JP2010026690A - Image processing device - Google Patents

Abstract

An object of the present invention is to obtain an image processing apparatus that does not cause color collapse in a high gradation region and a low gradation region.
According to the image processing apparatus of the present invention, the gradation correction parameter calculation means 8 reads a conversion ratio corresponding to the maximum component value of the input image signal from the gradation correction table and calculates it as a gradation correction parameter. Therefore, even if this gradation correction parameter is multiplied by another component smaller than the maximum component value, the maximum gradation is not exceeded and color collapse does not occur.
Further, since the local correction means 6 locally corrects the magnitude of the input image signal in accordance with the peripheral brightness of the target pixel, the local dark portion image signal is strengthened in advance and the gradation difference of this portion is expanded. Even when correction is performed such that the gradation difference in the dark portion is compressed according to the gradation correction table curve, gradation information is maintained and color collapse does not occur.
[Selection] Figure 1

Description

  The present invention relates to an image processing apparatus that corrects and outputs the gradation of an input image signal.

  In the conventional image processing apparatus, the input image signal is corrected as follows. First, a luminance signal is generated from each component (R (red), G (green), and B (blue)) of the input image signal, and for this luminance signal, a distribution of appearance frequencies in one screen is extracted, A gradation correction table is created so that many gradations are assigned to many luminance areas. Next, the conversion ratio corresponding to the gradation level of the luminance signal of each pixel is read from the gradation correction table, and correction is performed by multiplying each component of the input image signal of each pixel by this conversion ratio.

  Since each component of the input image signal is corrected using the gradation correction table created as described above, it is possible to subdivide light and darkness in a luminance region with many distributions and improve contrast. In addition, since each component of the input image signal is multiplied by a common conversion ratio, the color balance hardly changes between the input signal and the output signal (see, for example, Patent Document 1).

JP 2004-342030 A (page 5-14, FIG. 1-13)

Here, the luminance signal (Y) from each component (R, G, B) of the input image signal is calculated by the following equation.
Y = α · R + β · G + γ · B; α = 0.30, β = 0.59, γ = 0.11.
When the conversion ratio corresponding to the gradation level of the luminance signal calculated by the above equation is read from the gradation correction table, the conversion ratio may be larger than 1. If a component before correction is close to the maximum gradation and this component is multiplied by this conversion ratio, the corrected component will exceed the maximum gradation. The output is forcibly suppressed to the maximum gradation by a circuit or the like. Therefore, in such a case, the gradation difference near the maximum gradation is not properly expressed, and there is a problem that color collapse occurs.

  In addition, a screen in which a dark area (for example, a shadow portion) of a small area exists in a bright screen as a whole, that is, a pixel having a high gradation in all pixels constituting one screen is frequently present. When the appearance frequency of the low gradation pixel is low, if the gradation correction table in the conventional image processing apparatus is used, only a small gradation can be assigned to the low gradation pixel. However, there is a problem that the gradation difference after correction disappears and color collapse occurs in this case.

  The present invention has been made to solve the above-described problems, and an object thereof is to obtain an image processing apparatus that does not cause color collapse.

In the image processing apparatus according to the present invention, the maximum component value detection unit detects the maximum of the plurality of components of the input image signal for each pixel as the maximum component value, and the gradation correction parameter calculation means detects this from the gradation correction table The conversion ratio corresponding to the maximum component value is read and calculated as a gradation correction parameter, and the multiplication means multiplies each component of the input image signal by this gradation correction parameter to generate an output image signal.
Further, in the image processing apparatus according to the present invention, the peripheral lightness detection means calculates the peripheral lightness in a predetermined area around each pixel, and the local correction value calculation means calculates the local correction value for the pixel based on the peripheral lightness The local correction means corrects the magnitude of the input image signal in the pixel based on the local correction value.

