KR100888428B1 - Device and Method of Removing Noise of Image - Google Patents

Abstract

A method for removing noise in an image is disclosed. Image input unit for receiving the original image from the outside; In the original image divided into n regions in a first direction, a difference between the average value of pixel values of each line in the first direction included in the region and the average value between the lines is filtered to determine whether random noise is present. A random noise removing unit for calculating a corrected pixel value correcting the pixel value of a generation portion of random noise; A controller which reads the pixel value of the original image and outputs the pixel value to the random noise removing unit, and generates a corrected image from which the random noise is removed using the corrected pixel value; And an image output unit configured to output the corrected image. According to an exemplary embodiment of the present invention, noise generated in an image may be removed by a simple operation while minimizing loss of an original image.

Image, random noise, pattern noise.

Description

Device and Method of Removing Noise of Image

1 is a diagram illustrating a method of dividing an original image into n areas according to an embodiment of the present invention.

2 is a block diagram of a noise removing device according to an embodiment of the present invention.

3 is a flowchart of a noise removing method according to an embodiment of the present invention;

4 is a block diagram illustrating a random noise removing unit in accordance with an embodiment of the present invention.

5 is a diagram illustrating a process of calculating an average value of pixel values for each line of an original image according to an embodiment of the present invention.

6 is a flowchart of a random noise removal method according to an embodiment of the present invention.

7 is a block diagram illustrating a pattern noise removing unit in accordance with an embodiment of the present invention.

8 is a flowchart illustrating a method for removing pattern noise according to an embodiment of the present invention.

9 is a diagram illustrating a process of calculating a coping value according to an embodiment of the present invention.

10 is a view showing an original image that is not removed noise according to an embodiment of the present invention

11 is a diagram illustrating a corrected image from which random noise is removed according to an embodiment of the present invention.

12 is a diagram illustrating a corrected image from which random noise and pattern noise are removed according to an embodiment of the present invention.

<Explanation of symbols for the main parts of the drawings>

200: noise reduction device

210: video input unit

220: random noise removal unit

230: pattern noise removing unit

240: control unit

250: video output unit

514: correction target line

The present invention relates to a method and apparatus for removing noise in an image, and more particularly, to a method and apparatus for removing random noise and pattern noise generated in a stripe form in an image.

Random noise refers to indeterminate noise existing in nature, including radio interference. As its name suggests, random noise occurs randomly, that is, unpredictably, and is difficult to avoid. When VHS video and the like record a video, the noise mixed in the video is also a kind of random noise.

In addition, pattern noise refers to noise that is fixedly generated or exists in an image. Therefore, the pattern noise may be known in advance at the position or the degree of the noise.

In the related art, a band-reject filter, a notch filter, an adaptive median filter, and the like are removed to eliminate streak noise of an image.

A band-reject filter is a filter having a waveform and an effect opposite to that of a band pass filter. A band-reject filter is a filter that cuts out a signal in a frequency band between two specific cutoff frequencies and passes the other frequencies without attenuation. . That is, as a filter for removing only signals between specific frequencies at specific frequencies, it is frequently used to block interference between specific frequencies.

Notch filters are filters that have a sharp attenuation at certain frequencies. The notch filter is used to remove unnecessary frequency components by using a high frequency signal output from a television transmitter by reflecting the tuning frequency component by the tuning reflection element by branching from the transmission line.

An adaptive median filter is a filter that replaces a value for an arbitrary point with a median value in a window by covering a window around an arbitrary point, and in particular, changes the size of the window according to noise. The median filter covers the neighbors to the desired point or space point and replaces the value of the point or point point in the window with the median of the values in the window. It can effectively remove noise and shows excellent performance in removing shock signal. In addition, by varying the window size for the detection of the median value depending on the error, it is possible to minimize the error data selection rate of the median filter.

However, when the band cut filter or the notch filter is used, the operation is performed in the frequency domain. Therefore, the Fourier transform is troublesome. In addition, when a band cut filter or a notch filter is applied to an image acquired from an actual sensor, it may sometimes be difficult to find an aspect of noise in the frequency domain depending on the characteristics of the noise.

On the other hand, in the adaptive median filter, the calculation is performed in the spatial domain, which is easier to operate than the previously described method, and reduces the loss of high frequency components much more than other conventional noise removing filters. However, although this method reduces the loss, the resultant image can be magnified to show the loss of the blurring and edge components so that the loss of the high-frequency component can be felt visually enough when compared to the original image. In addition, the effect is good when the noise level is not large and occurs randomly. However, the effect is not very effective for the noise of the fixed pattern.

Accordingly, the present invention provides a noise removing apparatus and method capable of removing noise by performing an easier operation without Fourier transform.

The present invention also provides stable processing results in both random noise removal and pattern noise removal.

The present invention also provides a method and apparatus for removing noise that can be set by adjusting a threshold for distinguishing noise from image information.

In addition, the present invention provides a method and apparatus for removing noise, which facilitates tuning to the size of noise component when improving image quality.

In addition, the present invention provides a method and apparatus for removing noise in which the blurring and the high frequency components are maintained in the resultant image by minimizing the blurring and the loss of the high frequency components (including edge and line boundary parts) when comparing the input image with the resultant image. to provide.

The present invention also provides a noise removal method and apparatus capable of removing only noise components while minimizing loss of an original image.

According to an embodiment of the present invention, an image input unit for receiving an original image; In the original image divided into n regions in a first direction, a difference between the average value of pixel values of each line in the first direction included in the region and the average value between the lines is filtered to determine whether random noise is present. A random noise removing unit for calculating a corrected pixel value correcting the pixel value of a generation portion of random noise; A controller configured to generate a corrected image from which the random noise is removed using the corrected pixel value; And an image output unit configured to output the corrected image.

According to another embodiment of the present invention, the random noise removing unit comprises: a random data input unit configured to receive the pixel value of the original image divided into the n areas; First filtering in the first and second directions is performed on the average value to calculate pixel error and correction value, and second filtering is performed on the pixel error to calculate a comparison value, thereby calculating the absolute value of the correction value. An arithmetic and discriminating unit comparing random values and the comparison value with a first threshold value and a second threshold value to detect random noise; And a first image corrector configured to output, to the controller, a correction pixel value obtained by adding the correction value to all pixel values of the correction target line when the correction target line is recognized as random noise. Is provided.

According to another embodiment of the present invention, the calculating and determining unit comprises: an average value calculating unit obtaining an average value of the pixel values included in the line for each region from the pixel values; And a predicted value for filtering the average value corresponding to the correction target line, which is any one of the lines, and the average value corresponding to an adjacent line adjacent to the correction target line in a second direction, to calculate a prediction value corresponding to the correction target line. There is provided a noise removing device comprising a calculation unit.

