KR102195047B1 - Method and apparatus for enhancing quality of 3D image - Google Patents

Abstract

A method of improving 3D image quality is disclosed. The method of improving the 3D image quality is (a) classifying the input 3D image into a plurality of sub-regions based on noise characteristics, and (b) different noise characteristics according to the noise characteristics of the classified plurality of sub-regions. And removing noise from each sub-region of the input 3D image, obtaining a 3D image after removing the noise, and (c) improving the contrast ratio of the 3D image after removing the noise using a removal method.

Description

Method and apparatus for enhancing quality of 3D image TECHNICAL FIELD

The present invention relates to the processing of a 3D image captured by light interference tomography, and more particularly, to a method and apparatus for improving 3D image quality capable of removing noise from a 3D image and improving contrast ratio of a 3D image. .

Light interference tomography (OCT) technology is a high-resolution biological tissue imaging technology that has begun to develop in recent decades. This technology has already been successfully applied to ophthalmic clinical medicine, but its application in the field of imaging very dense tissues, for example, in tumor examination and skin disease examination, is not yet complete. It is difficult to interpret and diagnose organs and lesions in the image because very dense tissues highly dissipate light and have low permeability, seriously degrading the signal to noise ratio of the OCT system and the dynamic range of the generated image. Lose. In order to improve the image resolution of very dense tissues, there is a need to develop a method for quickly and efficiently removing image noise and improving the contrast ratio of image details.

The noise of the OCT image mainly includes additive noise and multiplicative noise, which is also called non-coherent noise and coherent noise. Noise removal methods are mainly divided into hardware-based removal and digital filter-based removal. Hardware-based removal is mainly a blending technique of space, frequency, angle and polarization state. In an ideal situation, this technique can physically remove the additive noise and multiplicative noise of the system, but additional system hardware needs to be added, increasing the manufacturing cost of the system. Since this blending technique requires a separate scanning process and image averaging process, the imaging speed of the system and the contrast ratio of the image are significantly reduced. In comparison, the software-based real-time post-treatment method can remove noise from the image and improve the contrast ratio without affecting the imaging speed, so the hardware-based processing method is used. It's a very good alternative to replace.

Most of the conventional noise reduction and image enhancement methods are applied to 2D images, and different noise reduction algorithms and filters are proposed, mainly based on some rational noise models. These filters are mainly divided into four types: linear filters, nonlinear filters, diffusion filters, and filters based on multi-scale analysis. All of these four conventional filters can suppress noise. However, the process of applying such a filter to an OCT image still has the following limitations: (1) A filter with relatively good effect does not meet the demands of high-speed real-time processing of the OCT system; (2) due to the limitation of the used filter model, some specific noise in the OCT image cannot be processed; (3) There is a problem in that the result of the filter is blurred and the details of the image cannot be displayed.

Recently, an algorithm for removing some 3D OCT image noise has been developed, and this algorithm has improved to some extent the image blurring in the 2D filter filter process. Algorithms based on the averaging method mainly include an averaging method based on motion compensation and an averaging method based on multi-scale wavelet analysis. L.fang, S.Li, Q.nie, JAIzatt, CAToth and S. Farsiu describe a method of removing noise from 3D OCT images based on sparse dictionary in “Sparsity based denoising of spectral domain optical coherence tomography images”. Was presented, and the blurring of the image using the averaging method could be further improved. However, the processing speed of these 3D noise reduction methods could not meet the actual processing needs.

As a general image, it is obtained by a CCD or a photosensitive film, and the noise characteristics of each pixel in the image are all the same. However, OCT images are obtained by scanning one by one point or line, and the noise property of each point is very closely related to the state at a certain point in the system and the state of the scanned object. have. Therefore, the current noise reduction method of 3D OCT images could not effectively remove various noises having different noise characteristics. In short, the following problems still exist in the current image noise removal method: (1) The 2D filter and the 3D filter, which are relatively effective, do not meet the demands of high-speed real-time processing of the OCT system, and (2) the model of the used filter. Due to the limitation of the OCT image, some specific noise and noise properties cannot process other noise; (3) As a result of the filter, the details of the image are blurred, and there is a problem that the image cannot be interpreted in detail.

