JP5137464B2 - Image processing apparatus, image processing method, image processing program, and integrated circuit - Google Patents

Description

  The present invention relates to an image processing device, an image processing method, an image processing program, and an integrated circuit, and in particular, an image processing device, an image processing method, an image processing program, and an integrated circuit that perform gradation conversion without enhancing noise of an input signal. About.

In general, in an imaging device or a display device, gradation conversion is performed to correct the contrast and contrast of an input signal. As gradation conversion, processing in units of pixels and processing for referring to a peripheral region are known.
The pixel-by-pixel process is a conversion process that is performed based only on the pixel value of the target pixel without referring to pixels other than the target pixel. As a specific example, there is a gamma correction process applied to a captured image in order to cancel the photoelectric conversion characteristics of the CRT display device.
On the other hand, the process of referring to the peripheral area is a conversion process performed by referring not only to the pixel value of the target pixel but also to the peripheral value of the target pixel or the pixel value of the entire image. As a specific example, histogram equalization is performed in which gradation conversion is performed on an input signal by obtaining an appearance frequency distribution of pixel values in the input signal and assigning many gradations to gradation levels having a high appearance frequency in the input signal. (In the case of the histogram equalization method, the range of gradation values having a high appearance frequency in the input signal is a narrow range (for example, when the gradation value of 5 to 20 is taken in the gradation value of 8-bit data) ), Gradation conversion is performed so that the output signal takes a wide range of gradation values (for example, gradation values of 10 to 120). As another specific example, there is visual processing that performs conversion processing based on the average value (average pixel value) of the pixel value of the target pixel and the pixel values of the peripheral region.

By applying these gradation conversions to the input signal, a converted signal with improved brightness and contrast appearance (feeling) can be obtained. However, when a pixel having a small pixel value such as gamma correction is processed with a high gain, a noise component slightly generated in the input signal is amplified, and the S / N ratio is greatly deteriorated. As a technique for improving this problem, a technique for performing noise reduction processing on a converted signal for a pixel having a small pixel value is disclosed (for example, Patent Document 1). According to this technique, noise reduction processing is performed only on pixels whose slope of the gamma curve representing the input / output characteristics of gradation conversion is larger than a predetermined threshold. When the input / output characteristics of gradation conversion are determined in advance, as in gamma correction, the input / output characteristics have a monotone increasing characteristic, and the slope of the input / output characteristic curve representing the input / output characteristics has a monotonic decreasing characteristic Since the gain increases as the input signal becomes lower (the level of the input signal having a smaller gradation value), a pixel whose S / N ratio deteriorates by detecting an input signal that is equal to or lower than the threshold value, that is, a pixel A pixel to be processed with a high gain is specified by a pixel having a small value. By performing noise reduction processing on the pixels thus identified, it is possible to improve the S / N ratio deterioration in the dark part.
JP 2001-309177 A

However, in processes that refer to peripheral areas such as histogram equalization and visual processing, the input / output characteristics of the gradation conversion depend on the frequency distribution of pixel values in the input signal and the average pixel value around the target pixel for each image. Or it changes for every pixel. In order to apply a conventional method such as gamma correction to the process of referring to these peripheral regions, it is necessary to calculate in advance a pixel value in which the slope of the input / output characteristic curve for gradation conversion is smaller than a predetermined threshold value. Therefore, when applied to a movie, search processing is performed on the input / output characteristics of the gradation conversion obtained for each frame (or each field) or for each pixel, and the obtained gradation conversion is input. It is necessary to calculate the pixel level (pixel value) of the input signal corresponding to the threshold value of the slope of the output characteristic curve. However, this processing requires a large amount of computation.
Also, in general tone conversion, there is no guarantee that the slope of the input / output characteristic curve of the input / output characteristics decreases monotonously like gamma correction, so the criterion for pixel values whose slope is below the threshold is the noise reduction process. There is a possibility that the range of the pixel level (pixel value) of the input signal to which is applied cannot be uniquely determined.

In addition, since it is only a process of switching whether to apply the noise reduction process by threshold processing for each pixel, the noise reduction process cannot be performed with an intensity suitable for each pixel.
According to the present invention, not only in the case of gradation conversion that refers only to the pixel of interest, but also in the case of gradation conversion that refers to the peripheral region of the pixel of interest, depending on the degree of deterioration of the S / N ratio due to amplification of noise components An object of the present invention is to provide an image processing apparatus, an image processing method, an image processing program, and an integrated circuit capable of performing noise reduction processing with different strength for each pixel of interest.

1st invention determines the noise reduction degree with respect to the conversion signal based on an input signal and a conversion signal, and a gradation conversion part which performs gradation conversion to the input signal which consists of pixel data which forms an image, and obtains a conversion signal An image processing apparatus includes: a noise reduction degree determination unit; and a noise reduction unit that performs noise reduction processing on a converted signal based on the noise reduction degree.
In this image processing apparatus, a noise reduction degree representing the strength of noise reduction processing applied to the converted signal is calculated from the input signal and the converted signal of each pixel, and noise is calculated based on the calculated noise reduction value. The strength of the reduction process can be adjusted. As a result, it is possible to perform noise reduction processing in accordance with the amplification degree of the noise component of each pixel accompanying gradation conversion.
The second invention is the first invention, wherein the gradation conversion unit performs gradation conversion based on histogram information on the gradation value of the pixel data of the input signal.

In this image processing apparatus, for example, even when a histogram equalization method is used as gradation conversion, it is possible to perform noise reduction processing according to the amplification degree of the noise component of each pixel accompanying gradation conversion.
3rd invention is 1st invention, Comprising: A gradation conversion part performs gradation conversion based on the signal which performed the predetermined process with respect to the pixel data of the attention pixel data of an input signal, and the surrounding pixel data of attention pixel data I do.
In this image processing apparatus, even when visual processing is performed on an input signal, conversion processing with suppressed noise enhancement can be performed. Here, visual processing (spatial visual processing) is processing that performs gradation correction on a pixel of interest (region) using gradation characteristics that change according to the brightness around the pixel of interest (region). Say. For example, when the gradation value of the pixel of interest in the input signal is a small gradation value (for example, “50” in the 8-bit gradation value) and the brightness around the pixel of interest is dark ( For example, for an 8-bit gradation value “20”, gradation correction is performed so that the gradation value of the target pixel becomes a large gradation value (for example, “150” for an 8-bit gradation value). . Conversely, when the gradation value of the pixel of interest in the input signal is a small gradation value (eg, 8-bit gradation value “50”), and the surrounding brightness of the pixel of interest is bright (For example, “100” for an 8-bit gradation value), gradation correction is performed so that the gradation value of the target pixel becomes a small gradation value (for example, “30” for an 8-bit gradation value). Do. Such processing is an example of spatial visual processing.

4th invention is 1st invention, Comprising: A noise reduction degree determination part determines noise reduction degree based on ratio of an input signal and a conversion signal.
According to this, the strength of the noise reduction process can be adjusted based on the gain applied to the input signal.
5th invention is 1st invention, Comprising: A noise reduction degree determination part determines a noise reduction degree based on the ratio of an input signal and a conversion signal, and an input signal.
According to this, the strength of the noise reduction process can be adjusted based on the input signal and the gain applied to the input signal.
6th invention is 5th invention, Comprising: The noise reduction degree determination part has the gradation conversion gain which is ratio (= (conversion signal) / (input signal)) with respect to the input signal of a conversion signal 1st When it is larger than the gain threshold and the signal level of the input signal is smaller than the first signal level threshold, the noise reduction degree is set to the first noise reduction degree. In addition, the noise reduction degree determination unit is configured when the gradation conversion gain is smaller than the second gain threshold value that is smaller than the first gain threshold value, and the signal level of the input signal is for the first signal level. If it is smaller than the threshold, the noise reduction degree is set to a second noise reduction degree that is smaller than the first noise reduction degree. Further, the noise reduction degree determination unit sets the noise reduction degree to be smaller than the first noise reduction degree when the signal level of the input signal is larger than the second signal level threshold value which is larger than the first signal level threshold value. The value is set to the third noise reduction degree.

In this image processing apparatus, pixel data having a small input signal level (gradation value of pixel data of the input signal) and a large gradation conversion gain is subjected to strong noise reduction processing, the input signal level is small, and For pixel data with a small gradation conversion gain, weak noise reduction processing is performed. Also, weak noise reduction processing is performed on pixel data having a large input signal level.
As a result, even if the pixel data constituting the dark portion in the image formed by the input signal is amplified with a large gain to generate an output signal and the output signal is displayed on the display device, the display screen The brightness and contrast of the portion corresponding to the pixel data can be improved while suppressing noise components.
7th invention is 5th invention, Comprising: The noise reduction degree determination part is so large that the gradation conversion gain which is ratio (= (conversion signal) / (input signal)) with respect to the input signal of a conversion signal becomes large. The gain reference noise reduction degree is set to a large value, and the signal level reference noise reduction degree is set to a smaller value as the signal level of the input signal increases. The noise reduction degree determination unit determines the noise reduction degree based on the gain reference noise reduction degree and the signal level reference noise reduction degree.