According to another embodiment of the present invention, the predictive value calculating unit includes one or more adjacent lines in the second direction with respect to the correction target line, the correction target line, and one or more in the second direction with respect to the correction target line. The noise removing device is provided by calculating the predicted value by averaging the average value of the average value except the smallest value and the largest value among the average values corresponding to the adjacent line.

According to another embodiment of the present invention, the calculation and determining unit obtains a pixel error that is a difference between the average value and the prediction value corresponding to a line on the same line in the first direction and the correction target line and the correction target line. And an error calculating unit for calculating a correction value by averaging the pixel error, wherein a line on the same line in the first direction is included in a region different from the region in which the correction target line is included. A noise canceling apparatus is provided.

According to another embodiment of the present invention, the calculation and determining unit compares the absolute value of the correction value received from the error calculating unit with the first threshold value, and when the absolute value of the correction value is smaller, the correction target line The noise removing device further comprises a random noise discrimination unit for recognizing the random noise.

According to another embodiment of the present invention, the calculation and determining unit obtains a pixel error that is a difference between the average value and the prediction value corresponding to a line on the same line in the first direction and the correction target line and the correction target line. A comparison value calculating unit for combining the pixel errors by two and calculating a comparison value that is a difference between the combined two pixel errors, wherein the comparison value is a positive number is provided. .

According to another embodiment of the present invention, the calculation and discriminating unit compares the comparison value received from the comparison value calculating unit with the second threshold value, and if the comparison value is smaller, recognizes the correction target line as random noise. There is provided a noise removing apparatus, further comprising a random noise discriminating unit.

According to another embodiment of the present invention, the error calculating unit is provided with a noise removing device, characterized in that to obtain a value obtained by subtracting the average value from the prediction value as the pixel error.

According to another embodiment of the present invention, if the correction value calculated by averaging the pixel error is not an integer, the error calculating unit calculates a value rounded to a decimal point value of the pixel error as the correction value. A noise canceling apparatus is provided.

According to another embodiment of the present invention, when the random noise removing unit removes the random noise generated horizontally, the first direction is a horizontal direction, the second direction is a vertical direction, and the random noise generated in the vertical direction is removed. When removing, the noise removing device is provided, wherein the random noise is removed by making the first direction the vertical direction and the second direction the horizontal direction.

According to another exemplary embodiment of the present invention, the pixel value of the point where the pattern noise occurs and the pixel value of a point adjacent to the pattern noise are filtered from the position information of the point where the pattern noise occurs in the original image. And a pattern noise remover configured to calculate a coping value for correcting the pixel value of the noise generating portion, wherein the controller reads the pixel value of the original image and outputs the pixel value to the random noise remover and the pattern noise remover. And outputting the positional information of the pattern noise to a pattern noise remover, and generating a corrected image from which the random noise and the pattern noise are removed using the corrected pixel value and the coping value. Is provided.

According to another embodiment of the present invention, the pattern noise removing unit comprises: a fixed data input unit configured to receive the position information from the controller, and to receive the pixel value of the original image from the image input unit; A pattern noise calculator configured to calculate a coping value by filtering the pixel values adjacent to the pattern noise in a second direction when the pattern noise occurs in the first direction; And a second image corrector configured to output the coping value to the controller instead of the pixel value at the point where the pattern noise occurs.

According to another embodiment of the present invention, the pattern noise calculation unit includes the pixel value at the point included in the pattern noise, the m number of pixel values adjacent to the pattern noise in the second direction, and the pattern noise in the -second direction. A noise canceller, wherein m is one or more natural numbers, is characterized by performing a median operation for calculating a median value among the m adjacent pixel values as the coping value.

According to another embodiment of the present invention, when the pattern noise occurs in the vertical direction, the first direction is a vertical direction, and the second direction is a horizontal direction, and when the pattern noise occurs in the horizontal direction, the first direction Is a horizontal direction, and the second direction is a vertical noise removal device is provided.

In addition, according to an embodiment of the present invention (a) receiving the original image; (b) In the original image divided into n regions in a first direction, the difference between the average value of pixel values per line in the first direction included in the region and the average value between the lines is filtered to determine whether random noise is present. Determining and calculating a corrected pixel value correcting the pixel value of the random noise generating portion; (c) generating a corrected image from which the random noise is removed using the corrected pixel value; And (d) outputting the corrected image.

According to another embodiment of the present invention, the step (b) may include: (b1) receiving the pixel value, the first threshold value, and the second threshold value of the original image divided into the n areas; (b2) performing a first filtering in the first and second directions on the average value to calculate a pixel error and a correction value, and performing a second filtering on the pixel error to calculate a comparison value, thereby correcting the correction Comparing the absolute value of the value and the comparison value with the first threshold value and the second threshold value to detect random noise; And (b3) outputting a correction pixel value obtained by adding the correction value to the pixel value of the correction target line when the correction target line is determined to be random noise.

According to another embodiment of the present invention, the step (b2) may include: (b21) obtaining an average value of the pixel values included in the line for each region from the pixel values; And (b22) a predicted value corresponding to the correction target line by filtering the average value corresponding to the correction target line which is any one of the lines and the average value corresponding to an adjacent line adjacent to the correction target line in a second direction. There is provided a noise removal method comprising the step of calculating the.

According to another embodiment of the present invention, the step (b22) is in the second direction with respect to the correction target line, at least one adjacent line in the second direction with respect to the correction target line and the correction target line. The noise removal method is provided by calculating the predicted value by averaging the average of the average values other than the smallest and largest values among the average values corresponding to one or more adjacent lines.

According to another embodiment of the present invention, the step (b2) may include (b23) a pixel error that is a difference between an average value corresponding to the line on the same line in the first direction and the prediction value and the correction target line and the correction target line. Obtaining a; And (b24) calculating a correction value by averaging the pixel error, wherein a line on the same line in the first direction is included in a region different from the region in which the correction target line is included. There is provided a method for removing noise, including.

According to another embodiment of the present invention, step (b2) (b25) compares the absolute value of the correction value with the first threshold value and randomizes the correction target line when the absolute value of the correction value is smaller. A noise removing method is provided, further comprising the step of recognizing it as noise.

According to another embodiment of the present invention, the step (b2) may include (b23) a pixel error that is a difference between an average value corresponding to the line on the same line in the first direction and the prediction value and the correction target line and the correction target line. Obtaining a; (b26) combining the pixel errors by two, and calculating a comparison value that is a difference between the two combined pixel errors, wherein the comparison value is a positive number. Is provided.

According to another embodiment of the present invention, the step (b2) (b27) compares the comparison value received from the comparison value calculator with the second threshold value, and if the comparison value is smaller, the correction target line; A noise removal method is provided, further comprising the step of recognizing it as random noise.

According to another embodiment of the present invention, the step (b23) is provided to the noise removal method, characterized in that the value obtained by subtracting the average value from the prediction value as the pixel error.

