JP5863236B2 - Image processing apparatus and image processing method - Google Patents

Description

  The present invention relates to a noise reduction technique.

  Usually, a captured image captured by an imaging device such as a digital camera includes noise. There are random noise and fixed pattern noise. Random noise is noise that appears regardless of the position on the image. Fixed pattern noise is noise that appears at a fixed position on an image.

  Since noise is an image component that is not directly related to the subject, there is a problem that if the subject image contains noise, the quality of the captured image is degraded. For this reason, in a digital camera or the like, noise reduction processing is performed by a signal processing circuit or the like, thereby suppressing deterioration in the image quality of a captured image due to noise.

  Conventionally, various noise reduction processing methods have been proposed, but they are mainly based on processing by a smoothing filter. Usually, since the subject image has a weaker high-frequency component than the low-frequency component, noise is dominant in the high-frequency region on the captured image. Therefore, if a smoothing filter that suppresses high frequency components is applied to a captured image, noise can be reduced while preserving low to medium frequency components of the subject image. A smoothing filter based on this idea is now widely used as a noise reduction technique.

  Further, as an improvement of the smoothing filter, an edge preserving smoothing filter called a bilateral filter has been proposed (Non-Patent Document 1). However, in the conventional noise reduction technique based on the smoothing filter as described above, the high-frequency component of the subject image is also suppressed, and a fine texture component may be removed. Even if a filter having edge preserving characteristics is used, the effect cannot be obtained unless the texture amplitude is large to some extent.

  In response to the above problem, in the noise reduction method described in Patent Document 1, it is determined whether or not the image region is a flat region. If it is determined that the image region is flat, the target pixel is set to the average value of the region. Perform the process of approaching. The outline of the determination of the flat region in Patent Document 1 is roughly as follows. The region determination method described in Patent Document 1 is performed based on the standard deviation of the region. Since the standard deviation is large in the edge region with high contrast, noise reduction is not performed when the standard deviation is large. Therefore, noise reduction processing is not performed in the edge region, and the edge is stored.

JP 2007-150441 A

“Fast Bilateral Filtering for the Display of High-Dynamic-Range IMAGES Proc. Of ACM SIGGRAPH 2002, pp. 257-266

  However, in the technique described in Patent Document 1, since an edge with low contrast has a small standard deviation, noise reduction processing is performed on this edge, and there is a problem that an edge with low contrast is blurred.

  The present invention has been made in view of the above problems, and an object thereof is to provide a technique for reducing noise while preserving edges with low contrast.

In order to achieve the object of the present invention, for example, an image processing apparatus of the present invention comprises the following arrangement. That is, an image processing apparatus that performs noise reduction processing on an input image,
Acquisition means for acquiring a pixel value of a target pixel in the input image and a pixel value of each peripheral pixel located around the target pixel;
A frequency calculating means for calculating a frequency for each pixel value according to the number of pixels for each pixel value in the region including the target pixel;
Based on the frequency for each pixel value obtained by the frequency calculation means, the frequency of the pixel value of the pixel of interest determined by the frequency of the pixel value of the pixel of interest and the frequency of neighboring pixel values of the pixel value of the pixel of interest An inclination calculating means for calculating a value correlated with the inclination;
Using the correction amount based on a value having a correlation with the inclination, have a correction means for correcting the pixel value of the target pixel,
When the number of pixels having a neighboring pixel value larger than the pixel value of the target pixel is larger than the number of pixels having a neighboring pixel value smaller than the pixel value of the target pixel, the correction amount When the pixel value is determined to be increased and the number of pixels having a neighboring pixel value smaller than the pixel value of the target pixel is larger than the number of pixels having a neighboring pixel value larger than the pixel value of the target pixel The pixel value of the pixel of interest is determined to be a value that decreases .

  According to the configuration of the present invention, it is possible to reduce noise while preserving edges with low contrast.

