68 is applied instead of the gradation correction circuit 8 described above with reference to FIG. In the tone correction circuit 68, the same components as those of the above-described tone correction circuit 8 and the like are denoted by the corresponding reference numerals, and redundant description will be omitted.

In the gradation correction circuit 68, the area determination filter 70 determines the determination results r0 (i, j) and r1 (i) of the area to which the input image data belongs based on the resolution based on the feature amount xmax (i, j). , J), r2 (i, j), ...
.., RN−1 (i, j) are output. That is, the area determination filters 70 are low-pass filters (LPFs) F0, F1, F2,...
And each low-pass filter (LPF) F0,
F1, F2,..., FN-1 have feature amounts xmax (i,
j), and the corresponding low-frequency component is identified by the identification signal r0.
(I, j), r1 (i, j), r2 (i, j),.
Output as rN-1 (i, j).

The coefficient calculation circuit 71 outputs the identification signal r0 (i,
j), r1 (i, j), r2 (i, j),..., rN−
Correction coefficients g0 (i, j), g corresponding to 1 (i, j)
1 (i, j), g2 (i, j),..., GN−1 (i,
j), and the correction coefficients g0 (i, j), g1 (i, j), g2 (i, j),.
.., GN-1 (i, j) are combined to obtain a correction coefficient g of 1
And a coefficient synthesis unit 71B that generates (i, j).

Among them, the coefficient generation section 71A is provided with a predetermined coefficient calculation function Gk (k = 0, 1, 2,..., N-
1), identification signals r0 (i, j), r1 (i, j)
j), r2 (i, j),..., rN-1 (i, j), and corresponding correction coefficients g0 (i, j), g1 (i, j), g
2 (i, j),..., GN−1 (i, j). The coefficient calculators L0, L1, L2,.

On the other hand, the coefficient synthesizing section 71B uses the multiplication circuits M0, M1, M2,..., MN-1 to correct the correction coefficients g0 (i, j), g1 (i, j), g2 (i,
j),..., gN-1 (i, j) are weighted and then added by an adder circuit 76, whereby a correction coefficient g (i, i, 1) of 1 is obtained.
j) is generated and output. At this time, the multiplication circuit M
The weighting factors w0, w1, w2,..., WN−1 at 0, M1, M2,.
3) It is set in advance so as to satisfy the relational expression of the expression.

According to the configuration shown in FIG. 16, the same effect as in the seventh embodiment can be obtained by generating one correction coefficient from each of the low-frequency components of a plurality of systems and then generating one correction coefficient. Obtainable.

(9) Other Embodiments In the above-described fifth to eighth embodiments, the case where the maximum value of the pixel value is detected from the predetermined range in the vicinity of each pixel and used as the feature amount has been described. However, the present invention is not limited to this, and as described above with respect to the second to fourth embodiments, the minimum value may be used as the feature amount as necessary, or the maximum value and the minimum value may be combined to obtain the characteristic value. It may be an amount.

In the above-described embodiment, the case where the correction coefficient is generated by the characteristics described above with reference to FIGS. 6, 18, and 19 has been described. However, the present invention is not limited to this, and the present invention is not limited thereto. In such a configuration, a characteristic correction coefficient different from the characteristic described above in the above-described embodiment may be generated as necessary. For example, as shown in FIG. A level conversion function based on input / output characteristics that can be reduced may be used.

That is, in the conventional method, when such a function is used, since this function is not a monotonically increasing function, a pseudo contour may be generated in an image as a processing result. However, in the case where the area is determined by the low-pass filter and the processing is performed as in the above-described embodiment, in the vicinity area having a size corresponding to the pass band of the low-pass filter, the pixel value of which the magnitude relation of the pixel value is reversed is reversed. Large changes can be prevented. As a result, generation of a false contour can be effectively avoided.

Further, in the above-described embodiment, the case has been described where the coefficient calculation function G is generated by the arithmetic processing of the equation (5) using the level conversion function T, but the present invention is not limited to this, and the level conversion function The coefficient calculation function G may be arbitrarily set without using the function T.

Further, in the above-described embodiment, the case has been described where the gradation is corrected by the gradation correction circuit and then the dynamic range is suppressed by the subsequent signal processing circuit. However, the present invention is not limited to this, and the level conversion is not limited to this. These processes can be collectively executed by setting a function T and a coefficient calculation function G corresponding thereto.