According to another embodiment of the present invention, the step (b23) is to calculate the value rounded to the decimal point value of the pixel error as the correction value when the correction value calculated by averaging the pixel error is not an integer. There is provided a noise removal method characterized in that.

According to another embodiment of the present invention, the step (b) is to remove the random noise generated horizontally, the first direction in the horizontal direction, the second direction in the vertical direction, random noise generated in the vertical direction In case of removing the noise, a method of removing noise is provided by removing the random noise using the first direction as the vertical direction and the second direction as the horizontal direction.

According to another embodiment of the present invention (e) by grasping the position information of the point where the pattern noise occurs in the original image, the pixel value of the point where the pattern noise occurs and the pixel value of the point adjacent to the pattern noise Calculating a coping value by correcting the pixel value of the pattern noise generating part by filtering, wherein step (c) removes the random noise and the pattern noise by using the corrected pixel value and the coping value. There is provided a method for removing noise, comprising generating a corrected image.

According to another embodiment of the present invention, the step (e) comprises: (e1) receiving the location information and receiving a pixel value of the original image; (e2) calculating a coping value by filtering pixel values adjacent to the pattern noise in a second direction when the pattern noise occurs in a first direction; and (e3) outputting the coping value instead of the pixel value at the point where the pattern noise occurs.

According to another exemplary embodiment of the present invention, the step (e2) may include pixel values of a point included in the pattern noise, m number of pixel values adjacent to the pattern noise in a second direction, and a second direction of the pattern noise. A noise removal method, wherein m is one or more natural numbers, is performed by performing a median operation of calculating a median value corresponding to an intermediate value among m adjacent pixel values.

According to another embodiment of the present invention, when the pattern noise occurs in the vertical direction, the first direction is a vertical direction, and the second direction is a horizontal direction, and when the pattern noise occurs in the horizontal direction, the first direction Is a horizontal direction and the second direction is a vertical direction.

As the invention allows for various changes and numerous embodiments, particular embodiments will be illustrated in the drawings and described in detail in the written description. However, this is not intended to limit the present invention to specific embodiments, it should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention.

The terms first, second, etc. may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be referred to as the second component, and similarly, the second component may also be referred to as the first component. The term and / or includes a combination of a plurality of related items or any item of a plurality of related items. The terminology used herein is for the purpose of describing particular example embodiments only and is not intended to be limiting of the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, the terms "comprising" or "stored" "mounted" "mounted" and the like are intended to indicate that there is a feature, number, step, action, component, part, or combination thereof described on the specification. It is to be understood that the present invention does not exclude in advance the possibility of the presence or the addition of one or more other features or numbers, steps, operations, components, parts, or a combination thereof. Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art. Terms such as those defined in the commonly used dictionaries should be construed as having meanings consistent with the meanings in the context of the related art, and shall not be construed in ideal or excessively formal meanings unless expressly defined in this application. Do not.

Other aspects, features, and advantages other than those described above will become apparent from the following drawings, claims, and detailed description of the invention.

Hereinafter, a preferred embodiment of the noise removing device and method according to the present invention will be described in detail with reference to the accompanying drawings, in the description with reference to the accompanying drawings, the same or corresponding components are given the same reference numerals and Duplicate explanations will be omitted.

Noise that can occur in an image can take many forms. However, in the present specification, description will be made mainly on the case where streak noise occurs. The streak noise can be largely classified into random noise and pattern noise. In the present specification, an apparatus and a method for removing noise generated in an image by dividing into a case of removing random noise and a case of removing pattern noise will be described.

According to an exemplary embodiment of the present invention, a process of obtaining an average of pixel values by dividing the entire original image into several regions may be included to remove noise. Here, the entire image is divided into n regions in the first direction or the second direction. Dividing an original image into n regions in a specific direction means that the original image divided by regions is divided so that n pieces are arranged in the specific direction.

1 is a diagram illustrating a method of dividing an original image into n areas according to an embodiment of the present invention.

Referring to FIG. 1A, an original image 110 divided into n regions 112, 114, and 116 in a first direction is illustrated. Here, the first direction means a horizontal direction, and n is three. As shown in FIG. 1A, when the random noise is removed by dividing the region of the original image, random noise generated in the horizontal direction may be removed. The noise removing apparatus calculates an average value of pixel values in units of lines included in each of the divided regions 112, 114, and 116. 118 shows an example of the line in the first direction.

Referring to FIG. 1B, an original image 120 divided into n regions 122, 124, and 126 in a second direction is illustrated. Here, the second direction means a vertical direction, and n is three. As shown in FIG. 1B, when the random noise is removed by dividing the region of the original image, random noise generated in the vertical direction may be removed. The noise removing apparatus calculates an average value of pixel values in units of lines included in each of the divided regions 122, 124, and 126. Where 128 is an example of a line in the second direction.

In each region constituting the original image, in the original image 110 divided into three regions in the horizontal direction, each region 112, 114, and 116 includes horizontal lines and 3 in the vertical direction. In the case of the original image 120 divided into two regions, each region 122, 124, and 126 includes vertical lines.

2 is a block diagram illustrating a noise removing apparatus 200 according to an embodiment of the present invention, and FIG. 3 is a flowchart illustrating a noise removing method according to an embodiment of the present invention. A function and an operation of the noise removing apparatus 200 will be described with reference to FIGS. 2 and 3.

The noise removing apparatus 200 according to an embodiment of the present invention includes an image input unit 210, a random noise remover 220, a pattern noise remover 230, a controller 240, and an image output unit 250. can do.

According to an exemplary embodiment, a noise removing device including only one of the random noise removing unit 220 and the pattern noise removing unit 230 may be implemented. However, in this embodiment, in order to describe the random noise removal and the pattern noise removal at a time, the noise removal apparatus including both the random noise removal unit 220 and the pattern noise removal unit 230 will be described.

Similarly, in the noise removal method, there may be a noise removal method that performs only one of random noise removal and pattern noise removal. However, for the sake of convenience of understanding and explanation, the present embodiment focuses on a noise removal method capable of removing both random noise and pattern noise in image correction.

The image input unit 210 receives an image to be displayed from the outside (S310). At this time, the input image corresponds to the original image. The image input unit 210 may be input together with an image and information about pixels of the input image. In the original image input to the image input unit 210, noise is removed by a method such as pixel value correction (S330, S360) by the random noise remover 220 and the pattern noise remover 230.

In order to remove the random noise, the random noise removing unit 220 examines whether the original image has random noise (S320), detects random noise, and calculates a corrected pixel value for correcting the pixel value of the portion where the random noise has occurred ( S330). In addition, when the pattern noise is present in the original image (S340), the pattern noise removing unit 230 first receives the position information of the pattern noise from the controller 240 or the image input unit 210 to identify the position information of the pattern noise (S350). )do. Then, a countermeasure value for correcting the pixel value of the pattern noise generating portion is calculated (S360).