FIG. 3 is a block diagram illustrating an example of a functional configuration of the image processing apparatus. FIG. 3 is a block diagram illustrating a configuration example of a signal processing unit 103. FIG. 3 is a block diagram illustrating a configuration example of a noise reduction processing unit 201. The flowchart of the process which the noise reduction process part 201 performs. The figure which shows a histogram. The figure which shows a histogram. The figure which shows a histogram.

  Preferred embodiments of the present invention will be described below with reference to the accompanying drawings. The embodiment described below shows an example when the present invention is specifically implemented, and is one of the specific examples of the configurations described in the claims.

[First Embodiment]
First, a configuration example of the image processing apparatus according to the present embodiment will be described with reference to the block diagram of FIG. In the present embodiment, an image processing apparatus applied to a digital camera will be described.

  The imaging unit 101 acquires external light as a video signal corresponding to the amount of light. The imaging unit 101 includes, for example, a zoom lens, a focus lens, a shake correction lens, a diaphragm, a shutter, an optical low-pass filter, an iR cut filter, a color filter, a sensor such as a CMOS or CCD, and the like. The A / D converter 102 converts the video signal (analog signal) sent from the imaging unit 101 into a digital signal (data).

  The signal processing unit 103 performs signal processing such as white balance processing, gamma correction processing, and noise reduction processing on the input image represented by the data obtained by the A / D conversion unit 102. The encoder unit 105 performs compression encoding processing such as Jpeg or Mpeg on the input image that has been subjected to signal processing by the signal processing unit 103, and generates encoded image data. Of course, there are various compression encoding methods, and the compression encoding method is not limited to a specific method. In the first place, this compression encoding is not essential.

  A PC (personal computer) and other media (for example, hard disk, memory card, CF card, SD card, USB memory) can be connected to the media I / F 106. However, the encoded image data (or uncompressed encoded image data) by the encoder unit 105 can be recorded on a PC or other media via the media I / F 106.

  The D / A converter 104 converts the input image (digital signal) that has been subjected to signal processing by the signal processor 103 into an analog signal. The display unit 113 is configured by a liquid crystal screen or the like, and displays an input image represented by this analog signal. The display unit 113 can also display characters, graphics, and the like generated by the character generation unit 112.

  The CPU 107 controls the operation of the entire image processing apparatus using computer programs and data stored in the ROM 108 and RAM 109. The ROM 108 stores computer programs and data for causing the CPU 107 to perform operation control of each unit constituting the image processing apparatus. Various kinds of information described later as known information (preliminary information) are also stored in the ROM 108.

  The RAM 109 has an area for temporarily storing various data, and a work area used when the CPU 107 executes various processes. That is, the RAM 109 can provide various areas as appropriate.

  The imaging system control unit 110 controls the imaging unit 101 described above, and performs various controls instructed by the CPU 107 such as focusing, opening a shutter, and adjusting an aperture.

  The operation unit 111 includes a user interface such as a button or a mode dial, and the user can input various instructions to the CPU 107 by operating the operation unit 111. When the display unit 113 is configured with a touch panel type liquid crystal screen, the user touches the touch panel type liquid crystal screen with his / her finger or the like, as is well known, so that various instructions can be given to the CPU 107. Input can be made.

  FIG. 1 shows only a part of the configuration including the configuration requirements used in the following description. However, the configuration applicable to the image processing apparatus according to the present embodiment is not limited to the configuration illustrated in FIG. 1, and may be a configuration in which appropriate configuration requirements are added to the configuration illustrated in FIG. However, if the following processing can be realized, an appropriate configuration requirement may be omitted. Further, the processing described later as one configuration requirement may be performed by another configuration requirement, or may be distributed by each.