That is, in the process of suppressing the dynamic range, it is required that the number of bits of the output pixel value is smaller than the number of bits of the input pixel value. Is set to the maximum value allowed for the output image, and the coefficient calculation function G is generated using the maximum value, whereby these processes can be executed collectively.

When the coefficient calculation function G is arbitrarily set without using the level conversion function T, the coefficient calculation function G may be set so as to satisfy the following equation. Where j
Represents the input pixel level, Lmax represents the maximum value of the input pixel level, and L0max represents the maximum value of the output pixel level.

[0119]

(Equation 15)

In the above embodiment, the case where the quantization circuit, the look-up table, and the interpolation circuit are used in the fifth and sixth embodiments has been described. However, the present invention is not limited to this. If necessary, all or any of these quantization circuits, look-up tables, and interpolation circuits can be applied to other than the fifth and sixth embodiments.

On the contrary, in the fifth and sixth embodiments, the quantization circuit may be omitted if necessary.

Further, in the above-described embodiment, a case has been described in which imaging results are processed by color filters having different arrangements between even-numbered lines and odd-numbered lines. However, the present invention is not limited to this. Can be widely applied to the processing of image data in which the amplitude-modulated color signal is superimposed on the luminance signal by time division.

Further, in the above-described embodiment, the case where the region to which the input image data belongs is determined by the processing of the low-pass filter based on the characteristic amount. However, the present invention is not limited to this. In the case where the feature amount is set for the similarity between the arbitrarily selected pixel and the neighboring pixels surrounding the pixel, and the area is sequentially enlarged from this pixel to determine the processing target image, the processing is performed by various processes. By determining the area of the target image, it is possible to obtain the same effect as in the above-described embodiment.

Further, in the above-described embodiment, the case where the present invention is applied to a television camera has been described. However, the present invention is not limited to this, and various images such as a television receiver, a video tape recorder, and a printer may be used. It can be widely applied to processing equipment.

[0125]

As described above, according to the present invention, a feature amount indicating a feature in a predetermined range in the vicinity of each pixel is detected, a region to which the input image data belongs is determined based on the feature amount, and the determination result is obtained. By correcting the pixel value by generating a correction coefficient based on the above, it is possible to correct the gradation by effectively avoiding partial contrast deterioration for an output signal of, for example, a single-chip solid-state imaging device. Can be.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a television camera according to a first embodiment of the present invention.

FIG. 2 is a plan view showing a color filter of a CCD solid-state imaging device of the television camera of FIG.

FIG. 3 is a signal waveform diagram showing an image pickup result by the color filter of FIG. 2;

FIG. 4 is a characteristic curve diagram for describing processing of an imaging result in the television camera of FIG. 1;

FIG. 5 is a schematic diagram illustrating an array of pixels in the television camera of FIG. 1;

FIG. 6 is a characteristic curve diagram for explaining a contrast correction coefficient g (i, j).

FIG. 7 is a signal waveform chart for describing processing of a gradation correction circuit in the television camera of FIG. 1;

FIG. 8 is a signal waveform diagram for describing processing of a gradation correction circuit at an input level different from that in FIG. 7;

FIG. 9 is a block diagram showing a gradation correction circuit applied to a television camera according to a second embodiment of the present invention.

FIG. 10 is a signal waveform diagram for explaining the operation of the gradation correction circuit of FIG. 9;

FIG. 11 is a block diagram showing a gradation correction circuit applied to a television camera according to a third embodiment of the present invention.

FIG. 12 is a block diagram showing a gradation correction circuit applied to a television camera according to a fifth embodiment of the present invention.

FIG. 13 is a block diagram showing a gradation correction circuit applied to a television camera according to a sixth embodiment of the present invention.

FIG. 14 is a block diagram showing a gradation correction circuit applied to a television camera according to a seventh embodiment of the present invention.

FIG. 15 is a signal waveform diagram for describing the operation of the gradation correction circuit in FIG. 14;

FIG. 16 is a block diagram showing a gradation correction circuit applied to a television camera according to an eighth embodiment of the present invention.

FIG. 17 is a characteristic curve diagram for explaining a level conversion function applied to a gradation correction circuit according to another embodiment.

FIG. 18 is a characteristic curve diagram for explaining a level conversion function applied to a conventional gradation correction suppression process.