Then, the controller 240 generates a correction image for output using the corrected pixel value including the corrected pixel value and / or the coping value (S370). The image output unit 250 outputs the corrected image generated by the controller 240 (S380).

As described above, the original image is composed of n regions, and the image corresponding to each region is composed of lines. Each line also includes an array of pixels with respective pixel values.

The controller 240 may read a pixel value from the original image input to the image input unit 210. The controller 240 may output pixel values of the original image read from the original image to the random noise remover 220 or the pattern noise remover 230. Alternatively, the image input unit 210 may output the pixels of the original image to the random noise remover 220 or the pattern noise remover 230 according to a control signal of the controller 240.

In addition, when a user of the noise removing apparatus 200 inputs the first threshold value and the second threshold value necessary to determine whether the random noise is present, the controller 240 receives and stores these threshold values, or the random noise removing unit 220. Can be printed as

The image output unit 250 outputs a corrected image that is an image from which noise is removed by the random noise remover 220 and the pattern noise remover 230 (S380). However, when there is no part determined as noise in the original image, the pixel value may not be corrected, and the image output through the image output unit 250 may be the same as the original image.

4 is a block diagram of a random noise removing unit 220 according to an embodiment of the present invention.

According to the exemplary embodiment illustrated in FIG. 4A, the random noise removing unit 220 may include a random data input unit 410, an operation and determination unit 415, and a first image correction unit 480. .

In addition, the operation and determination unit 415 may include an average value operation unit 430, a prediction value operation unit 440, an error operation unit 450, a comparison value operation unit 460, and a random noise determination unit 470. According to another exemplary embodiment, at least one of an average value calculator 430, a predicted value calculator 440, an error calculator 450, a comparison value calculator 460, and a random noise determiner 470 may be included in the calculation and determination unit 415. Combinations of these may be included. However, in the embodiment shown in (a) of Figure 4 will be described mainly in the case where all of these components are included.

Here, according to another exemplary embodiment shown in FIG. 4B, the average value calculator 430, the predicted value calculator 440, the error calculator 450, and the comparison value calculator 460 are random components of one component. 420.

However, in the present embodiment, in order to clearly describe each operation in the random noise removing unit 220 or the calculation and discriminating unit 415, different operations will be described based on the case where the different components are performed.

The random data input unit 410 receives data necessary for performing an operation in removing random noise. First, the pixel value of the original image may be input from the controller 240. In addition, the first threshold value and the second threshold value necessary for determining whether the correction target line corresponds to random noise may be input. The first threshold value and the second threshold value may be values previously stored or set in the noise removing apparatus 200. Or it may be a value received from the outside according to the noise level of the input original image. The first threshold value and the second threshold value will be described together in the description of the random noise discriminating unit 470.

According to another exemplary embodiment, the random data input unit 410 may be a component of the image input unit 210 due to being configured as a part of the image input unit 210. In this case, the first threshold value and the second threshold value are input by the image input unit 210, and the image input unit 210 determines the first threshold value and the second threshold value at the random noise detection step 470. Can be printed as

The calculation and determining unit 415 determines whether the correction target line corresponds to random noise (that is, detects random noise) and performs a calculation for calculating values necessary for the determination. The calculation and discriminating unit 415 performs the first filtering and the second filtering on the pixel value and the average value of the pixel values. The contents of the above filtering will be described in more detail below.

The average value calculator 430 may calculate an average value of pixel values of each line in line units. In the present specification, a description will be given of an embodiment for obtaining an average value by a method for calculating an arithmetic mean. However, in another embodiment, it may be possible to obtain pixel values per line by a different method.

The prediction value calculator 440 calculates the prediction value by filtering the average value of the correction target line and the average value of pixel values of adjacent lines in the second direction (vertical direction in this embodiment), that is, the correction target line. The predicted value may be a value as an estimate of the pixel value to be measured if the line to be corrected is not noise.

In addition, the error calculator 450 calculates a pixel error that is a difference between the predicted value and the average value. In particular, the present embodiment focuses on the operation of subtracting the average value from the predicted value. Therefore, when the average value is larger than the predicted value, the pixel error may be negative, and in the opposite case, the pixel error may be positive. The correction value is a value obtained by averaging pixel errors of the line to be corrected and pixel errors of the lines adjacent to the line to be corrected in the first direction, that is, the lines included in different areas and on the same line as the line to be corrected.

The average value calculator 440, the predicted value calculator 450, and the error calculator 460 may be one of operations performed by the first filtering. Therefore, the pixel error and the correction value can be obtained as a result of the first filtering.

The comparison value calculator 470 calculates a difference between pixel errors calculated as a result of the first filtering. Alternatively, the comparison value calculator 470 may calculate a pixel error and associate a difference between the pixel errors. A comparison value can be calculated by combining two pixel errors corresponding to the correction target line and the lines on the same line and calculating the difference between the two values. However, the comparison value is an absolute value of the difference between the pixel errors combined by two, and is not negative. That is, an operation of calculating a comparison value using the values (pixel errors) obtained as a result of the first filtering may be one of operations performed by the second filtering. That is, the comparison result of the second filtering may be calculated.

The random noise discriminating unit 470 compares the absolute value of the value (correction value) obtained as a result of the first filtering with the first threshold value, and compares the value (comparative value) obtained as a result of the second filtering with the second threshold value. It is determined whether the correction target line is random noise.

A component that performs the above process of determining whether a correction target line corresponds to random noise is referred to as an operation and detection unit. When the calculation of the values necessary for the detection of the random noise and the determination of the random noise are completed by the operation and detection unit, the first image correction unit 480 calculates the corrected pixel value correcting the pixel value of the random noise portion.

FIG. 5 is a diagram illustrating a process of calculating an average value of pixel values for each line of an original image according to an embodiment of the present invention, and FIG. 6 is a flowchart illustrating a random noise removing method according to an embodiment of the present invention. A process of removing random noise according to an embodiment of the present invention will be described with reference to FIGS. 5 and 6.

As shown in FIG. 5, the original image 500 is composed of pixels having pixel values corresponding to the original image 500. A ₁₁ , a ₁₂ ,... a _1x and b ₁₁ , b ₁₂ , c ₁₁ ,... , c ₃₁ ,. And the like are the pixel values of the pixels located therein. The arrangement of these points can constitute a line. That is, in the horizontal direction (first direction), the arrangement of the pixels from the pixel having the pixel value a ₁₁ to the pixel having the pixel value a _1x may constitute one line 510. Here, the pixel value of the original image is input to the random data input unit 410 (S610).

The random data input unit 410 may receive an input (S610) of the first threshold value and the second threshold value required when the random noise determination unit 470, which will be described later, detects or determines the random noise.