According to the image processing apparatus of the present invention, the gradation correction parameter calculating means reads the conversion ratio corresponding to the maximum component value of the input image signal from the gradation correction table and calculates it as the gradation correction parameter. Even if the gradation correction parameter is multiplied by another component smaller than the maximum component value, the maximum gradation is not exceeded, and color collapse does not occur.
Further, according to the image processing apparatus of the present invention, since the local correction unit locally corrects the magnitude of the input image signal according to the peripheral brightness of the target pixel, the local dark portion image signal is preliminarily obtained. The gradation difference of this part can be enlarged, and even when correction is performed so that the gradation difference of this dark part is compressed according to the gradation correction table curve, the gradation information is maintained. Color collapse does not occur.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing a configuration of an image processing apparatus according to Embodiment 1 of the present invention.
The image processing apparatus includes a first maximum value detection unit 1, a maximum value distribution extraction unit 2, a gradation correction table creation unit 3, a peripheral brightness detection unit 4, a local correction value calculation unit 5a, a local correction unit 6, and a second correction unit. Maximum value detecting means 7, gradation correction parameter calculating means 8, and multiplying means 9.
Input image signals R1, G1, and B1 are input to the image processing apparatus. Here, the input image signals R1, G1, and B1 are primary color signals representing the sizes of red, green, and blue components in the respective pixels.

  In this embodiment, the maximum value of each component of the input image signal at each pixel is used as a feature value for extracting the distribution when creating the gradation correction table. This is obtained from the input signal for each pixel. For example, a luminance signal calculated from each component may be used. In this case, a configuration of a luminance calculation unit instead of the first maximum value detection unit 1 and a luminance distribution extraction unit instead of the maximum value distribution extraction unit 2 may be adopted.

Next, the operation of each component will be described below.
The input image signal is input to the first maximum value detection means 1. The first maximum value detection means 1 detects the maximum gradation in real time from the components R1, G1, B1 of the input image signal for each pixel, and outputs this to the maximum value distribution extraction means 2. For example, when R1 = 10, G1 = 20, and B1 = 30, the value 30 of B1 is output.

The maximum value distribution extraction unit 2 extracts the distribution of the maximum value of the input image signal for each pixel for a certain period, for example, one screen (one frame), and outputs it to the gradation correction table creation unit 3. If the number of pixels for one screen (one frame) is 1 million pixels, it is extracted how the maximum value of the input image signal for 1 million pixels is distributed.
The distribution may be extracted for each gradation value (for example, when the total gradation is expressed by a 10-bit digital signal, the total number of gradations is 1024. In this case, 1024 gradation distributions are extracted. ) As the number of gradations increases, the circuit scale also increases. In this case, by optimizing the circuit scale to match the effect of improving contrast by collecting multiple gradation values as one class and extracting the distribution by calculating the appearance frequency for each class Can do. FIG. 2 shows an example of a histogram when the input frequency of 1024 gradations is divided into 16 classes and the appearance frequency is obtained.

The gradation correction table creation means 3 creates a gradation correction table from the maximum value distribution of the input image signal input from the maximum value distribution extraction means 2, for example, a histogram as shown in FIG. For example, FIG. 3 shows a case where a gradation correction table created from a 16-class histogram is created. The horizontal axis in FIG. 3 is the gradation of the maximum value of the input image signal, and the vertical axis is the normalized appearance frequency (bar graph) for each class or the corrected gradation (line graph).
The bar graph is obtained by dividing the appearance frequency for each class shown in FIG. 2 by dividing (the maximum value of the gradation range) by (the number of pixels in one frame), that is, multiplying by 1023 / 1,000,000 in this example. Indicates the normalized value (normalized appearance frequency). The curve extending from the lower left to the upper right is the curve corresponding to the gradation correction table for converting the input image signal to the output image signal, and the normalized value of the frequency up to that class, with the maximum value of each class as the horizontal axis coordinate value. Is the vertical axis coordinate value.
By creating a gradation correction table curve in this way, this curve has a large slope in a class or gradation area with a high appearance frequency, and conversely, a slope becomes small in a class or gradation area with a low appearance frequency. The maximum gradation (1023) of the signal matches before and after the correction by the gradation correction table.