  Next, a more detailed configuration of the signal processing unit 103 will be described with reference to the block diagram of FIG. The signal processing unit 103 is for improving the image quality of the input image sent from the A / D conversion unit 102. First, the noise reduction processing unit 201 performs noise reduction processing on the input image. The input image subjected to the noise reduction processing by the noise reduction processing unit 201 is subjected to white balance processing by the white balance control unit, edge enhancement processing by the edge enhancement unit, color conversion processing by the color conversion unit, and γ by the gamma processing unit. Correction processing is performed in order. The gamma processing unit outputs an input image on which the above processes have been performed. In order to improve the image quality, only the processing by the noise reduction processing unit 201 may be performed, or one or more of the processing by the noise reduction processing unit 201 and the white balance control unit, edge enhancement unit, color conversion unit, and gamma processing unit. You may implement | achieve by the process by.

  Next, a more detailed configuration of the noise reduction processing unit 201 will be described with reference to the block diagram of FIG. 3 and the flowchart of FIG. First, in step S <b> 401, the frequency calculation unit 301 acquires the input image sent from the A / D conversion unit 102.

  In step S <b> 402, the frequency calculation unit 301 first acquires pixel values of each of the pixel of interest in the input image and peripheral pixels located around the pixel of interest. For example, the pixel value of each of the pixel of interest and 8 pixels located around the pixel of interest is acquired.

  Then, the frequency calculation unit 301 obtains the number of pixels having the same pixel value as the pixel value x of the target pixel from the pixel values of the peripheral pixels located around the target pixel. Further, the frequency calculation unit 301 calculates the pixel value having the pixel value closest to the pixel value x of the target pixel from the pixel values of the peripheral pixels located around the target pixel among pixel values smaller than the pixel value x of the target pixel. The number H1 is obtained. Furthermore, the frequency calculation unit 301 calculates a pixel having a pixel value closest to the pixel value x of the target pixel from pixel values of peripheral pixels located around the target pixel, out of pixel values x of the target pixel. The number H2 is obtained.

  For example, assume that the pixel value of the pixel of interest is P3, and the pixel values of the eight pixels located around the pixel of interest are P2, P4, P2, P1, P5, P4, P3, and P2. Here, it is assumed that P1 <P2 <P3 <P4 <P5. At this time, the number of pixels having the pixel value P1 is 1, the number of pixels having the pixel value P2 is 3, the number of pixels having the pixel value P3 is 2, the number of pixels having the pixel value P4 is 2, and the pixel value P5 The number of pixels having 1 is 1. In this case, however, in step S402, the frequency calculation unit 301 obtains “1” as the number of peripheral pixels having the same pixel value as the pixel value x (= P3) of the target pixel. Furthermore, “3” is obtained as the number of peripheral pixels having a pixel value (= P2) closest to the pixel value x of the target pixel among pixel values smaller than the pixel value x of the target pixel. Further, the number “2” of peripheral pixels having a pixel value (= P4) closest to the pixel value x of the target pixel among pixel values larger than the pixel value x of the target pixel is obtained.

  Note that the number of pixels for each pixel value may be calculated using peripheral pixels located around the pixel of interest in the input image, but not all pixels may be used and is close to the pixel value of the pixel of interest. It is only necessary to obtain the number of peripheral pixels having a pixel value. For example, when the pixel value of the target pixel is “100”, the number of pixels within the range of the pixel value “80” to “120” may be obtained.

  In step S <b> 403, the inclination calculation unit 302 calculates an increase / decrease (= inclination a) of the number of pixels having the same pixel value in the vicinity region including the target pixel by calculating the following expression. However, other equations may be used as long as this increase / decrease can be obtained.

  Here, H (x) indicates the number of pixels having the pixel value x among the peripheral pixels located around the target pixel when the pixel value of the target pixel is x. H (x−1) corresponds to the above H1, and H (x + 1) corresponds to the above H2.

  In step S404, the correction amount adjustment unit 303 acquires a noise characteristic coefficient stored in the ROM 108 or the like. In the present embodiment, C1 and C2 are acquired as the noise characteristic coefficients. The coefficients C1 and C2 satisfy the following formula.