Thereby, noise reduction processing according to the gradation conversion gain and the signal level of the input signal can be performed.
Here, “the gain reference noise reduction degree is set to a larger value as the gradation conversion gain increases” means, for example, that the gain reference noise reduction degree monotonously increases with a change in the gradation conversion gain. Say relationship. Moreover, not only a strict monotonous increase but also a substantial monotonous increase (some parts may not include a monotonic increase) may be included. For example, when the gradation conversion gain is smaller than a predetermined value (this is referred to as “first gain value”), the gain reference noise reduction degree is set to a predetermined value (this is referred to as “first gain reference noise reduction degree”). When the gradation conversion gain is not less than the first gain value and not more than the second gain value (> first gain value), the gain reference noise reduction degree is set to a value that monotonously increases with respect to the gradation conversion gain. When the gradation conversion gain is larger than the second gain value, the value is larger than the first gain reference noise reduction degree. Such cases are also included.

  Further, as a method for determining the noise reduction degree by the noise reduction degree determination unit, for example, an average value (including an arithmetic average value and a geometric average value) of the gain reference noise reduction degree and the signal level reference noise reduction degree is obtained. It is preferable to use a method of determining the average value as the noise reduction degree or a method of obtaining a weighted average value of the gain reference noise reduction degree and the signal level reference noise reduction degree and determining the weighted average value as the noise reduction degree. . Here, the weighted average value is a smaller value of the gain reference noise reduction degree and the signal level reference noise reduction degree (this is called “value A”, and the larger value is called “value B”. ) Is first obtained, the value A is subjected to large weighting, the value B is subjected to small weighting, and then average processing is performed to obtain the obtained value. For example, the weighted average value can be obtained by (weighted average value) = ((value A) × 3 + (value B)) / 4.

8th invention is 7th invention, Comprising: A noise reduction degree determination part sets the smaller value of a gain reference noise reduction degree and a signal level reference noise reduction degree to a noise reduction degree.
As a result, even if the pixel data constituting the dark portion in the image formed by the input signal is amplified with a large gain to generate an output signal and the output signal is displayed on the display device, the display screen The brightness and contrast of the portion corresponding to the pixel data can be improved while suppressing noise components.
9th invention is 5th invention, Comprising: The noise reduction degree determination part has a 1st noise reduction degree calculation part, a 2nd noise reduction degree calculation part, and a noise reduction degree output part. When the signal level of the input signal is greater than a predetermined signal level threshold, the first noise reduction degree calculation unit has a signal level reference noise having a smaller value than when the signal level of the input signal is less than or equal to the predetermined signal level threshold. Output the degree of reduction. The second noise reduction degree calculation unit determines that the gradation conversion gain is predetermined when the gradation conversion gain that is the ratio of the converted signal to the input signal (= (converted signal) / (input signal)) is greater than a predetermined gain threshold. The gain reference noise reduction degree having a larger value than that when the gain threshold value is less than or equal to is output. The noise reduction degree output unit sets a value calculated based on the signal level reference noise reduction degree and the gain reference noise reduction degree as the noise reduction degree.

Thereby, noise reduction processing according to the gradation conversion gain and the signal level of the input signal can be performed.
The tenth invention is the ninth invention, wherein the noise reduction degree output unit sets a smaller value of the signal level reference noise reduction degree and the gain reference noise reduction degree as the noise reduction degree.
With this configuration, pixel data having a small input signal level (gradation value of pixel data of the input signal) and a large gradation conversion gain is subjected to strong noise reduction processing, the input signal level is small, and For pixel data having a small tone conversion gain, weak noise reduction processing can be performed. Also, weak noise reduction processing can be performed on pixel data having a large input signal level.
As a result, even if the pixel data constituting the dark portion in the image formed by the input signal is amplified with a large gain to generate an output signal and the output signal is displayed on the display device, the display screen The brightness and contrast of the portion corresponding to the pixel data can be improved while suppressing noise components.
An eleventh aspect of the invention is based on a gain calculation unit that calculates a gain for converting an input signal composed of pixel data that forms an image, a multiplication unit that multiplies the input signal by the gain to obtain a converted signal, and a gain. An image processing apparatus includes: a noise reduction degree determination unit that determines a noise reduction degree for a converted signal; and a noise reduction unit that performs a noise reduction process on the converted signal based on the noise reduction degree.

According to this configuration, the noise reduction degree indicating the strength of the noise reduction processing applied to the converted signal is calculated from the input signal and gain of each pixel, and the noise is calculated based on the calculated noise reduction degree value. The strength of the reduction process can be adjusted. As a result, it is possible to perform noise reduction processing in accordance with the amplification degree of the noise component of each pixel accompanying gradation conversion.

The twelfth invention is the eleventh invention, wherein the noise reduction degree determination unit further determines the noise reduction degree for the converted signal based on the input signal.
In a thirteenth aspect based on the eleventh aspect, the gain calculation unit calculates the gain based on the histogram information regarding the gradation value of the pixel data of the input signal.
In this image processing apparatus, for example, even when the histogram equalization method is applied, it is possible to perform noise reduction processing according to the amplification degree of the noise component of each pixel accompanying gradation conversion.

14th invention is 11th invention, Comprising: A gain calculation part calculates a gain based on the signal which performed the predetermined process with respect to the attention pixel data of an input signal, and the surrounding pixel data of attention pixel data .
According to this, even when visual processing is performed on the input signal, it is possible to perform noise reduction processing according to the amplification degree of the noise component of each pixel accompanying gradation conversion.
A fifteenth aspect of the present invention is the twelfth aspect of the present invention, in which the noise reduction degree determining unit is configured such that the gain of the converted signal is greater than the first gain threshold and the signal level of the input signal is the first signal level. If it is smaller than the threshold value, the noise reduction degree is set to the first noise reduction degree. In addition, the noise reduction degree determination unit is configured when the gain is smaller than the second gain threshold that is a value smaller than the first gain threshold, and the signal level of the input signal is smaller than the first signal level threshold. In this case, the noise reduction degree is set to a second noise reduction degree that is smaller than the first noise reduction degree. Further, the noise reduction degree determination unit sets the noise reduction degree to be smaller than the first noise reduction degree when the signal level of the input signal is larger than the second signal level threshold value which is larger than the first signal level threshold value. The value is set to the third noise reduction degree.

In this image processing apparatus, pixel data having a small input signal level (tone value of pixel data of the input signal) and a large gain of the conversion signal is subjected to strong noise reduction processing, the input signal level is small, and The pixel data having a small gain of the conversion signal is subjected to weak noise reduction processing. Also, weak noise reduction processing is performed on pixel data having a large input signal level.
As a result, even if the pixel data constituting the dark portion in the image formed by the input signal is amplified with a large gain to generate an output signal and the output signal is displayed on the display device, the display screen The brightness and contrast of the portion corresponding to the pixel data can be improved while suppressing noise components.
The sixteenth invention is the twelfth invention, wherein the noise reduction degree determining unit sets the gain reference noise reduction degree to a larger value as the gain increases, and the signal level of the input signal increases as the signal level increases. The level reference noise reduction degree is set to a small value, and the noise reduction degree is determined based on the gain reference noise reduction degree and the signal level reference noise reduction degree.

Thereby, noise reduction processing according to the gradation conversion gain and the signal level of the input signal can be performed.
Here, “the gain reference noise reduction degree is set to a larger value as the gain increases” means, for example, a relationship in which the gain reference noise reduction degree monotonously increases with respect to a change in gain. Moreover, not only a strict monotonous increase but also a substantial monotonous increase (some parts may not include a monotonic increase) may be included. For example, when the gain is smaller than a predetermined value (this is referred to as “third gain value”), the gain reference noise reduction degree is fixed to a predetermined value (this is referred to as “third gain reference noise reduction degree”). When the gain is not less than the third gain value and not more than the fourth gain value (> third gain value), the gain reference noise reduction degree is set to a value that monotonously increases with respect to the gain. When the gain is larger than the fourth gain value, the gain is set to a value larger than the third gain reference noise reduction degree. Such cases are also included.

Further, as a method for determining the noise reduction degree by the noise reduction degree determination unit, for example, an average value (including an arithmetic average value and a geometric average value) of the gain reference noise reduction degree and the signal level reference noise reduction degree is obtained. It is preferable to use a method of determining the average value as the noise reduction degree or a method of obtaining a weighted average value of the gain reference noise reduction degree and the signal level reference noise reduction degree and determining the weighted average value as the noise reduction degree. .
The seventeenth invention is the sixteenth invention, wherein the noise reduction degree determination unit sets the smaller value of the gain reference noise reduction degree and the signal level reference noise reduction degree as the noise reduction degree.
As a result, even if the pixel data constituting the dark portion in the image formed by the input signal is amplified with a large gain to generate an output signal and the output signal is displayed on the display device, the display screen The brightness and contrast of the portion corresponding to the pixel data can be improved while suppressing noise components.