In addition, a method other than the above is used to create a gradation correction table curve so that the slope increases in a class or gradation area with a high appearance frequency and conversely decreases in a class or gradation area with a low appearance frequency. Also good.
It is also possible to set a lower limit value and an upper limit value for the slope of the curve, and to limit the gradation correction parameter curve obtained. In this case, in each class, a class other than the class in which the slope of the curve is limited so that the sum of the increments in the vertical axis along the curve (Δy in FIG. 3) becomes the maximum gradation (1023 in this example). It is desirable to adjust the class together with the slope of the curve.

The peripheral brightness detection means 4 calculates a brightness signal from the input image signal at each pixel, and at the same time, a predetermined local region around the pixel (for example, a range of 5 pixels vertically and 5 pixels horizontally centering on the pixel). A lightness signal is calculated and output to the local correction value calculating means 5a as the peripheral lightness YL at the pixel.
Here, the lightness signal in each pixel can be obtained, for example, by adding the components R1, G1, and B1 of the input image signal in each pixel or adding each component with a weight. In addition, the brightness signal in a predetermined local area may be obtained as an average value of the brightness signal of each pixel constituting this area, or near the center when the brightness signal of each pixel is arranged from the minimum value to the maximum value. You may select and find what is in

The local correction value calculation means 5 a calculates the local correction value KL from the peripheral lightness YL obtained by the peripheral lightness detection means 4 and outputs it to the local correction means 6. The local correction value calculation means 5a can be realized, for example, by specifying the read address by the value of the peripheral brightness YL and storing the corresponding local correction value KL in the lookup table in advance.
FIG. 4 is a diagram illustrating an example of the relationship between the peripheral brightness YL and the local correction value KL. In FIG. 4, the local correction value KL is 1.0 when the peripheral lightness is greater than a predetermined value, and increases as the peripheral lightness decreases when the peripheral lightness is smaller than the predetermined value. Here, the predetermined value can be arbitrarily determined based on which brightness region is to be strengthened in advance according to the input image signal.

  The local correction unit 6 calculates the second image signals R2, G2, and B2 by multiplying the components R1, G1, and B1 of the input image signal by the local correction values KL, respectively. Therefore, when the peripheral brightness YL and the local correction value KL are in the relationship shown in FIG. 4, the input image signals R1, G1, and B1 are multiplied by a large value when the peripheral brightness YL is small (when the periphery is dark). Become. That is, when the periphery of the pixel is dark, a large value is multiplied, and the brightness is improved by improving the signal gain.

  The second maximum value detection means 7 detects the maximum value from the components R2, G2, B2 of the second image signal obtained by the local correction means 6 for each pixel, and calculates a gradation correction parameter. Output to means 8. The processing content is the same as that of the first maximum value detecting means 1.

The gradation correction parameter calculation means 8 is the gradation correction table curve obtained by the gradation correction table creation means 3 and the maximum value of each component of the second image signal obtained by the second maximum value detection means 7. Is used to calculate a conversion ratio to be multiplied with respect to the second image signals R2, G2, and B2, that is, a gradation correction parameter, and output the result to the multiplication unit 9.
A calculation example is shown in FIG. In FIG. 5, the curve extending from the lower left to the upper right is the same gradation correction table curve as in FIG. For a certain pixel, the maximum value of each component of the second image signal in the pixel is input from the second maximum value detecting means 7. A straight line drawn from the origin toward this point (x _m , y _m ), where x _m is the maximum value and y _m is the coordinate of the vertical axis corresponding to x _m on the gradation correction table curve. a gradient g _m is the conversion ratio to be obtained (tone correction parameter).
There are several methods for determining the slope g _m, as an example, the slope g ₂ in the upper tone x ₂ of the class that contains the maximum value x _m, floor upper ranks under one of the class from the slope g ₁ in tone x _1, there is a method of calculation by internally dividing by formula 1 below.