As described above, the original image 500 is divided into n regions, and FIG. 5 illustrates a case in which the original image is divided into three regions in the horizontal direction. The original image 500 according to the present exemplary embodiment is divided into a first region 540, a second region 542, and a third region 544. As illustrated, the first region 540 may include horizontal lines (eg, including 510, 512, 514, 516, 518,...).

Similarly, the second region 542 is a line (including 520, 524, etc.) in the horizontal direction (first direction) belonging to the region, and the third region 544 is a horizontal direction (first direction) belonging to the region. Of lines 530, 534, and the like.

In addition, in this specification, the line on the same line means the line located in the same position in a 2nd direction (corresponding to a vertical direction in FIG. 5) in each area | region. That is, for line 514, line 524 and line 534 correspond to the line on the same line.

In addition, an adjacent line means the line which adjoins parallel to a specific line among the lines in the same area | region. For line 514, 512 and 516 correspond to adjacent lines, and two adjacent lines indicate lines 510, 512, 516 and 518.

More specifically, for line 514, the adjacent line in the + second direction is 512, and the -line adjacent in the second direction is 516.

According to an exemplary embodiment of the present invention, as the n value increases, the accuracy or precision of random noise detection may increase.

The average value calculator 430 for each line may calculate an average value obtained by averaging pixel values of points belonging to the line (S620). If the average value of a pixel value is significantly higher or lower on only one line, that line is likely to contain noise.

According to the present embodiment, the average value may be calculated by Equation 1 below.

[Equation 1]

a ₁ = (a ₁₁ + a ₁₂ +… + a _1x ) / x

a ₂ = (a ₂₁ + a ₂₂ +… + a _2x ) / x

a ₃ = (a ₃₁ + a ₃₂ +… + a _3x ) / x

…

a _y = (a _y1 + a _y2 +… + a _yx ) / x

b ₁ = (b ₁₁ + b ₁₂ +… + b _1x ) / x

…

c ₁ = (c ₃₁ + c ₃₁ +… + c _3x ) / x

The average value of pixel values of the lines included in the first area 540 is included in the H_ave1 550, and the average value of the pixel values of the lines included in the second area 542 is included in the H_ave2 560 and the third area 544. The average value of the pixel values of the lines to be referred to as H_ave3 570. In the following description, a case in which random noise is detected and removed from the line 514 having an average value a ₃ among H_ave1 550, which is an average value of the first region 540, will be described. . Accordingly, the line 514 composed of the pixel values a ₃₁ to a _3x corresponds to the correction target line.

When the average value of the correction target line is calculated, the prediction value may be calculated using the average value of the adjacent lines (S630). The operation of calculating the predicted value may be performed by the predicted value calculator 440. The predicted value is the average value of the correction target line (514), (a ₃₎ and the average value of the adjacent line (510, 512, 516, 518, _{etc.) ((a 1, a 2} , a 4, a 5 , etc.) on the "alpha-stream This can be obtained by filtering by "alpha trimmed mean filter".

In calculating the predicted value by the alpha trimmed min filter, the case where the filter length is 5 and the alpha is 1 will be described as an embodiment. When the filter length is 5, the average value of the correction target line 514 and two lines 510 and 512 adjacent in the second direction (a ₁ , a ₂ ) and two lines adjacent in the second direction 516, The _five average values including the average values a ₄ and a ₅ of 518 and the average values a ₃ of the correction target line 514 are filtered.

In addition, if the alpha value is 1, the predicted value is calculated by performing an operation on three remaining values except for the smallest one and the largest one among a ₁ , a ₂ , a ₃ , a ₄ , and a ₅ .

If, according to another embodiment, the filter length is 7, the alpha value is 2, and the average value of the line to be corrected is a ₄ , the most of a ₁ , a ₂ , a ₃ , a ₄ , a ₅ , a ₆ , a ₇ The operation is performed with the smallest value, the second smallest value (or two smallest values), and the average of the three remaining values except the largest and second largest value (or two largest values).

In the present exemplary embodiment, the predicted value calculator 440 excludes one of the smallest one and the largest one among a ₁ , a ₂ , a ₃ , a ₄ , and a ₅ (that is, when alpha = 1). The difference between the arithmetic average value of the average value and the average value of the correction target line is calculated as a predicted value (S630).

This can be summarized as follows. Where pre_H1 is a predicted value.

[Equation 2]

min_val = min ([a ₁ , a ₂ , a ₃ , a ₄ , a ₅ ])

max_val = max ([a ₁ , a ₂ , a ₃ , a ₄ , a ₅ ])

pre_H1 = (a ₁ + a ₂ + a ₃ + a ₄ + a ₅ -min_val-max_val) / 3

min_val = min ([b ₁ , b ₂ , b ₃ , b ₄ , b ₅ ])

max_val = max ([b ₁ , b ₂ , b ₃ , b ₄ , b ₅ ])

pre_H2 = (b ₁ + b ₂ + b ₃ + b ₄ + b ₅ -min_val-max_val) / 3

min_val = min ([c ₁ , c ₂ , c ₃ , c ₄ , c ₅ ])

max_val = max ([c ₁ , c ₂ , c ₃ , c ₄ , c ₅ ])

pre_H3 = (c ₁ + c ₂ + c ₃ + c ₄ + c ₅ -min_val-max_val) / 3

However, there may be a case where there is no one or more adjacent lines in any one of the + second direction and the-second direction. In this case, the predicted value calculator 440 may calculate the predicted value by averaging the remaining average values except the smallest and largest values among the average values corresponding to two or more adjacent lines in the other direction.

In addition, according to the present exemplary embodiment, the pixel error corresponds to a difference between the predicted value calculated by Equation 2 and the average value of the correction target line, which may be calculated by subtracting the average value of the correction target line from the prediction value. have. Therefore, the pixel error becomes negative when the predicted value is smaller than the average value.

The pixel error is also calculated for the lines of all other areas on the same line as the correction target line. That is, the pixel error, which is obtained by subtracting the average value from the predicted values, can also be obtained for the lines 524 and 534. In Equation 3 below, diff_H1 is a pixel error.

[Equation 3]

diff_H1 = pre_H1-a ₃

diff_H2 = pre_H2-b ₃

diff_H3 = pre_H3-c ₃

The error calculator 450 may determine whether the line to be corrected is random noise by using the calculated pixel error, and calculate a correction value and a comparison value for correcting the pixel value of the noise part in the case of random noise. However, the comparison value may be calculated by the comparison value calculator 460.

First, the correction value is the average of pixel errors. The calculation of the correction value diff_all may be summarized by Equation 4 below.

[Equation 4]

diff_all = round ((diff_H1 + diff_H2 + diff_H3) / 3)

In Equation 4, round means rounding. Since the pixel value cannot be below the decimal point, if ((diff_H1 + diff_H2 + diff_H3) / 3) is not an integer, it can be changed to an integer value by rounding.