In this equation, I may be the pixel value of the pixel of interest, or may be the average pixel value of the pixel value of the pixel of interest and the pixel values of peripheral pixels located around the pixel of interest. The noise characteristic coefficient indicates the magnitude of noise included in the captured image. Specifically, noise variance calculated from an image obtained by photographing a chart. It is known that the amount of noise of shot noise increases according to the luminance. Therefore, it is preferable that the relationship between the variance σ ^{2 of the} noise included in the photographed image and the pixel value I is expressed as the above equation (2) using the coefficients C1 and C2 and the coefficients C1 and C2 are the noise characteristic coefficients. It is.

  However, C1 is a term mainly related to shot noise, and C2 is a term mainly related to dark current noise. C1 and C2 are obtained from the image of the gray scale chart. Specifically, C1 and C2 are obtained by performing regression analysis on the relationship between the average value and the variance value of the region of the captured image corresponding to the region having a constant reflectance in the gray scale chart using the equation (2). In addition to the method of obtaining from a captured image, the variance of noise that will be included in the captured image may be calculated from design parameters.

In any case, since the shot noise amount depends on the luminance, it is desirable to calculate the noise amount σ ² according to the pixel value of the pixel of interest or the surrounding pixels. Note that when the pixel value of the target pixel is large, the amount of noise σ ² that will be included in the target pixel increases according to Equation (2). Since the correction amount described below is proportional to the noise amount σ ^2, it can be seen that the correction amount of the pixel of interest increases. Note that the case where the slope a is positive means that the number of pixels having a larger value than the pixel value of the target pixel is larger than the number of pixels having a smaller value than the target pixel. . In that case, it can be seen from Equation (1) that the correction has the effect of increasing the pixel value of the pixel of interest. The reverse is also true.

In step S405, the correction amount adjustment unit 303 obtains C1 and C2 obtained in step S404, the pixel value of the pixel of interest (or the pixel value of the pixel of interest and the average pixel value of the pixel values of surrounding pixels located around the pixel of interest) I. , The above equation (2) is calculated to obtain σ ² . Then the correction amount adjusting unit 303, the sigma ² × a by using the slope a calculated by obtained sigma ² and the above equation (1), calculated as a correction amount, and sends to the adder 304.

In step S406, the adder 304 uses the result (x + σ ² × a) obtained by adding the correction amount (σ ² × a) to the pixel value x of the pixel of interest as a new pixel value of the pixel of interest, and the white balance control unit Output for.

  Since the flowchart of FIG. 4 is a flowchart of the process for one pixel (= target pixel), the noise reduction processing unit 201 performs the process according to the flowchart of FIG. 4 for each pixel constituting the input image. Become. An image obtained by performing the processing according to the flowchart of FIG. 4 for each pixel constituting the input image is an image after noise reduction.

  Note that there may be no surrounding pixels of the pixel of interest near the boundary of the image. In that case, it can be considered that a pixel having a predetermined pixel value virtually exists as a peripheral pixel, or the shape of the region is changed so that the pixel exists in the region.

  Next, it will be described that an edge having a low contrast cannot be produced in an image obtained by performing processing according to the flowchart of FIG. 4 for each pixel constituting the input image.

  First, the reason why noise can be satisfactorily reduced by the above processing according to the present embodiment will be described. The pixel value of the pixel of interest in the input image is x, the number of pixels having the pixel value x among the peripheral pixels is H (x), the result of differentiation of H (x) with respect to x is H ′ (x), and mixed in the input image Let σ be the standard deviation of the noise. At this time, the pixel value x of the target pixel is updated to the pixel value x ′ by the noise reduction processing unit 201 according to the following equation (3).

The reason why noise reduction is performed satisfactorily will be described using this equation (3). If the image before noise mixing has a uniform pixel value x ₀ and noise of standard deviation σ is mixed into this image, the histogram of the image after mixing is a normal distribution of the following equation (4). Can be represented.