FIG. 19 is a characteristic curve diagram for explaining a level conversion function applied to a gradation correction process according to another example different from FIG. 18;

FIG. 20 is a characteristic curve diagram for explaining a histogram equalization process.

[Explanation of symbols]

1. Television camera, 8, 18, 28, 38, 4
8, 58, 68 ... gradation correction circuit, 9, 19, 29 ...
Feature amount filter, 10, 40, 60, 70... Area determination filter, 11, 21, 31.
M0 to MN-1 multiplication circuit, 43 quantization circuit, 4
4, 54 ... look-up table, 55 ... interpolation circuit, 29C, F0 to FN-1 ... low-pass filter, L
0 to LN-1 coefficient calculating unit

[Procedure amendment]

[Submission Date] March 3, 2000 (200.3.3)

[Procedure amendment 2]

[Document name to be amended] Statement

[Correction target item name] Claims

[Correction method] Change

[Correction contents]

[Claims]

Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat II (Reference) H04N 9/69 H04N 1/40 101E (72) Inventor Kazuhiko Ueda 6-7-35 Kita Shinagawa, Shinagawa-ku, Tokyo Sony In-house F-term (reference) 5B057 AA20 BA02 BA24 BA29 CA01 CA16 CB01 CB16 CC03 CE11 CE16 DA08 DC03 DC05 DC36 5C065 AA01 BB12 CC01 DD02 DD17 EE03 EE07 GG03 GG05 GG15 A03 DD07 EC05 GB01 HA02 KA12 KC02 KC07 KD07 KE02 KE03 KE07 KE09 KE19 KG01 KM02 KM06 5C077 LL19 MM02 MM21 MM30 MP08 NP02 PP01 PP05 PP15 PP27 PP28 PP34 PP43 PP46 PP48 PP54 PP57 PP58 PP61 PQ08 PQ18 P22

Claims (30)