This method causes some mathematical errors, but not at a problematic level. Further, the divisor is a fixed value for calculating the g _m (in the above example, x 2 _{-x 1)} because it is, without using the divider, it is possible to reduce the circuit scale. For example, if this fixed value is 2 ⁿ (n is an integer of 1 or more), division can be performed by shift processing.
Of course, since the gradation correction table curve is known, the slope g _m may be directly obtained by the following equation 2 using a divider or the like. y ₁ and y ₂ are the coordinates of the vertical axis corresponding to each of the gradations x ₁ and x ₂ on the gradation correction table curve.

  The multiplication means 9 multiplies each component R2, G2, B2 of the second image signal by the gradation correction parameter obtained by the gradation correction parameter calculation means 8, and calculates the output image signals R3, G3, B3. . The multiplication means 9 can be configured using a normal multiplier that does not require any particular explanation.

Next, the effect of the image processing apparatus according to the present embodiment will be described with reference to FIG. The gradation correction table curve shown in FIG. 6 is different from those shown in FIGS. 3 and 5 in order to make the effect of the present invention easy to understand.
Symbols R, G, and B written below the horizontal axis are the gradations of the components R2, G2, and B2 of the second image signal, respectively. Y is the gradation of the luminance signal calculated from the gradation of each component R2, G2, B2. The gradients of the straight lines connecting the corresponding points on the R, G, B, and Y gradation correction table curves and the origin are g _r , g _g , g _b , and g _y , respectively. According to the image processing apparatus according to this embodiment, the maximum value of each component is tone of B, the conversion ratio of slope g _b are multiplied by the multiplication means 9, that is, tone correction parameter.

Here, if the gradient g _g calculated from G is multiplied by B, the multiplication result is BG0, and if the gradient g _y calculated from Y is multiplied by B as in the conventional image processing apparatus, The multiplication result is BY0, both of which exceed the maximum gradation (1023). Actually, since the output is limited to the maximum gradation value by a clip circuit or the like, the gradation difference in the vicinity is not properly expressed and color collapse occurs (in this example, the gradation of B (blue) is destroyed) ).
On the other hand, according to the image processing apparatus according to the present embodiment, the gradient g _b does not exceed the maximum gradation (1023 in this example) even if it is multiplied by B which is the component of the maximum value. since the correction table curves is defined, it does not exceed the maximum gradation be multiplied by g _b small R and G than B. Therefore, color collapse does not occur.

  Further, the local correction means 6 uses the local correction value KL shown in FIG. 4 to strengthen the second image signals R2, G2, and B2 in the dark part in advance, and enlarge the gradation difference in the dark part. For this reason, even if correction is performed on a screen with a low distribution of appearance frequency of dark areas so that the gradation difference in the dark areas is compressed according to the gradation correction table curve, the gradation information is maintained and the color is maintained. There is no crushing.

Further, as in the conventional image processing apparatus, the gradation correction table creating means 3 creates a gradation correction table so that many gradations are assigned to a gradation region having a high frequency distribution as shown in FIG. However, the contrast in the area can be subdivided, so that the contrast can be improved.
Further, since the multiplication means 9, to obtain an output image signal (R3, G3, B3) by multiplying the common tone correction parameter _{g b} a second image signal (R2, G2, B2), color change There is also an effect that can be prevented.

Embodiment 2. FIG.
FIG. 7 is a block diagram showing a configuration of an image processing apparatus according to Embodiment 2 of the present invention.
Since the present embodiment and Embodiment 1 are different only in the local correction value calculation means 5b, the description of other configurations is omitted, and the operation of the local correction value calculation means 5b will be described below.

The local correction value calculation means 5b refers to the gradation correction table curve created by the gradation correction table creation means 3, in particular the slope in the low gradation area, in addition to the surrounding brightness YL output from the surrounding brightness detection means 4. Thus, a local correction value KL is generated.
FIG. 8 is a diagram showing an example of the relationship between the peripheral brightness YL and the local correction value KL. FIG. 8A shows a case where the gradient of the low gradation region in the gradation correction table curve is small, and FIG. When the slope of the area is medium, (C) is when the slope of the low gradation area is large.