In addition, the comparison value refers to a difference between pixel errors in each region. That is, the two pixel errors (diff_H1, diff_H2, diff_H3) corresponding to the lines 514, 524, and 534 of the respective areas 540, 542, and 544 on the same line as the correction target line are combined, and the two The comparison value can be calculated by obtaining the difference between the values. Here, the comparison value is an absolute value of the difference between the pixel errors combined by two, and is not negative.

When the comparison value operation unit 460 expresses an operation for calculating the comparison value, it is as follows. In Equation 5 below, abs_diff1, abs_diff2 and abs_diff3 are comparison values, and "abs" means taking an absolute value. Therefore, the comparison value according to this embodiment is

[Equation 5]

abs_diff1 = abs (abs (diff_H1)-abs (diff_H2))

abs_diff2 = abs (abs (diff_H1)-abs (diff_H3))

abs_diff3 = abs (abs (diff_H2)-abs (diff_H2))

Can be obtained as

In the present embodiment, a method of determining whether the correction target line 514 is random noise is as follows. The random noise discriminating unit 470 compares the absolute value abs (diff_all) of diff_all, which is the correction value, with the first threshold value (S670). Also, the random noise discriminating unit 470 compares the comparison values abs_diff1, abs_diff2 and abs_diff3 with the second threshold value.

Here, the first threshold value and the second threshold value are variables that a user of the noise removing apparatus 200 can arbitrarily input. When the noise of the original image is severe, the first threshold value and the second threshold value may be input differently according to the noise level, such as inputting the first threshold value and the second threshold value in large numbers.

As described above, the first threshold value and the second threshold value may be input by the controller 240 from the outside, or may be preset and stored. The random noise discrimination unit 470 may receive a first threshold value and a second threshold value from the control unit 240 in order to perform the above-described large and small comparison (S610).

If the absolute value of the correction value is smaller than the first threshold value and all of the comparison values abs_diff1, abs_diff2 and abs_diff3 are smaller than the second threshold value, the correction target line 514 is determined as random noise (S680).

The random noise removal method may be largely divided into a random noise detection step and a random noise removal step by pixel value correction. The random noise detection step may correspond to the comparison value and the magnitude comparison step of the correction value.

If it is determined that the correction target line 514 is random noise, the first image correction unit 480 may remove random noise by a method of compensating pixel values of the correction target line 514. The first image corrector 480 adds the correction value calculated in Equation 4 to the pixel values a ₃₁ , a ₃₂ ,... A _3x of the correction target line 514 (S690).

The pixel value of the image (corrected image) from which the random noise obtained through this operation is removed is called a correction pixel value. The first image corrector 480 outputs 695 the calculated correction pixel value to the controller 240.

As described above, the controller 240 may generate the corrected image by using the corrected pixel value received from the first image corrector 480.

In other cases, that is, when the absolute value of the correction value and the first threshold value are compared with each other, and the comparison values abs_diff1, abs_diff2 and abs_diff3 with the second threshold value, at least one of the correction value or the comparison values is large. In this case, the correction target line 514 may be determined not to be random noise.

Therefore, if the random noise is detected by repeating the above-described process for the other lines, and the random noise is not hidden in the other lines, the original image may be output through the image output unit 250 without removing the random noise. However, if there is pattern noise in the original image, it may be necessary to further remove the pattern noise.

In FIG. 4 and FIG. 5, the first direction is a horizontal direction and the second direction is a vertical direction, and the noise removal apparatus 200 removes random noise generated in the horizontal direction. However, since random noise may also occur in the vertical direction, it is also necessary to remove random noise in the vertical direction.

Therefore, the above-described process is repeated, but the first direction is a vertical direction and the second direction is a horizontal direction, and the average value, the prediction value, the pixel error, the correction value, and the comparison value are calculated to obtain the corrected pixel value. Can be removed. Repeating the above process by changing the horizontal direction and the vertical direction is obvious to those skilled in the art, and thus redundant description will be omitted.

7 is a block diagram illustrating a pattern noise removing unit 230 according to an embodiment of the present invention, and FIG. 8 is a flowchart illustrating a pattern noise removing method according to an embodiment of the present invention. The pattern noise remover 230 may remove pattern noise, which is noise generated at a fixed position of the original image by the method illustrated in FIG. 8.

As illustrated in FIG. 7, the pattern noise remover 230 may include a fixed data input unit 710, a pattern noise calculator 720, and a second image corrector 730.

The fixed data input unit 710 may receive relatively constant data necessary for pattern noise removal (S810 and S820). That is, position information (which may be recognized as coordinates on the original image) of a predetermined position where pattern noise occurs may be input. In addition to the positional information, the pixel values of the original image are also required to remove the pattern noise, and thus this value can also be input.

According to an embodiment, the fixed data input unit 710 may receive the position information or the pixel value of the original image from the controller 240 (S810 and S820).

Alternatively, according to another embodiment, the fixed data input unit 710 may be configured as a part of the image input unit 210 as in the case of the random noise remover.

The pattern noise present in the first direction may be removed using the pixel value of the point included in the pattern noise and the pixel value of the periphery thereof. The pixel values around the pattern noise may mean pixel values of a point adjacent to the pattern noise in the second direction.

That is, the pattern noise calculator 720 filters pixel values of a point included in the pattern noise and pixel values of m points (for example, one or two) adjacent to the point in the + second direction and the-second direction (S830). In this case, the coping value is calculated (S840), and the calculated coping value copes with the pixel value generating the pattern noise. The value calculated at this time is called a coping value. A detailed example of the filtering method will be described with reference to FIG. 9.

The second image correcting unit 730 copes with the pixel value at the point where the pattern noise occurs as the coping value calculated by the pattern noise calculating unit 720 (S840), and includes a pixel value including the coping value of the image from which the pattern noise is removed. These may be output to the control unit 240 (S850).

There may be a case where there are no m pixel values adjacent to the pattern noise in either the + second direction or the-second direction, that is, there are less than m pixels in one direction. In this case, all pixel values (l) adjacent in the + second direction or-second direction and pixel values (2m-1) adjacent in the other direction and pixel values at points included in the pattern noise are thus (2m The pixel values of +1) are subjected to median calculation. As a result of the calculation, the coping value can be calculated by calculating an intermediate value among the (2m + 1) pixel values.

In addition to the above-described method, according to the exemplary embodiment, various mediation targets for mediation operation for removing pattern noise on edges of the original image, such as performing median operation on remaining values except pixel values corresponding to pattern noise, may be used. There may be ways.

9 is a diagram illustrating a process of calculating a coping value according to an embodiment of the present invention.

Referring to FIG. 9, an original image 900 in which vertical pattern noise 910 exists is illustrated. The points constituting the pattern noise 910 have pixel values of q ₀ , q ₁ to q _y . The noise removing apparatus 200 may use pixel values of points adjacent to the pattern noise 910 in the horizontal direction to calculate a coping value.