18th invention is 12th invention, Comprising: The noise reduction degree determination part has a 1st noise reduction degree calculation part, a 2nd noise reduction degree calculation part, and a noise reduction degree output part. When the signal level of the input signal is greater than a predetermined signal level threshold, the first noise reduction degree calculation unit has a signal level reference noise having a smaller value than when the signal level of the input signal is less than or equal to the predetermined signal level threshold. Output the degree of reduction. The second noise reduction degree calculation unit outputs a gain reference noise reduction degree having a larger value than when the gain is equal to or smaller than the predetermined gain threshold when the gain is larger than the predetermined gain threshold. The noise reduction degree output unit sets a value calculated based on the signal level reference noise reduction degree and the gain reference noise reduction degree as the noise reduction degree.
Thereby, noise reduction processing according to the gradation conversion gain and the signal level of the input signal can be performed.

In a nineteenth aspect based on the eighteenth aspect, the noise reduction degree output unit sets a smaller value of the signal level reference noise reduction degree and the gain reference noise reduction degree as the noise reduction degree.
In this image processing apparatus, pixel data having a small input signal level (tone value of pixel data of the input signal) and a large gain of the conversion signal is subjected to strong noise reduction processing, the input signal level is small, and The pixel data having a small gain of the conversion signal is subjected to weak noise reduction processing. Also, weak noise reduction processing is performed on pixel data having a large input signal level.
As a result, even if the pixel data constituting the dark portion in the image formed by the input signal is amplified with a large gain to generate an output signal and the output signal is displayed on the display device, the display screen The brightness and contrast of the portion corresponding to the pixel data can be improved while suppressing noise components.

In a twentieth aspect, a gradation conversion step for obtaining a converted signal by performing gradation conversion on an input signal composed of pixel data forming an image, and determining a noise reduction degree for the converted signal based on the input signal and the converted signal. An image processing method comprising: a noise reduction degree determination step; and a noise reduction step of performing noise reduction processing on a converted signal based on the noise reduction degree.
According to this configuration, the noise reduction degree indicating the strength of the noise reduction processing applied to the converted signal is calculated from the input signal and the converted signal of each pixel, and based on the calculated noise reduction value. The strength of the noise reduction process can be adjusted. As a result, it is possible to perform noise reduction processing in accordance with the amplification degree of the noise component of each pixel accompanying gradation conversion.
A twenty-first invention is based on a gain calculating step for calculating a gain for converting an input signal composed of pixel data forming an image, a multiplying step for multiplying the input signal by the gain to obtain a converted signal, and a gain. An image processing method comprising: a noise reduction degree determination step for determining a noise reduction degree for a converted signal; and a noise reduction step for applying a noise reduction process to the converted signal based on the noise reduction degree.

According to this configuration, the noise reduction degree indicating the strength of the noise reduction processing applied to the converted signal is calculated from the input signal and gain of each pixel, and the strength of the noise reduction processing is calculated based on this value. Can be adjusted. As a result, it is possible to perform noise reduction processing in accordance with the amplification degree of the noise component of each pixel accompanying gradation conversion.
The twenty-second invention is the twenty-first invention, wherein in the noise reduction degree determination step, the noise reduction degree for the converted signal is further determined based on the input signal.
In a twenty-third invention, a gradation conversion step for obtaining a conversion signal by performing gradation conversion on an input signal composed of pixel data forming an image in a computer, and a noise reduction degree for the conversion signal based on the input signal and the conversion signal And a noise reduction step of performing a noise reduction process on the converted signal based on the noise reduction degree.

According to this configuration, the noise reduction degree representing the strength of the noise reduction processing applied to the converted signal is calculated from the input signal and the conversion signal of each pixel, and the strength of the noise reduction processing is calculated based on this value. Can be adjusted. As a result, it is possible to perform noise reduction processing in accordance with the amplification degree of the noise component of each pixel accompanying gradation conversion.
According to a twenty-fourth aspect of the invention, a gain calculating step for calculating a gain for converting an input signal composed of pixel data forming an image in a computer, a multiplying step for multiplying the input signal by the gain to obtain a converted signal, and a gain An image processing program for executing a noise reduction degree determination step for determining a noise reduction degree to be applied to a converted signal based on the noise reduction step and a noise reduction step for reducing noise of the converted signal based on the noise reduction degree is there.
According to this configuration, the noise reduction degree indicating the strength of the noise reduction processing applied to the converted signal is calculated from the input signal and gain of each pixel, and the noise is calculated based on the calculated noise reduction degree value. The strength of the reduction process can be adjusted. As a result, it is possible to perform noise reduction processing in accordance with the amplification degree of the noise component of each pixel accompanying gradation conversion.

In a twenty-fifth aspect based on the twenty-fourth aspect, in the noise reduction degree determination step, the noise reduction degree for the converted signal is further determined based on the input signal.
According to a twenty-sixth aspect of the present invention, a gradation conversion unit that performs gradation conversion on an input signal composed of pixel data forming an image to obtain a converted signal, and determines a noise reduction degree for the converted signal based on the input signal and the converted signal. An integrated circuit comprising: a noise reduction degree determination unit; and a noise reduction unit that performs noise reduction processing on a converted signal based on the noise reduction degree.
Thus, an integrated circuit that exhibits the same effect as that of the first invention can be realized.
According to a twenty-seventh aspect, a gain calculating unit that calculates a gain for converting an input signal composed of pixel data that forms an image, a multiplying unit that multiplies the input signal by the gain to obtain a converted signal, and a gain based on the gain. An integrated circuit comprising: a noise reduction degree determination unit that determines a noise reduction degree for a converted signal; and a noise reduction unit that performs a noise reduction process on the converted signal based on the noise reduction degree.

  As a result, an integrated circuit having the same effect as that of the eleventh invention can be realized.

In the present invention, not only in the case of gradation conversion that refers only to the pixel of interest, but also in the case of gradation conversion that refers to the peripheral region of the pixel of interest, depending on the degree of degradation of the S / N ratio due to amplification of the noise component. It is possible to provide an image processing apparatus, an image processing method, an image processing program, and an integrated circuit capable of performing noise reduction processing with different strength for each pixel of interest.
Therefore, according to the present invention, it is possible to perform gradation conversion of the input signal without amplifying the noise component.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[First Embodiment]
The image processing apparatus according to the first embodiment of the present invention will be described with reference to FIGS.
FIG. 1 is a block diagram showing a configuration of an image processing apparatus 1000 according to the first embodiment of the present invention.
As shown in FIG. 1, the image processing apparatus 1000 includes a gradation conversion unit 100 that calculates a conversion signal TS from an input signal IS, and a noise reduction degree that determines a noise reduction degree NR based on the input signal IS and the conversion signal TS. A determination unit 200 and a noise reduction unit 300 that performs noise reduction processing on the converted signal TS based on the noise reduction degree NR to obtain an output signal OS are provided.
With this configuration, the noise reduction degree determination unit 200 determines the noise reduction degree to be applied to each pixel based on the degree to which the noise component is amplified in the gradation conversion unit 100, and the noise reduction unit 300 further reduces the noise reduction. Noise reduction processing can be performed based on the noise reduction degree NR calculated by the degree determining unit 200. As a result, it is possible to reproduce an image having preferable gradation characteristics without amplifying noise components when converting the gradation of the input signal.

Hereinafter, each functional unit in FIG. 1 will be described.
The gradation conversion unit 100 receives the input signal IS, performs gradation conversion for correcting the brightness and contrast on the input signal IS, generates a conversion signal TS, and converts the conversion signal TS into the noise reduction unit 300 and the noise. It outputs to the reduction degree determination unit 200. In the gradation conversion in the gradation conversion unit 100, for example, a histogram equalization method that performs gradation conversion of the input signal IS based on histogram information of the entire image can be applied. FIG. 2 is a block diagram showing a configuration of the gradation conversion unit 100 that performs gradation conversion by the histogram equalization method.
As shown in FIG. 2, first, the histogram calculation unit 101 calculates histogram information HG from the input signal IS, and outputs the histogram information HG to the input / output characteristic determination unit 102. Next, the input / output characteristic determination unit 102 receives the histogram information HG as input, and based on the histogram information HG calculated by the histogram calculation unit 101, adds many gradations to a gradation level (gradation value) having a high appearance frequency. The input / output characteristic HC is determined so as to be assigned. Then, the input / output characteristic determination unit 102 outputs data constituting the input / output characteristic HC to the gradation processing unit 103.