From the above equation (3) and (4), the pixel value x after the noise reduction 'is as shown in the following equation (5), the pixel value x ₀ before noise contamination pixel value x after the noise reduction' and I understand that

  However, it should be noted that the normal distribution represented by Equation (4) is a smooth function. If the peripheral area of the pixel of interest is too small, the histogram H (x) is calculated from a small number of samples, and a smooth ideal histogram such as the equation (4) cannot be obtained, and the equation (3) is calculated. There is no guarantee that the result will be (Expression (5)).

  FIG. 5 shows a histogram in which random numbers according to the normal distribution are generated by a computer. In FIG. 5, the solid line is a histogram, and the wavy line is a normal distribution (ideal histogram). Due to the finite number of samples, the histogram is not smooth and does not match the normal distribution. In order to avoid such a situation, it is necessary to devise a method for obtaining a good result even if the image area for taking the histogram is widened or the number of samples is small. For example, Equation (1) is a general equation for obtaining an amount indicating a slope from discrete data. However, since the histogram is not smooth when the number of samples is small, a may take a value larger than the ideal value. . Therefore, as shown in the following equation (6), a device for correcting using the number of samples N can be considered.

  According to Equation (6), when the number of samples is small, the denominator is large, and the amount a indicating the inclination is small. That is, the formula (6) has an effect of preventing the amount “a” indicating the slope from becoming excessively large, and a good result can be obtained. Note that the number of samples N in Equation (6) is the sum of the frequencies of x-1, x, and x + 1.

  Next, consider a case where the image before noise mixing is not uniform. Let h (α) be the histogram of the image before noise mixing. Then, the histogram of the image after mixing noise can be expressed as in the following formula (7) using h (α).

  Equation (7) provides a new view of the histogram of the captured image. The content of the sum symbol is equal to the number of pixels whose pixel value is α before noise is mixed and which has changed to the pixel value x due to noise. Looking at Expression (7) as a whole, the frequency of the pixel value x of the input image is equal to the sum of the pixel values α before noise mixing that changes to the pixel value x due to noise mixing. Substituting equation (7) into equation (3) yields the following equation (8).

  Consider the meaning of equation (9), which is part of equation (8).

  The amount represented by Equation (9) is obtained by dividing the number of pixels whose pixel value has changed from α to x due to noise mixing by the number of pixels x after noise mixing. That is, the amount represented by Expression (9) is an amount indicating what percentage of the pixels having the pixel value x of the captured image that were the pixel value α before the noise was mixed. This is equal to the posterior probability P (α | x) of the pixel value α before the noise is mixed after obtaining the pixel value x with noise. If the amount represented by equation (9) is P (α | x), equation (8) can be modified as shown in equation (10) below.

  When formula (10) is expanded, the following formula (11) is obtained.

  Here, since ΣP (α | x) = 1 (the last term on the right side of Expression (11)), Expression (11) can be finally transformed into Expression (12) below.

  Expression (12) is conditional on the acquisition of the pixel value x, and the expected value of the pixel value before noise mixing is calculated using the probability that the pixel value is α before noise mixing. In other words, obtaining the pixel value after noise reduction using Equation (3) substantially determines the expected value of the pixel value before noise mixing. The above is the reason why the noise can be reduced by the equation (3). It should be noted that the expression (3) can calculate the expected value of the pixel value before noise mixing by the expected value calculation without directly obtaining the histogram before noise mixing.

  Next, the reason why low-contrast edges can be preserved by the above processing according to the present embodiment will be described. The operation of the noise reduction method by the above processing according to the present embodiment is shown in FIG. FIG. 6 shows the relationship between the histogram of the captured image and the correction direction of the pixel value.

  In FIG. 6, the pixel value having a value smaller than the average value has a positive gradient, and correction is performed so as to approach the peak of the histogram. On the other hand, the pixel value having a value larger than the average value has a negative gradient, and as a correction result, the pixel value of the pixel of interest decreases and is corrected so as to approach the peak of the histogram. Since the histogram spreads due to noise mixing, it can be seen that the noise reduction method by the above processing according to the present embodiment has an effect of bringing the spread pixel value closer to the state before noise mixing.