In order to solve the above defects in the existing technology, an object of the present invention is to provide a method and apparatus for improving the quality of a 3D image capable of removing 3D image noise and improving a 3D image contrast ratio.

A method of improving the quality of a 3D image according to an embodiment of the present invention includes: (a) classifying an input 3D image into a plurality of sub-regions based on noise characteristics; (b) According to the noise characteristics of each of the classified plurality of sub-regions, noise in each sub-region of the input 3D image is removed using different noise removal methods, and a 3D image after noise removal is obtained. Step to do; And (c) improving the contrast ratio of the 3D image after removing the noise.

In the method for improving the quality of a 3D image according to an embodiment of the present invention, in step (a), based on the noise characteristic, the input 3D image is classified into a foreground region and a background region. .

In the method for improving the quality of a 3D image according to an embodiment of the present invention, the step (a) comprises: receiving the input 3D image; Sequentially selecting a target image from the first frame to the last frame of the input 3D image; And classifying the selected target image into a foreground region and a background region based on the noise characteristic.

In the method for improving the quality of a 3D image according to an embodiment of the present invention, a pixel whose brightness exceeds a predetermined threshold in the input 3D image is classified as a foreground area, and the brightness is a predetermined threshold in the input 3D image. Pixels less than or equal to the value are classified as a background area.

In the method of improving the quality of a 3D image according to an embodiment of the present invention, in step (b), noise of the input 3D image is removed using a double-linear noise removal model represented by Equation 5 below. and;

[Equation 5]

, 와

는 각각 상기 입력된 3D이미지의 전경 영역과 배경 영역의 분류 파라미터를 표시하며, 하기의 수학식4에 따라, 상기 픽셀i를 중심으로 상기 입력된 3D이미지의 3D이미지물 속에서 a_{1 i} 와 b_{1 i} 를 계산하고, 상기 픽셀i를 중심으로 상기 입력된 3D이미지의 평면 구역 속에서 a_{2 i} 와 b_{2 i} 를 계산하며:In Equation 5, f _i is the brightness of the pixel i included in the 3D image after noise removal, g _i is the brightness of the pixel i included in the input 3D image, and g _i is the background of the input 3D image Average brightness of all pixels in the area

, Wow

Represents the classification parameters of the foreground area and the background area of the input 3D image, respectively, and a _{1 i} and b in the 3D image of the input 3D image centered on the pixel i according to Equation 4 below. Calculate _{1 i,} and calculate a _{2 i} and b _{2 i} in the plane area of the input 3D image around the pixel i:

[Equation 4]

는 상기 3D이미지물 또는 상기 평면 구역

에 포함된 픽셀의 개수, g _k 는 상기 3D이미지물 또는 상기 평면 구역

에 포함된 픽셀k의 밝기, g _{w i} 는 상기 3D이미지물 또는 상기 평면 구역

에 포함된 모든 픽셀의 밝기의 평균값 및

는 상기 입력된 3D이미지의 각 픽셀의 밝기의 분산값(variance)을 표시하며, 상기 수학식 5에서

와

는 각각 상기 3D이미지물에 포함된각 픽셀의 a_{1 i} 와 b_{1 i} 의 평균값을 표시하고,

와

는 각각 상기 평면 구역에 포함된 각 픽셀의 a_{2 i} 와 b_{2 i} 의 평균값을 표시하는 것을 특징으로 한다.In Equation 4,

Is the 3D image object or the flat area

The number of pixels included in, g _k is the 3D image object or the plane area

The brightness of the pixel k included in, g _{w i} is the 3D image object or the plane area

The average value of the brightness of all pixels included in

Represents the variance of the brightness of each pixel of the input 3D image, in Equation 5

Wow

_Represents the average value of a _{1 i} and b _{1 i} of each pixel included in the 3D image, respectively,

Wow

Is characterized in that the average values of a _{2 i} and b _{2 i} of each pixel included in the planar area are respectively displayed.