Here, the gradation processing unit 103 receives the data constituting the input / output characteristic HC as input, converts the input signal IS of the pixel of interest according to the input / output characteristic HC calculated by the input / output characteristic determination unit 102, and converts The signal TS is acquired. Then, the gradation processing unit 103 outputs the acquired converted signal TS to the noise reduction unit 300 and the noise reduction degree determination unit 200.
Note that the gradation conversion method is not limited to the method realized by the gradation conversion unit 100 illustrated in FIG. Further, the configuration of the gradation conversion unit 100 is not limited to the configuration shown in FIG.
<Modification (Spatial Visual Processing)>
Hereinafter, a modified example using a gradation conversion method different from the above-described gradation conversion method as a gradation conversion method will be described.
The gradation conversion method according to the modification shown here is a method based on visual processing that performs gradation conversion based on a target pixel of an input signal and a signal obtained by performing predetermined processing on a peripheral pixel of the target pixel. .

FIG. 7 is a block diagram showing a configuration of a gradation conversion unit 150 that performs gradation conversion by visual processing according to this modification. As shown in FIG. 7, the gradation conversion unit 150 includes a spatial processing unit 151 that extracts a peripheral image signal US including peripheral image information from the input signal IS, and a conversion in which the input signal IS is visually processed according to the peripheral image signal US. A visual processing unit 152 that calculates the signal TS.
Here, the spatial processing unit 151 performs filter processing on the target pixel and the peripheral pixels of the input signal IS. For example, the spatial processing unit 151 performs the following low-pass filter processing on the target pixel and the peripheral pixels of the input signal IS to calculate the peripheral image signal US.
US = (Σ [Wij] × [Aij]) ÷ (Σ [Wij])
Here, [Wij] is a weight coefficient of the pixel located in the i-th row and j-th column in the target pixel and the peripheral pixels, and [Aij] is a pixel weight coefficient in the i-th row and j-th column in the target pixel and the peripheral pixels. Value. Further, “Σ” means that the sum of the values for each of the target pixel and the surrounding pixels is calculated (calculation of the sum of several sequences).

Note that a smaller weight coefficient may be given as the absolute value of the pixel value difference is larger, or a smaller weight coefficient may be given to a pixel having a larger distance from the target pixel.
Next, the visual processing unit 152 receives the input signal IS and the peripheral image signal US output from the spatial processing unit as input, and performs tone conversion of the input signal IS according to the peripheral image signal US, thereby converting the converted signal TS. And outputs a conversion signal TS. The visual processing unit 152 can perform gradation conversion based on, for example, the two-dimensional gradation conversion characteristics shown in FIG. Here, the two-dimensional gradation conversion refers to gradation conversion in which output values are determined for two inputs of the peripheral image signal US and the input signal IS.
FIG. 8 shows a graph of two-dimensional gradation conversion characteristics. In FIG. 8, the horizontal axis represents the pixel value (gradation value) of the input signal IS, and the vertical axis represents the pixel value (gradation value) of the conversion signal TS. In FIG. 8, the input signal and the conversion signal TS are 8-bit signals, and the range of pixel values (gradation values) is 0 to 255.

  As shown in FIG. 8, the two-dimensional gradation conversion is performed according to the signal level (gradation value) of the peripheral image signal US, US0 to USn (n is an integer associated with the signal level (gradation value)). It has a predetermined gradation conversion characteristic. That is, in the two-dimensional gradation conversion, any one of the gradation conversion curves US0 to USn is selected according to the signal level (gradation value) of the peripheral image signal US, and the selected gradation conversion curve is used. This is realized by converting the input signal IS (IS gradation value) into a conversion signal TS (TS gradation value). For example, when the level (gradation value) of the US signal is “1”, the gradation conversion curve US1 of FIG. 8 is selected, and when the level (gradation value) of the US signal is “120”, the gradation Conversion curve US120 is selected. However, it is not always necessary to prepare the gradation conversion curves US0 to USn in a number corresponding to the number of gradation values of the US signal. For example, the gradation conversion curves US0 to USn are set to the number of gradation values of the US signal. Prepare a gradation conversion curve corresponding to the gradation value of the US signal by interpolation from the prepared gradation conversion curve for the gradation conversion curves that are less than the corresponding number. Thus, two-dimensional gradation conversion may be realized.

In the two-dimensional gradation conversion, for example, if the peripheral image signal US has a value of 8 bits, there are gradation conversion characteristics divided into 256 stages, which are expressed as gradation conversion curves each having a predetermined gamma conversion characteristic. be able to.
As shown in FIG. 8, the gradation conversion characteristic is expressed by a plurality of gradation conversion curves having a predetermined gamma conversion characteristic, and the output of the plurality of gradation conversion curves is monotonous with respect to the subscript of the peripheral image signal US. (When the gradation value of the input signal is the same value, the gradation value to be output becomes smaller as the subscript (number) of the peripheral image signal US becomes larger). The “relationship in which the output monotonously decreases” here is not limited to a relationship in which the output monotonously decreases. Even if there is a portion where the output is not partially monotonously reduced, the subscript of the peripheral image signal US Therefore, it may be monotonically decreasing.

In the two-dimensional gradation conversion characteristics shown in FIG. 8, for all the gradation values of the pixels of the input signal IS, (output value when US = US0) ≧ (output value when US = US1) ≧ ... ≧ (output value when US = USn) is satisfied. This gradation conversion characteristic enhances the contrast of the local area.
According to the two-dimensional gradation conversion characteristics shown in FIG. 8, when the input signal IS has a value (gradation value) “a”, the visual processing unit 152 determines the level of the conversion signal TS according to the surrounding image signal US. The adjustment value can take a value from “P” to “Q”. That is, even if the input signal IS has the value (gradation value) “a”, the value (gradation value) of the conversion signal TS to be output according to the peripheral image signal US is changed from “P” to “Q”. Will vary greatly.
As described above, according to the gradation conversion unit 150 of the modification, it is possible to perform gradation conversion according to the peripheral image information for each pixel in the image.

Next, in the noise reduction degree determination unit 200, the input signal IS and the converted signal TS are input, and the noise reduction degree NR representing the strength of the noise reduction processing applied to the converted signal TS is input from the input signal IS and the converted signal TS. calculate. Then, the noise reduction degree determination unit 200 outputs the calculated noise reduction degree NR to the noise reduction unit 300. In general, the noise component is greatly amplified as the pixel has a smaller signal level (gradation value) of the input signal IS and is subjected to gradation conversion with a large gain value. Therefore, as the signal level (gradation value) of the input signal IS is smaller, the gain of gradation conversion, that is, the ratio of the conversion signal TS to the input signal IS (= (gradation value of the conversion signal TS) / The larger the (tone value of the input signal IS)), the stronger the noise reduction processing is preferably performed.
In the following description, the noise reduction degree NR has a value (real value) from 0 to 1, and the larger the value, the stronger the noise reduction processing will be described.

FIG. 3 shows a block diagram of the noise reduction degree determination unit 200 that performs processing for calculating the noise reduction degree NR.
As shown in FIG. 3, the noise reduction degree determination unit 200 mainly includes a first noise reduction degree calculation unit 201 that calculates a first noise reduction degree NRA from the input signal IS, a conversion signal TS, and an input signal IS. A division unit 202 that calculates the ratio DIV (= (TS gradation value) / (IS gradation value)), and a second noise reduction degree that calculates the second noise reduction degree NRB from the calculated ratio DIV value. The calculation unit 203 includes a minimum value output unit 204 as a noise reduction degree determination unit that outputs a smaller one of the first noise reduction degree NRA and the second noise reduction degree NRB.
Here, since the first noise reduction degree calculation unit 201 wants to perform noise reduction processing more strongly for pixels with a lower signal level (gradation value), a larger noise level with a lower signal level (gradation value) of the input signal IS. It is preferable that the reduction degree NRA is output. An example of such characteristics is shown in FIG.

On the other hand, since the second noise reduction degree calculation unit 203 wants to perform noise reduction processing more strongly for pixels that have undergone gradation conversion with a high gain, the larger the ratio DIV between the converted signal TS and the input signal IS, the larger the noise reduction degree. It is preferable that NRB is output. An example of such characteristics is shown in FIG.
Next, in the minimum value output unit 204, the first noise reduction degree NRA calculated by the first noise reduction degree calculation unit 201 and the second noise reduction degree NRB calculated by the second noise reduction degree calculation unit 203 are calculated. By outputting the smaller one, a larger noise reduction degree NR can be calculated for a pixel whose input signal IS has a smaller signal level (gradation value) and whose gradation is converted with a larger gain. . Then, the noise reduction degree determination unit 200 outputs the calculated noise reduction degree NR to the noise reduction unit 300.