  FIG. 7A shows the relationship between the histogram and the correction direction of the pixel value when the input image includes a low contrast edge. The pixel values a and b in FIG. 7A are average values of the pixel values constituting the edge. In the state before mixing noise, the pixel values a and b have non-zero frequencies as in the histogram shown in FIG. When viewing the correction direction of the pixel values in FIG. 7A, the pixel values of the input image other than the peaks a and b are corrected so as to approach the peaks a and b. A pixel having a pixel value equal to a and b which is a pixel value before noise is mixed is not corrected because the inclination is zero. Therefore, as an effect of the correction, there is an action of suppressing a decrease in edge contrast and bringing it closer to a state before noise mixing.

  The above is true even if the pixel values a and b located at the peak of the histogram take close values. Therefore, in the noise reduction method by the above processing according to the present embodiment, noise is reduced while maintaining contrast even at an edge with low contrast.

[Second Embodiment]
In the first embodiment, the case where the noise reduction processing unit 201 is mounted on a digital camera has been described. However, the operation of the noise reduction processing unit 201 described in the first embodiment is various images that require noise reduction processing. It can be incorporated into a processing device.

  The noise reduction processing unit 201 may be configured by hardware, but may be implemented by software and stored in a ROM or the like. In this case, in the device having the ROM and the CPU, when the CPU executes the software stored in the ROM, each processing described above is performed by the noise reduction processing unit 201 in the first embodiment. Will be executed.

(Other examples)
The present invention can also be realized by executing the following processing. That is, software (program) that realizes the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and a computer (or CPU, MPU, or the like) of the system or apparatus reads the program. It is a process to be executed.

Claims (7)