[Claims] 1. An image processing apparatus for correcting a gradation of image data in which a color signal is sequentially superimposed on a luminance signal by time division, wherein a characteristic amount indicating a characteristic of a predetermined range in the vicinity is provided for each pixel of the image data. Feature value detecting means for sequentially detecting, area determining means for determining a region to which the image data belongs based on the feature value and outputting a determination result, and correcting a pixel value of the image data according to the determination result An image processing apparatus comprising: a coefficient calculation unit that outputs a correction coefficient; and a correction unit that corrects a pixel value of the image data according to the correction coefficient. 2. The image processing apparatus according to claim 1, wherein the characteristic amount detecting means sequentially detects a maximum value of pixel values in the predetermined range in the vicinity as the characteristic amount. 3. The image processing apparatus according to claim 1, wherein the characteristic amount detecting means sequentially detects a minimum value of pixel values in the predetermined range in the vicinity as the characteristic amount. 4. The method according to claim 1, wherein the characteristic amount detecting means sequentially detects a maximum value and a minimum value of the pixel values in the predetermined range, and detects the characteristic amount based on the maximum value and the minimum value. The image processing device according to claim 1. 5. The feature amount detecting means sequentially detects a maximum value and a minimum value of pixel values in the predetermined range in the vicinity, and combines the maximum value and the minimum value according to an average value of the image data. The image processing apparatus according to claim 1, wherein the feature amount is detected. 6. The apparatus according to claim 1, wherein the area determining unit is a low-pass filter that extracts a low-frequency component of the feature amount, and the coefficient calculating unit generates the correction coefficient according to the low-frequency component. The image processing device according to claim 1. 7. The area determining means includes: a quantizing means for quantizing the feature quantity; and a low-pass filter for extracting a low frequency component from the feature quantity quantized by the quantizing means. The image processing apparatus according to claim 1, wherein the calculating unit generates the correction coefficient according to the low frequency component. 8. The region determining means generates one composite signal based on a plurality of low-pass filters for extracting low-frequency components of the feature amount, respectively, and a low-frequency component output from the plurality of low-pass filters. The image processing apparatus according to claim 1, further comprising: a signal synthesizing unit configured to generate the correction coefficient based on the synthesized signal. 9. The image processing apparatus according to claim 8, wherein said signal synthesizing unit generates the synthesized signal by performing weighted averaging of low-frequency components output from the plurality of low-pass filters. 10. The signal synthesizer according to claim 1, wherein the low-frequency components output from the plurality of low-pass filters are weighted and added by a weighting coefficient set in advance to generate the synthesized signal. 9. The image processing device according to 8. 11. The area determination means has a plurality of low-pass filters for extracting low-frequency components of the feature quantity, respectively, and the coefficient calculation means calculates a low-frequency component from the low-frequency components output from the plurality of low-pass filters. The image processing apparatus according to claim 1, further comprising: a partial coefficient calculation unit configured to generate a correction coefficient; and a coefficient synthesis unit configured to generate the correction coefficient based on the correction coefficient. 12. The image processing apparatus according to claim 11, wherein said coefficient synthesizing unit generates the correction coefficient by performing a weighted average of the correction coefficient. 13. The image processing apparatus according to claim 11, wherein said coefficient synthesizing unit generates the correction coefficient by weighting and adding the correction coefficient using a weighting coefficient set in advance. . 14. The image processing apparatus according to claim 1, wherein said correction means corrects a pixel value of said image data by multiplying a pixel value of said image data by said correction coefficient. 15. The image processing apparatus according to claim 1, wherein the number of bits of the image data output from the correction unit is reduced as compared with the number of bits of the image data input to the feature amount detection unit. An image processing apparatus according to claim 1. 16. An image processing method for correcting the gradation of image data in which a color signal is sequentially superimposed on a luminance signal by time division, wherein a characteristic amount indicating a characteristic of a predetermined range in the vicinity is provided for each pixel of the image data. A feature value detection process of sequentially detecting, a region determination process of determining a region to which the image data belongs based on the feature value and outputting a determination result, and correcting a pixel value of the image data according to the determination result An image processing method, comprising: a coefficient calculation process of outputting a correction coefficient; and a correction process of correcting a pixel value of the image data according to the correction coefficient. 17. The image processing method according to claim 16, wherein in the feature amount detection processing, a maximum value of pixel values in the vicinity predetermined range is sequentially detected as the feature amount. 18. The image processing method according to claim 16, wherein in the feature amount detection processing, a minimum value of pixel values in the predetermined range in the vicinity is sequentially detected as the feature amount. 19. The feature amount detection process includes sequentially detecting a maximum value and a minimum value of the pixel values in the predetermined range in the vicinity, and detecting the feature amount based on the maximum value and the minimum value. 17. The image processing method according to claim 16, wherein: 20. The feature amount detecting process sequentially detects a maximum value and a minimum value of pixel values in the predetermined range in the vicinity, and combines the maximum value and the minimum value according to an average value of the image data. 17. The image processing method according to claim 16, wherein the feature amount is detected. 21. The method according to claim 16, wherein the area determination process extracts a low frequency component of the feature amount, and the coefficient calculation process generates the correction coefficient according to the low frequency component. The image processing method described in the above. 22. The area determination processing includes: quantizing the feature value; extracting a low-frequency component from the feature value quantized by the quantization process; 17. The image processing method according to claim 16, wherein the correction coefficient is generated according to the value. 23. The area determination processing includes: a signal extraction processing of extracting a plurality of low frequency components of the feature amount in different bands; and a signal synthesis processing of generating one synthesized signal based on the plurality of low frequency components. 17. The image processing method according to claim 16, wherein the coefficient calculation process generates the correction coefficient based on the composite signal. 24. The image processing method according to claim 23, wherein said signal synthesizing process generates the synthesized signal by weighted averaging the plurality of low frequency components. 25. The image processing apparatus according to claim 23, wherein in the signal combining processing, the combined signal is generated by weighting and adding the plurality of low frequency components using a weighting coefficient set in advance. Method. 26. The area determination processing, wherein a plurality of low frequency components of the feature amount are extracted from different bands, and the coefficient calculation processing generates a correction coefficient from each of the plurality of low frequency components. 17. The image processing method according to claim 16, further comprising: a coefficient calculation process; and a coefficient synthesis process for generating the correction coefficient based on the correction coefficient. 27. The image processing method according to claim 26, wherein in the coefficient combining processing, the correction coefficient is generated by weighted averaging of the correction coefficient. 28. The image processing method according to claim 26, wherein in the coefficient synthesizing process, the correction coefficient is generated by weighting and adding the correction coefficient using a weighting coefficient set in advance. . 29. The image processing method according to claim 16, wherein said correction processing corrects a pixel value of said image data by multiplying a pixel value of said image data by said correction coefficient. 30. The image processing method according to claim 16, wherein the number of bits of the image data output from the correction processing is reduced as compared with the number of bits of the input image data.