In a method for improving the quality of a 3D image according to an embodiment of the present invention, the 3D image object is a 3D cube of 8×8×10 pixels centered on the pixel i, and the planar area is centered on the pixel i. It is characterized by a flat area of one 8×8 pixels.

와 상기

를 결정하며,A method of improving the quality of a 3D image according to an embodiment of the present invention is based on Equation 3 below.

And remind

To determine,

[Equation 3]

는 상기 픽셀i를 중심으로 한 평면 구역

에 포함된 모든 픽셀의 밝기의 평균값,

는 소정의 임계값으로서, 상기 입력된 3D이미지의 배경 구역에 포함된 픽셀의 밝기의 평균값을 표시하는 것을 특징으로 한다.From here,

Is the plane area centered on the pixel i

The average value of the brightness of all pixels included in

Is a predetermined threshold value, and displays an average value of brightness of pixels included in the background area of the input 3D image.

In the method for improving the quality of a 3D image according to an embodiment of the present invention, step (c) comprises: classifying the 3D image after noise removal into a high frequency component and a low frequency component; And improving the contrast ratio of the 3D image after removing the noise by weighting synthesizing the high frequency component and the low frequency component.

In the method for improving the quality of a 3D image according to an embodiment of the present invention, in step (c), the contrast ratio of the 3D image after noise removal is improved based on Equation 6 below:

[Equation 6]

는 향상 후의 3D이미지 속 픽셀i의 밝기,

는 상기 잡음 제거 후의 3D이미지 속의 픽셀i의 밝기, k는 고주파 분량 증가 요소를 표시하고 또한 상기 k는 1보다 큰 실수(實數)이며, 그 중, 하기의 수학식 4에 따라 상기 픽셀i을 중심으로 한 잡음 제거 후의 3D이미지의 평면구역에 포함된

와

를 계산하며:From here,

Is the brightness of the pixel i in the 3D image after enhancement,

Denotes the brightness of the pixel i in the 3D image after noise removal, k denotes a high frequency component, and k denotes a real number greater than 1, of which the pixel i is centered according to Equation 4 below. Included in the plane area of the 3D image after removing the noise

Wow

To calculate:

[Equation 7]

는 상기 평면 구역

에 포함된 픽셀의 개수, g _k 는 상기 평면 구역

에 포함된 픽셀k의 밝기, g _{w i} 는 상기 평면 구역

에 포함된 모든 픽셀의 밝기의 평균값 및

는 잡음 제거 후의 3D이미지의 각 픽셀의 밝기의 분산값(variance)을 표시하며, 그 중에서

와

는 각각 상기 평면 구역 속 각 픽셀에 대해 계산한 a _if 와 b _if 의 평균값을 표시하는 것을 특징으로 한다.In Equation 7,

Is the above flat area

The number of pixels included in, g _k is the plane area

The brightness of the pixel k contained in, g _{w i} is the plane area

The average value of the brightness of all pixels included in

Represents the variance of the brightness of each pixel of the 3D image after noise removal, among which

Wow

Is characterized in that the average values of a _if and b _if calculated for each pixel in the planar region are respectively displayed.

An apparatus for improving the quality of a 3D image according to an embodiment of the present invention includes an image classification module for classifying an input 3D image into a plurality of sub-regions based on noise characteristics; A noise removal module for obtaining a 3D image after noise removal by removing noise in each sub-region of the input 3D image by using different noise removal strategies according to noise characteristics of each of the plurality of sub-regions; And And a detail enhancement module for improving the contrast ratio of the 3D image after removing the noise.