  The processing in the minimum value output unit 204 is not limited to the above processing. For example, the average value (arithmetic average value and geometric average) of the first noise reduction degree NRA and the second noise reduction degree NRB is used. And the average value of the noise reduction degree NR or the weighted average value of the first noise reduction degree NRA and the second noise reduction degree NRB is obtained, and the weighted average value is used as the noise reduction degree. It may be a process of setting NR. Here, the weighted average value is a smaller value of the first noise reduction degree NRA and the second noise reduction degree NRB (this smaller value is referred to as “value A”, and the larger value is referred to as “value B”). "). First, a large weighting is performed on the value A, a small weighting is performed on the value B, and then an averaging process is performed. For example, the weighted average value can be obtained by (weighted average value) = ((value A) × 3 + (value B)) / 4.

Next, the noise reduction unit 300 performs noise reduction processing on the converted signal TS based on the noise reduction degree NR. FIG. 6 is a diagram illustrating an example of the configuration of the noise reduction unit 300 that adjusts the strength of the noise reduction processing based on the noise reduction degree NR.
As shown in FIG. 6, the noise reduction unit 300 mainly smoothes the converted signal TS to obtain a smoothed signal LPS, and smoothes the converted signal TS based on the noise reduction degree NR. And an internal division processing unit 302 that internally divides the signal LPS and calculates an output signal OS.
Here, the smoothing processing unit 301 can be realized by performing, for example, a low-pass filter process on a plurality of pixels around the target pixel.
Further, the internal division processing in the internal division processing unit 302 can be realized by the internal division processing according to the following expression using the noise reduction degree NR calculated by the noise reduction degree determination unit 200.

OS = LPS × NR + TS × (1−NR)
According to this equation, when the noise reduction degree NR is large, the smoothed signal LPS is output with a large weight, and a result (output) in which noise is strongly removed is obtained. On the other hand, when the noise reduction degree NR is small, the converted signal TS is output with a large weight, and a result (output) in which noise is removed weakly is obtained.
A process of applying a large gain to the input signal IS corresponding to the pixel of interest having a small gradation value (such a pixel of interest is displayed dark on the display screen) and the acquired output signal OS to the display device. When displayed, noise is easily noticeable on the display screen.
That is, the input signal IS corresponding to the pixel of interest having a small gradation value has a small signal level (gradation value) of the pixel of interest. The S / N ratio is worse than when it is large. Therefore, when the output signal OS obtained by amplifying the signal with the poor S / N ratio is displayed on the display device, the display is in a state with a poor S / N, and noise is easily noticeable.

Further, even if a process of blurring (for example, a process of applying an LPF) is performed on a target pixel having a small gradation value, deterioration on the display screen is difficult to be detected in terms of visual characteristics.
Therefore, it is preferable to perform strong noise reduction processing on a portion corresponding to a pixel having a small gradation value and a portion amplified by a large gain.
On the other hand, the input signal IS corresponding to the pixel of interest having a large gradation value has a large signal level (gradation value) of the pixel of interest. Therefore, if the noise components are the same, the signal level (gradation value) of the pixel of interest is the same. Compared to the small case, the S / N ratio is good. Therefore, even if the display screen is amplified with a large gain, noise is not noticeable on the display screen.
On the other hand, when a process of blurring (for example, a process of applying an LPF) is performed on a pixel of interest having a large gradation value, deterioration on the display screen is likely to be detected in terms of visual characteristics.

Therefore, it is preferable to perform weak noise reduction processing on a portion corresponding to a pixel having a large gradation value.
Based on the above concept, the characteristics shown in FIG. 4, the characteristics shown in FIG. 5, and the configuration of the noise reduction degree determination unit 200 are determined. Needless to say, the characteristics shown in FIG. 4, the characteristics shown in FIG. 5, and the configuration of the noise reduction degree determination unit 200 can be changed to other ones based on the above concept.
As described above, according to the image processing apparatus 1000 according to the present embodiment, from the signal level (gradation value) of the input signal IS and the gain value when the input signal IS is converted into a converted signal TS. Since the degree of amplification of the noise component of each pixel is known, by applying noise reduction processing with an appropriate strength according to the amplification degree of the noise component, the noise component is suppressed and preferable gradation characteristics are obtained. It is possible to obtain an output signal OS having the same. The image processing apparatus 1000 according to the present embodiment can perform preferable gradation conversion on the input signal IS without amplifying the noise component of the input signal IS, and can output it as the output signal OS. By displaying the output signal OS output from the image processing apparatus 1000 as an image (video) on a display device (not shown), it is possible to reproduce an image (video) having a preferable gradation.

The noise reduction degree NR may be calculated from only the ratio DIV between the converted signal TS and the input signal IS. This makes it possible to strongly apply noise reduction processing to all pixels processed with a high gain regardless of the value of the input signal IS.
Further, the noise reduction degree NR may be calculated based on a difference between the converted signal TS and the input signal IS. As a result, it is possible to strongly perform noise reduction processing on the pixels in which the conversion signal TS has greatly increased from the input signal IS.
Note that the adjustment of the strength of the noise reduction processing in the noise reduction unit 300 may be realized by changing the filter coefficient of the low-pass filter applied to the converted signal TS. Thereby, it is possible to adjust the degree of reduction of the high frequency component by the low-pass filter processing, and it is possible to adjust the degree of noise reduction.

[Second Embodiment]
Next, an image processing apparatus according to a second embodiment of the present invention will be described with reference to FIGS.
In the image processing apparatus 1000 according to the first embodiment of the present invention, the tone conversion unit 100 calculates the conversion signal TS from the input signal IS, and then the noise reduction degree determination unit 200 calculates noise from the input signal IS and the conversion signal TS. The reduction degree NR was calculated. In the image processing apparatus 2000 according to the second embodiment of the present invention, after calculating the gain GN for gradation conversion from the input signal IS, the input signal IS is multiplied by the gain GN to calculate the converted signal TS, An embodiment in which the noise reduction degree NR is calculated from the input signal IS and the gain GN will be described with reference to FIG.

FIG. 9 is a block diagram showing a configuration of an image processing apparatus 2000 according to the second embodiment of the present invention. Hereinafter, the same parts as those in the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.
In FIG. 9, an image processing apparatus 2000 according to the second embodiment of the present invention includes a gain calculation unit 110 that outputs a gain GN from an input signal IS, and a multiplication unit that calculates a conversion signal TS based on the input signal IS and the gain GN. 120, a noise reduction degree determination unit 210 that calculates a noise reduction degree NR based on the input signal IS and the gain GN, and a noise reduction process on the converted signal TS based on the noise reduction degree NR to obtain an output signal OS And a noise reduction unit 300.
Even when this configuration is used, the noise reduction degree determination unit 210 can specify a pixel whose noise component is greatly amplified when the input signal IS is converted into the conversion signal TS. Can appropriately perform noise reduction processing based on the noise reduction degree NR calculated by the noise reduction degree determination unit 210.

Thereby, in the image processing apparatus 2000 according to the present embodiment, when the tone conversion is performed by multiplying the input signal by the gain, the noise reduction process is appropriately performed even for the pixel in which the noise component is greatly amplified. Since the output signal OS processed and output by the image processing apparatus 2000 according to the present embodiment is displayed as an image (video) on a display device (not shown), the noise component in the image is amplified. It is possible to reproduce an image (video) having a preferable gradation without causing the image to appear.
Hereinafter, of the functional units shown in FIG. 9, portions different from the first embodiment will be described.
The gain calculation unit 110 calculates a gain GN that is a gain value for correcting the brightness and contrast of the input signal IS. By multiplying the input signal IS by the gain GN, the brightness and contrast of the input signal IS are corrected.

In order to realize the gain calculation unit 110, for example, a technique based on a histogram equalization method that performs gradation conversion of the input signal IS based on histogram information can be applied. FIG. 10 is a block diagram illustrating a configuration of the gain calculation unit 110 based on the histogram equalization method.
As shown in FIG. 10, the gain calculation unit 110 mainly includes a histogram calculation unit 101 that calculates histogram information HG from the input signal IS, and a tone conversion gain characteristic determination unit that determines the gain output characteristic GC from the histogram information HG. 112 and a gradation processing gain calculation unit 113 that calculates the gain GN based on the gain output characteristic GC and the input signal IS.
In FIG. 10, similarly to the histogram equalization method described in the first embodiment, first, the histogram calculation unit 101 calculates histogram information HG from the input signal IS.

Next, the gradation conversion gain characteristic determination unit 112 is set based on the histogram information HG so that a larger number of gradations are assigned to a gradation level (gradation value) having a higher appearance frequency in the input signal IS. The gain output characteristic GC is determined for each target pixel of the input signal IS, for each predetermined block (region) including a plurality of pixels including the target pixel, or for the entire image.
The gradation processing gain calculation unit 113, based on the gain output characteristic GC calculated by the gradation conversion gain characteristic determination unit 112, a gain GN for multiplying the input signal IS of the target pixel by the multiplication unit 120 and the noise reduction. Output to the degree determination unit 210.
The method for calculating the gain is not limited to the method realized by the gain calculation unit 110 shown in FIG. Further, the configuration of gain calculation section 110 is not limited to the configuration shown in FIG.