In the present embodiment, the operation for removing the pattern noise is described as an example of the median operation. According to the median operation, the pattern noise calculator 720 selects an intermediate value between two pixel values of the normal portion other than the pattern noise and the pixel value of the pattern noise portion to cope with the pixel value of the pattern noise portion. As a result, the pattern noise calculator 720 corrects the pixel value of the pattern noise part so that the controller 240 generates a corrected image.

Referring to FIG. 9, (p ₀ , q ₀ , r ₀ ) 920, (p ₁ , q ₁ , r ₁ ) 930,. (p _y , q _y , r _y ) 950 are each subject to median operation.

When the median value of p ₀ , q ₀ , and r ₀ is p ₀ , when the pattern noise operator 720 performs the median operation on (p ₀ , q ₀ , r ₀ ) 920, median ([p ₀ , q ₀ , r ₀ ]) = p ₀ . The second image correction unit 730 for coping q ₀ pattern noise portion to p _0.

_{Similarly, (p 3, q 3,} r 3) (940) when the middle case with a value of _{r 3 (p 3, q 3} , r 3) pattern noise calculating section 720 for 940 of performing a median operation results in median ([p ₃ , q ₃ , r ₃ ]) = r ₃ . Then, the second image corrector 730 copes with q ₃ , which is a pattern noise part, as r ₃ .

If this process is repeated for the points q ₀ to q _y constituting the pattern noise 910, the second image correction unit 730 copes with all the pixel values of the points generating the pattern noise 910 as coping values. By doing this, the pattern noise 910 can be corrected.

In addition, the second image corrector 730 outputs the corrected pixel value including the coping values to the controller 240, and the controller 240 generates a corrected image from which pattern noise has been removed and outputs it to the image outputter 250. You can print

In addition, a program of instructions that can be executed by a terminal such as a computer is tangibly implemented to remove random noise and / or pattern noise, and as a recording medium that can be read by a terminal such as a computer, A recording medium on which a program for performing the noise removing method is recorded can be implemented.

10 to 12 are diagrams illustrating an original image that is an image before noise is removed and a corrected image from which random noise and pattern noise of the original image are removed according to an embodiment of the present invention.

FIG. 10 is a view showing an original image without noise removal according to an embodiment of the present invention, FIG. 11 is a view showing a correction image from which random noise is removed according to an embodiment of the present invention, and FIG. FIG. 1 is a diagram illustrating a corrected image from which random noise and pattern noise are removed according to an embodiment. FIG.

The stripes seen in the vertical direction at the portions indicated by the arrows in FIG. 10 are pattern noise. The stripes excluding the noise indicated by the arrow, which exist in the vertical or horizontal direction throughout the image, correspond to random noise. As described above, random noise occurs at an arbitrary position unlike pattern noise. As shown in FIG. 10, the pattern noise is thicker and clearer than the random noise.

According to FIG. 11, the image after the random noise is removed can be seen that the original image is clearer than the image shown in FIG. 10 except that some thick and dark noise (pattern noise) is visible. This is because the blurred image of the image shown in FIG. 10 is reduced by random noise removal according to an embodiment of the present invention.

According to FIG. 12, the vertical stripes at the point indicated by the arrow in FIG. 10 are removed or lightened. The image shown in FIG. 12 is an image from which random noise and pattern noise are removed according to an embodiment of the present invention. Therefore, the image shown in FIG. 12 can be seen that the object shown in the image is clearly visible compared to the image shown in FIG. 10 or 11.

The above-described embodiments of the present invention are disclosed for the purpose of illustration, and those skilled in the art may make various modifications, changes, and additions within the spirit and scope of the present invention. Should be considered to be within the scope of the following claims.

According to an embodiment of the present invention, the loss of the original image may be minimized in removing noise generated in the form of stripes at a random position or a fixed position of the image.

In addition, according to the embodiment of the present invention, since the processing is performed in the spatial domain without Fourier transform, the operation for removing noise is easy.

Further, according to the embodiment of the present invention, it is possible to provide stable processing results in both random noise removal and pattern noise removal.

In addition, according to an exemplary embodiment of the present invention, since a threshold for distinguishing from image information may be adjusted and set, tuning for image quality improvement according to the size of noise component is easy.

In addition, according to the embodiment of the present invention, when comparing the input image and the resultant image, the resultant image is maintained without blurring and loss of high frequency components (edge and line boundary portions).

Therefore, according to the exemplary embodiment of the present invention, only noise components may be removed while minimizing loss of the original image.

Claims (30)