The method and apparatus for improving the quality of a 3D image according to an embodiment of the present invention is an efficient means that can improve image quality by directly applying it to a commercial OCT system due to its fast calculation speed, strong noise removal and detailed enhancement capability. to be.

In the following description, another aspect and/or advantage of the present invention is partially described, and another portion thereof becomes more apparent through the following description or can be known through the practice of the present invention.

By combining the drawings in detail below, the above and other objects, features and advantages of the present invention will become more apparent.
1 is a flowchart illustrating a method of improving the quality of a 3D image according to an embodiment of the present invention.
2A and 2B are views showing an image before noise removal and an image after noise removal.
3A to 3C are comparison diagrams showing images obtained by applying different noise removal methods.
4 is a block diagram showing an apparatus for improving the quality of a 3D image according to an embodiment of the present invention.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the drawings. In the drawings, the same reference numerals always indicate the same structure, characteristics and components.

1 is a flowchart illustrating a method of improving the quality of a 3D image according to an embodiment of the present invention.

Referring to FIG. 1, in step 101, an apparatus for improving 3D image quality classifies an input 3D image into a plurality of sub-areas based on noise characteristics. Specifically, in step 101, the apparatus for improving 3D image quality receives an input 3D image, and sequentially selects a target image from the first frame to the last frame of the input 3D image. An apparatus for improving 3D image quality classifies a selected target image into a foreground region and a background region based on noise characteristics. Additive noise and multiplicative noise mainly exist in 3D OCT (Optical Coherence Tomography) images. Additive noise exists in the entire image area, and multiplicative noise mainly exists in the signal area.

As shown in Fig. 2A, the image is clearly divided into a relatively bright foreground area (signal area) and a relatively dark background area (area with weak or no signal). Based on this point, an apparatus for improving 3D image quality divides an image into a foreground region mainly composed of multiplicative noise and a background region mainly composed of additive noise based on a threshold value. In other words, the device for improving the 3D image quality classifies pixels included in the input 3D image as a foreground area with a brightness greater than a predetermined threshold, and a pixel whose brightness is equal to or less than a predetermined threshold in the input 3D image Is classified as a background area. Here, a predetermined threshold value is expressed as N back , and the threshold value indicates an average value of brightness of all pixels in the background area. Device to improve 3D image quality in the embodiment, on the basis of experiments may be determined for N back suitable for 3D image, and based on the result performed in the existing can determine the appropriate N back to the 3D image. However, the present invention is not limited thereto, and the apparatus for improving the 3D image quality may determine N back suitable for the input 3D image by using any method among existing technologies. In addition, the classification of the sub-region is not limited to classification based on a threshold value, and any method applicable to classifying an image based on noise characteristics of a 3D image may be applied.

In step 102, the apparatus for improving the 3D image quality removes noise from each sub-region of the input 3D image by applying a different noise removal method to each sub-region, based on the noise characteristics of each classified sub-region, A 3D image is obtained after noise removal. Specifically, the model of the 3D OCT image may be expressed by Equation 1.

[Equation 1]

In Equation 1 above, a _i represents the multiplication noise of the pixel i included in the input 3D image, b _i represents the additive noise of the pixel i included in the input 3D image, and f _i represents the 3D noise after noise removal. It represents the brightness of pixel i included in the image, and g _i represents the brightness of pixel i included in the input 3D image.

The apparatus for improving the 3D image quality averages the additive noise in the background area of the input 3D image to remove the noise in the background area, removes most of the additive noise, and then restores the averaged background area using Equation 1 do. Meanwhile, the apparatus for improving the 3D image quality restores the input 3D image by using Equation 1 to remove noise in the foreground region. The noise reduction strategy can be realized by using the double-linear noise reduction model shown in Equation 2 below.