<Modification (Gain-type spatial visual processing)>
Hereinafter, a modified example using a gain calculation method different from the gain calculation method described above will be described as a method for calculating the gain used for gradation conversion (gain calculation method).
The method of gradation conversion (gradation conversion using gain) according to the modified example shown here is based on a signal obtained by performing predetermined processing on the target pixel of the input signal IS and peripheral pixels of the target pixel. This is a technique based on visual processing to be calculated.
FIG. 11 is a block diagram showing a configuration of the gain calculation unit 160 for gradation conversion by visual processing according to this modification. As shown in FIG. 11, the gain calculation unit 160 extracts a peripheral image signal US including peripheral image information from the input signal IS, and visually processes the input signal IS according to the peripheral image signal US. A visual processing gain calculation unit 162 that calculates the gain GN.

Since the spatial processing unit 151 is the same process as that of the first embodiment described with reference to FIG. 7, detailed description thereof is omitted here.
Next, the visual processing gain calculation unit 162 calculates a gain GN for multiplying the input signal IS in accordance with the input signal IS and the peripheral image signal US. Similar to the modification of the first embodiment, the processing in this case also determines the gain GN for gradation conversion of the input signal IS based on the two inputs of the peripheral image signal US and the input signal IS. Two-dimensional gradation conversion is performed. Here, in order to realize the same gradation conversion characteristic as the two-dimensional gradation conversion characteristic shown in FIG. 8, the output value of the two-dimensional gradation conversion characteristic is converted into the conversion signal TS and the input signal IS shown in FIG. (Ie, gain GN). This input / output characteristic is shown in FIG.
As shown in FIG. 12, in the process of outputting the gain GN, depending on the signal level (gradation value) of the peripheral image signal US, US0 to USn (n is an integer associated with the signal level (gradation value)). Output characteristics of a predetermined gain GN. That is, in the two-dimensional gain conversion, any one of the gain conversion curves US0 to USn is selected according to the signal level (gradation value) of the peripheral image signal US, and the input signal is determined by the selected gain conversion curve. This is realized by converting IS (IS gradation value) into gain GN. For example, when the level (tone value) of the US signal is “1”, the curve US1 in FIG. 12 is selected, and when the level (tone value) of the US signal is “120”, the curve US120 is selected. The However, the gain conversion curves US0 to USn do not necessarily have to be prepared in a number corresponding to the number of gradation values of the US signal. For example, the gain conversion curves US0 to USn correspond to the number of gradation values of the US signal. By preparing less than the number of gain conversion curves that are not prepared, and calculating the gradation conversion curve corresponding to the gradation value of the US signal by interpolation processing from the prepared gain conversion curve, You may make it implement | achieve two-dimensional gain conversion.

In the two-dimensional gain conversion, for example, if the peripheral image signal US is an 8-bit value, there are gain conversion characteristics divided into 256 stages, each of which has an output characteristic (gain conversion characteristic) of a predetermined gain GN. It can be expressed as a curve.
As shown in FIG. 12, the gain conversion characteristic associated with the gradation conversion characteristic includes a plurality of gain conversion curves having a gain conversion characteristic associated with a predetermined gamma conversion characteristic (converting gradation values into gain values). The plurality of gain conversion curves are in a relationship in which the output monotonously decreases with respect to the subscript of the peripheral image signal US (if the gradation value of the input signal is the same value, the peripheral image signal US The larger the subscript (number), the smaller the output gain value). The “relationship in which the output monotonously decreases” here is not limited to a relationship in which the output monotonously decreases. Even if there is a portion where the output is not partially monotonously reduced, the subscript of the peripheral image signal US Therefore, it may be monotonically decreasing.

In the gain conversion characteristic associated with the two-dimensional gradation conversion characteristic shown in FIG. 12, (output value when US = US0) ≧ (US = The output value in the case of US1) ≧... ≧ (output value in the case of US = USn) is satisfied. The contrast of the local area is enhanced by the gain conversion characteristic associated with the gradation conversion characteristic.
In other words, the two-dimensional gradation conversion described in the first embodiment is realized by multiplying the input signal IS by the gain GN determined by the gain conversion characteristic shown in FIG. Therefore, by using the gain conversion characteristic shown in FIG. 12, it is possible to realize a process for enhancing the contrast of the local region as in the case described in the first embodiment.
According to the conversion characteristics shown in FIG. 12, when the input signal IS has a value (gradation value) “a”, the value of the gain GN varies from “R” to “S” according to the surrounding image signal US. Can take a value. That is, even if the input signal IS has the value (gradation value) “a”, the output gain GN (gain value) varies greatly from “R” to “S” according to the peripheral image signal US. .

Next, the multiplication unit 120 obtains the converted signal TS by multiplying the input signal IS by the obtained gain GN.
Next, the noise reduction degree determination unit 210 receives the input signal IS and the gain GN, and calculates the noise reduction degree NR to be applied to the converted signal TS from the input signal IS and the gain GN. Then, the noise reduction degree determination unit 210 outputs the calculated noise reduction degree NR to the noise reduction unit 300. In general, the noise component is greatly amplified as the pixel has a smaller signal level (gradation value) of the input signal IS and is subjected to gradation conversion with a large gain GN. For this reason, it is preferable that the noise reduction process be performed more strongly as the signal level (gradation value) of the input signal IS is smaller and as the gain GN of gradation conversion is larger.
FIG. 13 is a block diagram of the noise reduction degree determination unit 210 that performs processing for calculating the noise reduction degree NR.

As illustrated in FIG. 13, the noise reduction degree determination unit 210 mainly includes a first noise reduction degree calculation unit 201 that calculates the first noise reduction degree NRA from the input signal IS, and a second noise reduction degree NRC from the gain GN. And a minimum value output unit 204 as a noise reduction degree determination unit that outputs a smaller one of the first noise reduction degree NRA and the second noise reduction degree NRC. Composed.
As described in the first embodiment, the first noise reduction degree calculation unit 201 preferably outputs a larger noise reduction degree NRA as the input signal IS is smaller, as shown in FIG. On the other hand, it is preferable that the second noise reduction degree calculation unit 213 outputs a larger second noise reduction degree NRC as the gain GN increases. An example of such input / output characteristics is shown in FIG.
Next, in the minimum value output unit 204, the first noise reduction degree NRA calculated by the first noise reduction degree calculation unit 201 and the second noise reduction degree NRC calculated by the second noise reduction degree calculation unit 213 are calculated. By outputting the smaller one, a pixel having a smaller signal level (gradation value) and a pixel processed with a larger gain GN calculates a larger value as the noise reduction degree NR. Then, the noise reduction degree determination unit 210 outputs the calculated noise reduction degree NR to the noise reduction unit 300. The processing in the minimum value output unit 204 is not limited to the above processing. For example, the average value (arithmetic average value and geometric average) of the first noise reduction degree NRA and the second noise reduction degree NRC is used. And the average value of the noise reduction degree NR or the weighted average value of the first noise reduction degree NRA and the second noise reduction degree NRC is obtained, and the weighted average value is used as the noise reduction degree. It may be a process of setting NR. Here, the weighted average value is a smaller value of the first noise reduction degree NRA and the second noise reduction degree NRC (this smaller value is referred to as “value A”, and the larger value is referred to as “value B”). "). First, a large weighting is performed on the value A, a small weighting is performed on the value B, and then an averaging process is performed. For example, the weighted average value can be obtained by (weighted average value) = ((value A) × 3 + (value B)) / 4.

Next, the noise reduction unit 300 performs noise reduction processing on the converted signal TS based on the noise reduction degree NR. Since this process is the same as that of the noise reduction unit 300 in the first embodiment, a detailed description thereof is omitted.
As described above, according to the image processing apparatus 2000 according to the present embodiment, the signal level (gradation value) of the input signal IS and the gradation conversion (gain-type gradation conversion) by converting the input signal IS into the converted signal TS. ), The degree of amplification of the noise component of each pixel can be determined from the gain value, so that noise component can be suppressed by applying noise reduction processing with appropriate strength according to the amplification degree of the noise component. And an output signal OS having preferable gradation characteristics can be obtained. In the image processing apparatus 2000 according to the present embodiment, it is possible to perform a preferable gain type gradation conversion on the input signal IS without amplifying the noise component of the input signal IS, and to output it as the output signal OS. By displaying the output signal OS output from the image processing apparatus 2000 according to the above as an image (video) on a display device (not shown), it is possible to reproduce an image (video) having preferable gradation. .

Note that the noise reduction degree NR may be calculated only from the gain GN. This makes it possible to strongly apply noise reduction processing to all pixels processed with a high gain regardless of the value of the input signal IS.
Further, the adjustment of the strength of the noise reduction processing in the noise reduction unit 300 may be realized by changing the filter coefficient of the low-pass filter applied to the converted signal TS. Thereby, it is possible to adjust the degree of reduction of the high frequency component by the low-pass filter processing, and it is possible to adjust the degree of noise reduction.
[Other Embodiments (Other Modifications)]
Although the present invention has been described based on the above embodiment, it is needless to say that the present invention is not limited to the above embodiment. The following cases are also included in the present invention.