An image input unit for receiving an original image; In the original image divided into n regions in a first direction, an average value of pixel values for each line in the first direction included in the region and a difference between the average values for each line are filtered to determine whether random noise is present. A random noise removing unit for calculating a corrected pixel value correcting the pixel value of a generation portion of random noise; A controller configured to generate a corrected image from which the random noise is removed using the corrected pixel value; And And an image output unit configured to output the corrected image. The method of claim 1, The random noise removing unit, A random data input unit configured to receive the pixel values of the original image divided into the n areas; First filtering in the first and second directions is performed on the average value to calculate pixel error and correction value, and second filtering is performed on the pixel error to calculate a comparison value, thereby calculating the absolute value of the correction value. An arithmetic and discriminating unit comparing random values and the comparison value with a first threshold value and a second threshold value to detect random noise; And And a first image correcting unit configured to output, to the controller, a correction pixel value obtained by adding the correction value to all the pixel values of the correction target line when the correction target line, which is any of the lines, is recognized as random noise. Noise canceller The method of claim 2, The operation and determination unit An average value calculating unit for obtaining an average value of the pixel values included in the line for each region from the pixel values; And And a prediction value calculator configured to calculate the prediction value corresponding to the correction target line by filtering the average value corresponding to the correction target line and the average value corresponding to an adjacent line adjacent to the correction target line in a second direction. Noise reduction device. The method of claim 3, The predictive value calculator includes the average values corresponding to the correction target line, the one or more adjacent lines in a + second direction with respect to the correction target line, and the one or more adjacent lines in a −2 direction with respect to the correction target line. And calculating the predicted value by averaging the remaining average values except for the smallest and largest values. The method of claim 3, The operation and determination unit An error calculation unit for obtaining a pixel error that is a difference between an average value corresponding to the line on the same line in the first direction and the prediction value, and averaging the pixel error to calculate a correction value And a line that is collinear with the correction target line in the first direction is included in a region different from the region in which the correction target line is included. The method of claim 5, The operation and determination unit The apparatus may further include a random noise discrimination unit comparing the absolute value of the correction value input from the error calculating unit with the first threshold value and recognizing the correction target line as random noise when the absolute value of the correction value is smaller. Noise canceller characterized in that. The method of claim 3, The operation and determination unit Obtaining a pixel error that is a difference between an average value corresponding to a line on the same line in the first direction and the prediction value in the correction target line and the correction target line, And a comparison value calculator which combines the pixel errors by two and calculates a comparison value that is a difference between the two combined pixel errors, wherein the comparison value is a positive number. The method of claim 7, wherein The operation and determination unit And comparing the comparison value received from the comparison value calculator with the second threshold value and, if the comparison value is smaller, further comprising a random noise discrimination unit for recognizing the correction target line as random noise. Device. The method of claim 5, And the error calculating unit obtains a value obtained by subtracting the average value from the predicted value as the pixel error. The method of claim 5, And the error calculating unit calculates a value obtained by rounding a value below the decimal point of the pixel error as the correction value when the correction value calculated by averaging the pixel error is not an integer. The method of claim 1, When the random noise removing unit removes the random noise generated horizontally, the first direction is a horizontal direction and the second direction is a vertical direction. When the random noise generated in the vertical direction is removed, the random noise is removed by using the first direction in the vertical direction and the second direction in the horizontal direction. The method of claim 1, Correcting the pixel value of the pattern noise generation part by filtering the pixel value of the point where the pattern noise occurs and the pixel value of a point adjacent to the pattern noise from the positional information of the point where the pattern noise occurs in the original image. Further comprising a pattern noise removing unit for calculating a coping value, The controller reads the pixel values of the original image and outputs them to the random noise remover and the pattern noise remover, and outputs the position information of the pattern noise to the pattern noise remover, and corrects the corrected pixel values and the response. And a corrected image from which the random noise and the pattern noise are removed by using a value. The method of claim 12, The pattern noise removing unit may receive the position information from the controller, and the fixed data input unit receives the pixel value of the original image from the image input unit; A pattern noise calculator configured to calculate a coping value by filtering the pixel values adjacent to the pattern noise in a second direction when the pattern noise occurs in the first direction; And a second image corrector configured to output the coping value to the controller instead of the pixel value at the point where the pattern noise occurs. The method of claim 13, The pattern noise calculator may be configured to generate the pixel value at the point included in the pattern noise, m of the pixel values adjacent to the pattern noise in the second direction, and m of the pixel values adjacent to the pattern noise in the second direction. And a median operation for calculating a median value as the coping value, wherein m is a natural number of 1 or more. The method of claim 13, When the pattern noise occurs in the vertical direction, the first direction is a vertical direction, and the second direction is a horizontal direction. When the pattern noise occurs in the horizontal direction, the first direction is a horizontal direction and the second direction is Noise canceller, characterized in that the vertical direction. (a) receiving an original image; (b) In the original image divided into n regions in a first direction, the difference between the average value of pixel values per line in the first direction and the average values per line included in the region is filtered to determine whether random noise exists. Determining and calculating a corrected pixel value correcting the pixel value of the random noise generating portion; (c) generating a corrected image from which the random noise is removed using the corrected pixel value; And and (d) outputting the corrected image. The method of claim 16, In step (b), (b1) receiving the pixel value, the first threshold value, and the second threshold value of the original image divided into the n areas; (b2) performing a first filtering in the first and second directions on the average value to calculate a pixel error and a correction value, and performing a second filtering on the pixel error to calculate a comparison value, thereby correcting the correction Comparing the absolute value of the value and the comparison value with the first threshold value and the second threshold value to detect random noise; And and (b3) outputting a correction pixel value obtained by adding the correction value to the pixel value of the correction target line when the correction target line, which is any one of the lines, is determined to be random noise. Way. The method of claim 17, Step (b2) obtaining an average value of the pixel values included in the line for each region from the pixel values; And (b22) calculating a predicted value corresponding to the correction target line by filtering the average value corresponding to the correction target line and the average value corresponding to an adjacent line adjacent to the correction target line in a second direction. Noise reduction method The method of claim 18, Step (b22) is The smallest value among the average values corresponding to the correction target line, the one or more adjacent lines in the second direction with respect to the correction target line and the one or more adjacent lines in the -second direction with respect to the correction target line. And calculating the predicted value by averaging the average of the average value excluding the largest value. The method of claim 18, Step (b2) is (b23) obtaining a pixel error that is a difference between an average value corresponding to a line on the same line in the first direction and the prediction value in the correction target line and the correction target line; And (b24) calculating a correction value by averaging the pixel error, wherein a line on the same line in the first direction is included in a region different from the region in which the correction target line is included; Noise reduction method characterized in that. The method of claim 20, Step (b2) is (b25) further comprising comparing the absolute value of the correction value with the first threshold value and recognizing the correction target line as random noise when the absolute value of the correction value is smaller. Way. The method of claim 18, Step (b2) is (b23) obtaining a pixel error that is a difference between an average value corresponding to a line on the same line in the first direction and the prediction value in the correction target line and the correction target line; (b26) combining the pixel errors by two to calculate a comparison value that is a difference between the two combined pixel errors, wherein the comparison value is positive; Noise removal method further comprising. The method of claim 22, Step (b2) is (b27) further comprising comparing the comparison value calculated in the step (b26) with the second threshold value and recognizing the correction target line as random noise when the comparison value is smaller. How to remove noise. The method of claim 20, And the step (b23) is obtained by subtracting the average value from the predicted value as the pixel error. The method of claim 20, And the step (b23) is to calculate the value obtained by rounding a value below the decimal point of the pixel error as the correction value when the correction value calculated by averaging the pixel error is not an integer. The method of claim 16, In the step (b), when the random noise generated horizontally is removed, the first direction is a horizontal direction and the second direction is a vertical direction. When the random noise generated in the vertical direction is removed, the random noise is removed using the first direction as the vertical direction and the second direction as the horizontal direction. The method of claim 16, (e) grasping the position information of the point where the pattern noise occurs in the original image, and filtering the pixel value of the point where the pattern noise occurs and the pixel value of the point adjacent to the pattern noise to perform the pixel of the pattern noise generating part. Comprising the step of calculating the corrected value of the correction value, And (c) generating a corrected image from which the random noise and the pattern noise are removed using the corrected pixel value and the coping value. The method of claim 27, Step (e) is (e1) receiving the location information and receiving a pixel value of the original image; (e2) calculating a coping value by filtering pixel values adjacent to the pattern noise in a second direction when the pattern noise occurs in a first direction; and (e3) outputting the coping value instead of the pixel value of the point where the pattern noise occurs. The method of claim 28, In the step (e2), among the pixel values at the point included in the pattern noise, the m pixel values adjacent to the pattern noise in the second direction, and the m pixel values adjacent to the pattern noise in the second direction. Noise removal method, characterized in that for performing the median operation to calculate the value corresponding to the median value as a coping value, where m is a natural number of 1 or more. The method of claim 28, When the pattern noise occurs in the vertical direction, the first direction is a vertical direction, and the second direction is a horizontal direction. When the pattern noise occurs in the horizontal direction, the first direction is a horizontal direction and the second direction is Noise canceling method characterized in that the vertical direction.

Images