[Equation 2]

와 상기

는 각각 입력된 3D이미지의 전경 영역과 배경 영역을 구분하는 파라미터를 표시한다. 상기

와 상기

하기의 수학식 3에 기초하여 확정될 수 있다.Here, g _i represents the average brightness of all pixels included in the background area of the input 3D image, and

And remind

Represents a parameter that separates the foreground area and the background area of each input 3D image. remind

And remind

It may be determined based on Equation 3 below.

[Equation 3]

는 상기 픽셀i를 중심으로 한 평면 구역

에 포함된 모든 픽셀의 밝기의 평균값,

는 소정의 임계값으로서, 상기 입력된 3D이미지의 배경 구역에 포함된 픽셀의 밝기의 평균값을 표시하는 것을 특징으로 한다. 여기서 상기 평면 구역

는 픽셀i를 중심으로 한 8×8픽셀의 평면 구역이다. 하지만, 이는 일 실시예일 뿐, 평면 구역은 다른 크기로 선택될 수 있다. 예를 들어, 는 픽셀i를 중심으로 한 3D 정육면체에 포함된 모든 픽셀의 밝기 평균값을 나타낼 수 있다.From here,

Is the plane area centered on the pixel i

The average value of the brightness of all pixels included in

Is a predetermined threshold value, and displays an average value of brightness of pixels included in the background area of the input 3D image. Where the above flat area

Is an 8×8 pixel planar area centered on pixel i. However, this is only an example, and the planar area may be selected in different sizes. For example, may denote an average brightness value of all pixels included in a 3D cube centered on pixel i.

Calculate a _{1 i} and b _{1 i} in the input 3D image object centered on the pixel i based on Equation 4 below, and a in the plane area of the input 3D image centered on the pixel i _{2 i} and b _{2 i} can be calculated. Here, the 3D image may be an 8×8×10 pixel 3D cube centered on the pixel i, and the plane area may be a flat area of 8×8 pixels centered on the pixel i. However, the present invention is not limited thereto, and a 3D image object and a flat area of any suitable size may be selected.

[Equation 4]

는 상기 3D이미지물 또는 상기 평면 구역

에 포함된 픽셀의 개수, g _k 는 상기 3D이미지물 또는 상기 평면 구역

에 포함된 픽셀k의 밝기, g _{w i} 는 상기 3D이미지물 또는 상기 평면 구역

에 포함된 모든 픽셀의 밝기의 평균값 및

는 상기 입력된 3D이미지의 각 픽셀의 밝기의 분산값(variance)을 표시한다.In Equation 4,

Is the 3D image object or the flat area

The number of pixels included in, g _k is the 3D image object or the plane area

The brightness of the pixel k included in, g _{w i} is the 3D image object or the plane area

The average value of the brightness of all pixels included in

Represents the variance of the brightness of each pixel of the input 3D image.

에 포함된 각 픽셀k에 대해 계산한 a _k 의 평균값이 실제 기준값으로 사용되었고, 수학식2는 아래 수학식5처럼 수정될 수 있다.In an embodiment of the present invention, a 3D image object or a flat area

The average value of a _k calculated for each pixel k included in is used as an actual reference value, and Equation 2 may be modified as in Equation 5 below.

[Equation 5]

와

는 각각 3D이미지물에 포함된 각 픽셀에 대해 계산한 a_{1 i} 와 b_{1 i} 의 평균값을 표시하고,

와

는 각각 평면 구역에 포함된 각 픽셀에 대해 계산한 a_{2 i} 와 b_{2 i} 의 평균값을 표시한다. 단계 102에서 처리된 입력된 3D이미지는, 입력된 3D이미지의 제1 프레임부터 마지막 프레임까지 순차적으로 선택된 목표 이미지이다.Among them,

Wow

_Represents the average value of a _{1 i} and b _{1 i} calculated for each pixel included in the 3D image,

Wow

_Denotes the average value of a _{2 i} and b _{2 i} calculated for each pixel included in the planar area. The input 3D image processed in step 102 is a target image sequentially selected from the first frame to the last frame of the input 3D image.