(1) Each of the above devices is specifically a computer system including a microprocessor, a ROM, a RAM, and the like. A computer program is stored in the RAM. Each device achieves its functions by the microprocessor operating according to the computer program. Here, the computer program is configured by combining a plurality of instruction codes indicating instructions for the computer in order to achieve a predetermined function.
(2) A part or all of the constituent elements constituting each of the above-described devices may be configured by one system LSI (Large Scale Integration). The system LSI is an ultra-multifunctional LSI manufactured by integrating a plurality of components on a single chip, and specifically, a computer system including a microprocessor, ROM, RAM, and the like. . A computer program is stored in the RAM. The system LSI achieves its functions by the microprocessor operating according to the computer program.

(3) Part or all of the constituent elements constituting each of the above devices may be configured from an IC card that can be attached to and detached from each device or a single module. The IC card or the module is a computer system including a microprocessor, a ROM, a RAM, and the like. The IC card or the module may include the super multifunctional LSI described above. The IC card or the module achieves its function by the microprocessor operating according to the computer program. This IC card or this module may have tamper resistance.
(4) The present invention may be the method described above. Further, the present invention may be a computer program that realizes these methods by a computer, or may be a digital signal composed of the computer program. The present invention also provides a computer-readable recording medium such as a flexible disk, hard disk, CD-ROM, MO, DVD, DVD-ROM, DVD-RAM, BD (Blu-ray Disc). ), Recorded in a semiconductor memory or the like. The digital signal may be recorded on these recording media.

In the present invention, the computer program or the digital signal may be transmitted via an electric communication line, a wireless or wired communication line, a network represented by the Internet, a data broadcast, or the like.
The present invention may be a computer system including a microprocessor and a memory, wherein the memory stores the computer program, and the microprocessor operates according to the computer program.
In addition, the program or the digital signal is recorded on the recording medium and transferred, or the program or the digital signal is transferred via the network or the like and executed by another independent computer system. You may do that.

(5) The above embodiment and the above modifications may be combined.
Further, in each process in the above-described embodiment, a part or all of each process may be performed in units of pixels, or a part or all of each process may be performed in units of blocks composed of a plurality of pixels. May be performed.
Moreover, each process of the said embodiment may be implement | achieved by hardware, and may be implement | achieved by software. Further, it may be realized by mixed processing of software and hardware.
The specific configuration of the present invention is not limited to the above-described embodiment, and various changes and modifications can be made without departing from the scope of the invention.

  The image processing apparatus, the image processing method, the image processing program, and the integrated circuit according to the present invention can reproduce an image having a preferable gradation without amplifying a noise component in the image. The image processing apparatus, the image processing method, the image processing program, and the integrated circuit according to the present invention can be implemented in this field.

1 is a block diagram of an image processing apparatus according to a first embodiment of the present invention. Block diagram of a gradation conversion unit in the same embodiment Block diagram of the noise reduction degree determination unit in the same embodiment The figure which shows the relationship between the input signal and the 1st noise reduction degree in the 1st noise reduction degree calculation part in the embodiment. The figure which shows the relationship between the ratio of the converted signal and input signal in the 2nd noise reduction degree calculation part in the same embodiment, and a 2nd noise reduction degree. Block diagram of the noise reduction unit in the same embodiment Block diagram of a modification of the gradation conversion unit in the embodiment The figure which shows the relationship between the input signal and conversion signal of visual processing in the embodiment The block diagram of the image processing apparatus in 2nd Embodiment of this invention. Block diagram of the gain calculation unit in the same embodiment The block diagram of the modification of the gain calculation part in the embodiment The figure which shows the relationship between the input signal and gain of visual processing in the embodiment Block diagram of the noise reduction degree determination unit in the same embodiment The figure which shows the relationship between the gain and 2nd noise reduction degree in the embodiment

Explanation of symbols

1000, 2000 Image processing apparatus 100, 150 Gradation conversion unit 101 Histogram calculation unit 102 Input / output characteristic determination unit 103 Gradation processing unit 110, 160 Gain calculation unit 112 Gradation conversion gain characteristic determination unit 113 Gradation processing gain calculation unit 120 Multiplication unit 151 Spatial processing unit 152 Visual processing unit 162 Visual processing gain calculation unit 200, 210 Noise reduction degree determination unit 201 First noise reduction degree calculation unit 202 Division units 203, 213 Second noise reduction degree calculation unit 204 Minimum value output unit 300 Noise reduction unit 301 Smoothing processing unit 302 Internal processing unit

Claims (20)