An image processing apparatus that performs noise reduction processing on an input image,
Acquisition means for acquiring a pixel value of a target pixel in the input image and a pixel value of each peripheral pixel located around the target pixel;
A frequency calculating means for calculating a frequency for each pixel value according to the number of pixels for each pixel value in the region including the target pixel;
Based on the frequency for each pixel value obtained by the frequency calculation means, the frequency of the pixel value of the pixel of interest determined by the frequency of the pixel value of the pixel of interest and the frequency of neighboring pixel values of the pixel value of the pixel of interest An inclination calculating means for calculating a value correlated with the inclination;
Using the correction amount based on a value having a correlation with the inclination, have a correction means for correcting the pixel value of the target pixel,
When the number of pixels having a neighboring pixel value larger than the pixel value of the target pixel is larger than the number of pixels having a neighboring pixel value smaller than the pixel value of the target pixel, the correction amount When the pixel value is determined to be increased and the number of pixels having a neighboring pixel value smaller than the pixel value of the target pixel is larger than the number of pixels having a neighboring pixel value larger than the pixel value of the target pixel The image processing apparatus is determined to have a value that decreases the pixel value of the pixel of interest .   The image processing apparatus according to claim 1, wherein the correction amount is a correction amount corresponding to a noise characteristic in the input image. An image processing apparatus that performs noise reduction processing on an input image,
Acquisition means for acquiring a pixel value of a target pixel in the input image and a pixel value of each peripheral pixel located around the target pixel;
Correction means for correcting the pixel value of the pixel of interest so as to approach the pixel value that is the largest number of pixels among the pixel values of the respective surrounding pixels,
The correction means includes
Frequency calculation means for calculating the number of pixels having the pixel value among the respective peripheral pixels, using the pixel values of the respective peripheral pixels;
Determining means for determining a correction amount for the pixel value of the target pixel according to the result calculated by the frequency calculating means;
Updating means for updating the pixel value of the pixel of interest using the correction amount;
The frequency calculation means includes
Let H be the number of pixels having the same pixel value as the pixel value x of the pixel of interest,
From the pixel values of the respective peripheral pixels, the number H1 of pixels having the pixel value closest to the pixel value x among the pixel values smaller than the pixel value x is calculated, and from the pixel values of the respective peripheral pixels, Calculating the number H2 of pixels having a pixel value closest to the pixel value x among the pixel values greater than the pixel value x;
The determining means includes
When a = (H2−H1) / (2 × H) is calculated and I is the pixel value x or the average pixel value of the pixel of interest and each of the surrounding pixels, it is set in advance. An image processing apparatus, wherein σ2 = C1 × I + C2 is calculated using the obtained coefficients C1 and C2, and a × σ2 is determined as a correction amount. When the pixel value of the pixel of interest is x and the number of pixels having the pixel value x calculated by the frequency calculation means is H (x),
The inclination calculation means, an image processing apparatus according to claim 1 or 2, characterized in that to calculate the result of said H (x) is differentiated with respect to x as a value correlated to the inclination. An image processing method performed by an image processing apparatus that performs noise reduction processing on an input image,
An acquisition step in which the acquisition unit of the image processing apparatus acquires a pixel value of a pixel of interest in the input image and a pixel value of each peripheral pixel located around the pixel of interest;
A frequency calculation step in which the frequency calculation means of the image processing device calculates a frequency for each pixel value according to the number of pixels for each pixel value in the region including the target pixel;
The inclination calculation means of the image processing device is determined by the frequency of the pixel value of the pixel of interest and the frequency of neighboring pixel values of the pixel value of the pixel of interest based on the frequency for each pixel value obtained by the frequency calculation step. An inclination calculating step of calculating a value correlated with the inclination of the frequency of the pixel value of the target pixel;
Correction means of the image processing apparatus, by using the correction amount based on a value having a correlation with the inclination, have a correction step of correcting the pixel value of the target pixel,
When the number of pixels having a neighboring pixel value larger than the pixel value of the target pixel is larger than the number of pixels having a neighboring pixel value smaller than the pixel value of the target pixel, the correction amount When the pixel value is determined to be increased and the number of pixels having a neighboring pixel value smaller than the pixel value of the target pixel is larger than the number of pixels having a neighboring pixel value larger than the pixel value of the target pixel The image processing method is characterized in that it is determined to be a value that decreases the pixel value of the pixel of interest . An image processing method performed by an image processing apparatus that performs noise reduction processing on an input image,
An acquisition step in which the acquisition unit of the image processing apparatus acquires a pixel value of a pixel of interest in the input image and a pixel value of each peripheral pixel located around the pixel of interest;
A correction step of correcting the pixel value of the pixel of interest so that the correction means of the image processing apparatus approaches the pixel value that is the largest number of pixels among the pixel values of the respective peripheral pixels,
The correction step includes
A frequency calculation step of calculating the number of pixels having the pixel value among the respective peripheral pixels, using the pixel values of the respective peripheral pixels;
A determination step of determining a correction amount for the pixel value of the target pixel according to the result calculated in the frequency calculation step;
Updating the pixel value of the pixel of interest using the correction amount, and
In the frequency calculation step,
Let H be the number of pixels having the same pixel value as the pixel value x of the pixel of interest,
From the pixel values of the respective peripheral pixels, the number H1 of pixels having the pixel value closest to the pixel value x among the pixel values smaller than the pixel value x is calculated, and from the pixel values of the respective peripheral pixels, Calculating the number H2 of pixels having a pixel value closest to the pixel value x among the pixel values greater than the pixel value x;
In the determination step,
When a = (H2−H1) / (2 × H) is calculated and I is the pixel value x or the average pixel value of the pixel of interest and each of the surrounding pixels, it is set in advance. An image processing method, wherein σ2 = C1 × I + C2 is calculated using the obtained coefficients C1 and C2, and a × σ2 is determined as a correction amount. The computer program for functioning a computer as each means of the image processing apparatus of any one of Claims 1 thru | or 4 .

Images