  When the gradient of the low gradation area in the gradation correction table curve is small, the gradation difference of the input image signal in this area is greatly compressed and corrected. It will happen. In such a case, by using the local correction value KL having the characteristics shown in FIG. 8A, the dark portion signal is increased in advance and the number of gradations assigned to this portion is increased. Can be maintained and color collapse can be prevented.

On the contrary, when the gradient of the low gradation area of the gradation correction table curve is large, the gradation difference of the input image signal in this area is not compressed, and color crushing occurs even if the gradation difference in the dark part is not enlarged in advance. There is no waking. Therefore, a curve shown in FIG.
In the first embodiment, the same local correction value KL is used uniformly regardless of the condition of the gradation correction table curve. Therefore, when the gradient of the low gradation area of the gradation correction table curve is large, the dark portion is bright. However, according to the present embodiment, it is possible to improve such a state and to prevent the dark part from becoming too bright.

It is a block diagram which shows the structure of the image processing apparatus which concerns on Embodiment 1 of this invention. In Embodiment 1 of this invention, it is a figure which shows the example of distribution of the maximum value of an input image signal. In Embodiment 1 of this invention, it is a figure which shows the example which produces a gradation correction table curve from distribution of the maximum value of an input image signal. In Embodiment 1 of this invention, it is a figure which shows an example of the relationship between the peripheral brightness YL and the local correction value KL. It is a figure which shows the example which calculates a gradation correction parameter in Embodiment 1 of this invention. It is a figure which shows the effect of the image processing apparatus which concerns on Embodiment 1 of this invention. It is a block diagram which shows the structure of the image processing apparatus which concerns on Embodiment 2 of this invention. In Embodiment 2 of this invention, it is a figure which shows an example of the relationship between the peripheral brightness YL and the local correction value KL.

Explanation of symbols

2 Maximum value distribution extraction means 3 Gradation correction table creation means 4 Peripheral brightness detection means 5a Local correction value calculation means 5b Local correction value calculation means 6 Local correction means 7 Second maximum value detection means 8 Tone correction parameter calculation means 9 Multiplication means

Claims (6)

A distribution extracting means for extracting a distribution of a certain period with respect to the appearance frequency of the feature amount of the input image signal;
A gradation correction table creating means for creating a gradation correction table based on the distribution;
Surrounding brightness detection means for calculating a surrounding brightness in a predetermined area around the pixel from the input image signal for each pixel;
Local correction value calculating means for calculating a local correction value in the pixel based on the peripheral brightness;
Local correction means for generating a second image signal by correcting the magnitude of the input image signal at the pixel based on the local correction value;
Maximum component value detecting means for detecting a maximum one of the plurality of components of the second image signal as a maximum component value;
Gradation correction parameter calculation means for calculating a gradation correction parameter from the maximum component value and the gradation correction table;
Multiplication means for multiplying the gradation correction parameter and the second image signal to generate an output image signal;
An image processing apparatus. The feature amount is a maximum one of a plurality of components of the input image signal,
The image processing apparatus according to claim 1. The fixed period is one frame period,
The image processing apparatus according to claim 1. The local correction value calculation means sets the local correction value to 1 when the peripheral brightness is larger than the predetermined value, and sets the local correction value to 1 or more when the surrounding lightness is equal to or lower than the predetermined value.
The local correction means generates the second image signal by multiplying the input image signal by the local correction value,
The image processing apparatus according to claim 1. The distribution extraction means divides the gradation range of the feature amount of the input image signal into a plurality of classes, extracts a distribution of the appearance frequency of the feature quantity for each class for a certain period,
The gradation correction table creating means creates a line graph in which the cumulative value of the normalized appearance frequency from the minimum class of the gradation range to each class is associated with the maximum gradation of each class. Features
The image processing apparatus according to claim 1. The local correction value calculating means calculates the local correction value with reference to the gradation correction table,
The image processing apparatus according to claim 1.

Images