A gradation conversion unit that performs gradation conversion on an input signal composed of pixel data forming an image to obtain a converted signal;
Based on both the signal level of the input signal and the gradation conversion gain which is the ratio of the conversion signal to the input signal (= (the conversion signal) / (the input signal)), the noise reduction degree for the conversion signal A noise reduction degree determination unit for determining
A noise reduction unit that performs noise reduction processing on the converted signal based on the noise reduction degree ,
The noise reduction degree determination unit
When the gradation conversion gain is larger than the first gain threshold and the signal level of the input signal is smaller than the first signal level threshold, the noise reduction degree is set to the first noise reduction degree. And
When the gradation conversion gain is smaller than a second gain threshold that is smaller than the first gain threshold, and when the signal level of the input signal is smaller than the first signal level threshold, Setting the noise reduction degree to a second noise reduction degree that is smaller than the first noise reduction degree;
When the signal level of the input signal is larger than a second signal level threshold value that is larger than the first signal level threshold value, the noise reduction degree is a value smaller than the first noise reduction degree. Set the noise reduction level to
Image processing device. The gradation conversion unit performs gradation conversion based on histogram information about a gradation value of the pixel data of the input signal;
The image processing apparatus according to claim 1. The gradation conversion unit performs gradation conversion based on a signal obtained by performing predetermined processing on the target pixel data of the input signal and the peripheral pixel data of the target pixel data;
The image processing apparatus according to claim 1. A gradation conversion unit that performs gradation conversion on an input signal composed of pixel data forming an image to obtain a converted signal;
Based on both the signal level of the input signal and the gradation conversion gain which is the ratio of the conversion signal to the input signal (= (the conversion signal) / (the input signal)), the noise reduction degree for the conversion signal A noise reduction degree determination unit for determining
A noise reduction unit that performs noise reduction processing on the converted signal based on the noise reduction degree,
The noise reduction degree determination unit
As the gradation conversion gain increases, the gain reference noise reduction degree is set to a larger value,
As the signal level of the input signal increases, the signal level reference noise reduction degree is set to a small value,
Determining the noise reduction degree based on the gain reference noise reduction degree and the signal level reference noise reduction degree ;
Image processing device. The noise reduction degree determination unit sets a smaller value of the gain reference noise reduction degree and the signal level reference noise reduction degree to the noise reduction degree;
The image processing apparatus according to claim 4 . A gradation conversion unit that performs gradation conversion on an input signal composed of pixel data forming an image to obtain a converted signal;
Based on both the signal level of the input signal and the gradation conversion gain which is the ratio of the conversion signal to the input signal (= (the conversion signal) / (the input signal)), the noise reduction degree for the conversion signal A noise reduction degree determination unit for determining
A noise reduction unit that performs noise reduction processing on the converted signal based on the noise reduction degree,
The noise reduction degree determination unit
When the signal level of the input signal is greater than a predetermined signal level threshold value, a signal level reference noise reduction degree having a smaller value than when the signal level of the input signal is equal to or less than the predetermined signal level threshold value is output. 1 noise reduction degree calculation unit,
When the gradation conversion gain, which is the ratio of the converted signal to the input signal (= (the converted signal) / (the input signal)) is greater than a predetermined gain threshold, the gradation conversion gain is used for the predetermined gain. A second noise reduction degree calculation unit that outputs a gain reference noise reduction degree with a larger value than when the threshold value is equal to or lower than a threshold value;
A noise reduction degree output unit that sets a value calculated based on the signal level reference noise reduction degree and the gain reference noise reduction degree to the noise reduction degree;
The a, the image processing apparatus. The noise reduction degree output unit sets a smaller value of the signal level reference noise reduction degree and the gain reference noise reduction degree to the noise reduction degree.
The image processing apparatus according to claim 6 . A gain calculation unit for calculating a gain for converting an input signal composed of pixel data forming an image;
A multiplier for multiplying the input signal by the gain to obtain a converted signal;
A noise reduction degree determining unit that determines a noise reduction degree for the converted signal based on both the signal level of the input signal and the gain;
A noise reduction unit that performs noise reduction processing on the converted signal based on the noise reduction degree ,
The noise reduction degree determination unit
When the gain of the converted signal is greater than the first gain threshold and the signal level of the input signal is less than the first signal level threshold, the noise reduction degree is set to the first noise reduction degree. Set,
When the gain is smaller than a second gain threshold that is smaller than the first gain threshold, and when the signal level of the input signal is smaller than the first signal level threshold, the noise reduction is performed. Setting the degree to a second noise reduction degree that is smaller than the first noise reduction degree;
When the signal level of the input signal is larger than a second signal level threshold value that is larger than the first signal level threshold value, the noise reduction degree is a value smaller than the first noise reduction degree. An image processing apparatus that sets the noise reduction degree of The noise reduction degree determination unit further determines a noise reduction degree for the converted signal based on the input signal.
The image processing apparatus according to claim 8 . The gain calculation unit calculates the gain based on histogram information about a gradation value of the pixel data of the input signal;
The image processing apparatus according to claim 8 . A gain calculation unit for calculating a gain for converting an input signal composed of pixel data forming an image;
A multiplier for multiplying the input signal by the gain to obtain a converted signal;
A noise reduction degree determining unit that determines a noise reduction degree for the converted signal based on both the signal level of the input signal and the gain;
A noise reduction unit that performs noise reduction processing on the converted signal based on the noise reduction degree,
The noise reduction degree determination unit
As the gain increases, the gain reference noise reduction degree is set to a larger value,
As the signal level of the input signal increases, the signal level reference noise reduction degree is set to a small value,
Determining the noise reduction degree on the basis of the gain reference noise reduction degree and the signal level reference noise reduction degree, the image processing apparatus. The noise reduction degree determination unit sets a smaller value of the gain reference noise reduction degree and the signal level reference noise reduction degree to the noise reduction degree;
The image processing apparatus according to claim 1 1. A gain calculation unit for calculating a gain for converting an input signal composed of pixel data forming an image;
A multiplier for multiplying the input signal by the gain to obtain a converted signal;
A noise reduction degree determining unit that determines a noise reduction degree for the converted signal based on both the signal level of the input signal and the gain;
A noise reduction unit that performs noise reduction processing on the converted signal based on the noise reduction degree,
The noise reduction degree determination unit
When the signal level of the input signal is greater than a predetermined signal level threshold value, a signal level reference noise reduction degree having a smaller value than when the signal level of the input signal is equal to or less than the predetermined signal level threshold value is output. 1 noise reduction degree calculation unit,
A second noise reduction degree calculation unit that outputs a gain reference noise reduction degree with a larger value than when the gain is less than or equal to the predetermined gain threshold when the gain is larger than the predetermined gain threshold;
A noise reduction degree output unit that sets a value calculated based on the signal level reference noise reduction degree and the gain reference noise reduction degree to the noise reduction degree;
The a, the image processing apparatus. The noise reduction degree output unit sets a smaller value of the signal level reference noise reduction degree and the gain reference noise reduction degree to the noise reduction degree.
The image processing apparatus according to claim 1 3. A gradation conversion step of performing gradation conversion on an input signal composed of pixel data forming an image to obtain a converted signal;
Noise reduction for the converted signal based on both the signal level of the input signal and the gradation conversion gain which is the ratio of the converted signal to the input signal (= (the converted signal) / (the input signal)) In determining the degree ,
When the gradation conversion gain is larger than the first gain threshold and the signal level of the input signal is smaller than the first signal level threshold, the noise reduction degree is set to the first noise reduction degree. And
When the gradation conversion gain is smaller than a second gain threshold that is smaller than the first gain threshold, and when the signal level of the input signal is smaller than the first signal level threshold, Setting the noise reduction degree to a second noise reduction degree that is smaller than the first noise reduction degree;
When the signal level of the input signal is larger than a second signal level threshold value that is larger than the first signal level threshold value, the noise reduction degree is a value smaller than the first noise reduction degree. A noise reduction degree determination step for setting the noise reduction degree of
A noise reduction step of applying a noise reduction process to the converted signal based on the noise reduction degree;
An image processing method comprising: A gain calculating step for calculating a gain for converting an input signal composed of pixel data forming an image;
A multiplication step of multiplying the input signal by the gain to obtain a converted signal;
In determining the noise reduction degree for the converted signal based on both the signal level of the input signal and the gain ,
When the gain of the converted signal is greater than the first gain threshold and the signal level of the input signal is less than the first signal level threshold, the noise reduction degree is set to the first noise reduction degree. Set,
When the gain is smaller than a second gain threshold that is smaller than the first gain threshold, and when the signal level of the input signal is smaller than the first signal level threshold, the noise reduction is performed. Setting the degree to a second noise reduction degree that is smaller than the first noise reduction degree;
When the signal level of the input signal is larger than a second signal level threshold value that is larger than the first signal level threshold value, the noise reduction degree is a value smaller than the first noise reduction degree. A noise reduction degree determination step for setting the noise reduction degree of
A noise reduction step of applying a noise reduction process to the converted signal based on the noise reduction degree;
An image processing method comprising: On the computer,
A gradation conversion step of performing gradation conversion on an input signal composed of pixel data forming an image to obtain a converted signal;
Noise reduction for the converted signal based on both the signal level of the input signal and the gradation conversion gain which is the ratio of the converted signal to the input signal (= (the converted signal) / (the input signal)) In determining the degree ,
When the gradation conversion gain is larger than the first gain threshold and the signal level of the input signal is smaller than the first signal level threshold, the noise reduction degree is set to the first noise reduction degree. And
When the gradation conversion gain is smaller than a second gain threshold that is smaller than the first gain threshold, and when the signal level of the input signal is smaller than the first signal level threshold, Setting the noise reduction degree to a second noise reduction degree that is smaller than the first noise reduction degree;
When the signal level of the input signal is larger than a second signal level threshold value that is larger than the first signal level threshold value, the noise reduction degree is a value smaller than the first noise reduction degree. A noise reduction degree determination step for setting the noise reduction degree of
A noise reduction step of applying a noise reduction process to the converted signal based on the noise reduction degree;
An image processing program for executing A gain calculating step for calculating a gain for converting an input signal composed of pixel data forming an image in a computer;
A multiplication step of multiplying the input signal by the gain to obtain a converted signal;
When determining the noise reduction degree to be applied to the converted signal based on both the signal level of the input signal and the gain ,
When the gain of the converted signal is greater than the first gain threshold and the signal level of the input signal is less than the first signal level threshold, the noise reduction degree is set to the first noise reduction degree. Set,
When the gain is smaller than a second gain threshold that is smaller than the first gain threshold, and when the signal level of the input signal is smaller than the first signal level threshold, the noise reduction is performed. Setting the degree to a second noise reduction degree that is smaller than the first noise reduction degree;
When the signal level of the input signal is larger than a second signal level threshold value that is larger than the first signal level threshold value, the noise reduction degree is a value smaller than the first noise reduction degree. A noise reduction degree determination step for setting the noise reduction degree of
A noise reduction step for reducing noise of the converted signal based on the noise reduction degree;
An image processing program for executing A gradation conversion unit that performs gradation conversion on an input signal composed of pixel data forming an image to obtain a converted signal;
Noise reduction for the converted signal based on both the signal level of the input signal and the gradation conversion gain which is the ratio of the converted signal to the input signal (= (the converted signal) / (the input signal)) In determining the degree ,
When the gradation conversion gain is larger than the first gain threshold and the signal level of the input signal is smaller than the first signal level threshold, the noise reduction degree is set to the first noise reduction degree. And
When the gradation conversion gain is smaller than a second gain threshold that is smaller than the first gain threshold, and when the signal level of the input signal is smaller than the first signal level threshold, Setting the noise reduction degree to a second noise reduction degree that is smaller than the first noise reduction degree;
When the signal level of the input signal is larger than a second signal level threshold value that is larger than the first signal level threshold value, the noise reduction degree is a value smaller than the first noise reduction degree. A noise reduction degree setting step to set the noise reduction degree of
A noise reduction unit that performs noise reduction processing on the converted signal based on the noise reduction degree;
An integrated circuit comprising: A gain calculation unit for calculating a gain for converting an input signal composed of pixel data forming an image;
A multiplier for multiplying the input signal by the gain to obtain a converted signal;
In determining the noise reduction degree for the converted signal based on both the signal level of the input signal and the gain ,
When the gain of the converted signal is greater than the first gain threshold and the signal level of the input signal is less than the first signal level threshold, the noise reduction degree is set to the first noise reduction degree. Set,
When the gain is smaller than a second gain threshold that is smaller than the first gain threshold, and when the signal level of the input signal is smaller than the first signal level threshold, the noise reduction is performed. Setting the degree to a second noise reduction degree that is smaller than the first noise reduction degree;
When the signal level of the input signal is larger than a second signal level threshold value that is larger than the first signal level threshold value, the noise reduction degree is a value smaller than the first noise reduction degree. A noise reduction degree determination unit that sets the noise reduction degree of
A noise reduction unit that performs noise reduction processing on the converted signal based on the noise reduction degree;
An integrated circuit comprising:

Images