2B is a 3D image after noise removal obtained by using the double linear model presented in Equation 5. As compared with FIG. 2A, in FIG. 2B, noise was significantly suppressed, but image details were blurred in the process of removing noise. Therefore, you need to proceed with the procedure below.

In step 103, the apparatus for improving the 3D image quality improves the contrast ratio of the 3D image after noise removal. Specifically, in step 103, the apparatus for improving the 3D image quality first classifies the 3D image after noise removal into a high frequency component and a low frequency component. Then, the apparatus for improving the 3D image quality weights and synthesizes the high frequency component and the low frequency component to improve the contrast ratio of the 3D image. According to an embodiment of the present invention, the contrast ratio of the 3D image after noise removal can be improved based on Equation 6 below.

[Equation 6]

는 향상 후의 3D이미지 속 픽셀i의 밝기,

는 상기 잡음 제거 후의 3D이미지 속의 픽셀i의 밝기, k는 고주파 분량 증가 요소를 표시하고 또한 상기 k는 1보다 큰 실수(實數)이다. From here,

Is the brightness of the pixel i in the 3D image after enhancement,

Denotes the brightness of the pixel i in the 3D image after noise removal, k denotes a high-frequency component, and k denotes a real number greater than 1.

와

으로 한다. 여기에서, 평면 구역은 픽셀i를 중심을 한 8×8픽셀의 평면 구역일 수 있다. 하지만, 본 발명은 이에 한정되지 않고, 3D 이미지 품질을 향상시키는 장치는 다른 크기의 평면 구역을 선택할 수 있다. 그 외에, 실시예에서는, 1보다 크고 10보다 작은 범위에서 적합한 값을 선택하여 고주파 성분 증가 요소k로 사용할 수 있다. In addition, through [Equation 4], a _if and b _if can be calculated in the plane area of the 3D image after noise reduction centered on the pixel i, and then calculated for each pixel in the plane area. the average of a _if and b _if

Wow

To do. Here, the planar area may be a planar area of 8×8 pixels centered on the pixel i. However, the present invention is not limited thereto, and an apparatus for improving 3D image quality may select a planar area of a different size. In addition, in the embodiment, a suitable value may be selected from a range greater than 1 and less than 10 and used as the high frequency component increasing factor k .

FIG. 3A shows an image obtained by processing the image as in FIG. 2A by applying a method for improving the quality of a 3D image according to an embodiment of the present invention. Comparing FIGS. 2A and 2B, according to an embodiment of the present invention The method of improving the quality of 3D images improves image detail and suppresses image noise. FIG. 3B shows an image obtained after removing noise from the image shown in FIG. 2A using a guided filter in the paper “Optical coherence tomography” written by AFFercher, and in FIG. 3C, C. Tomasi and R. Manduchi An image obtained after removing noise from an image as shown in FIG. 2A using a bilateral filter in the paper “Bilateral filtering for gray and color images” is shown. Compared with the part shown by the square in FIG. 3A, the best noise removal effect can be obtained by applying the method of improving the quality of the 3D image according to the embodiment of the present invention.

4 is a block diagram showing an apparatus for improving the quality of a 3D image according to an embodiment of the present invention.

Referring to FIG. 4, an apparatus for improving the quality of a 3D image may include an image classification module 401, a noise removal module 402, and a detail enhancement module 403. The image classification module 401 classifies the input 3D image into a plurality of sub areas based on noise characteristics. For example, the image classification module 401 receives the input 3D image, sequentially selects a target image from the first frame to the last frame of the input 3D image, and converts the target image into a foreground area and a background area according to noise characteristics. It can be classified as As described above, the image classification module 401 may classify the 3D image input through the threshold method into a foreground region and a background region. The noise removal module 402 may remove noise in each sub-region of the 3D image by using different noise removal strategies according to noise characteristics of each sub-region, and obtain a 3D image after noise removal. For example, the noise removal module 402 may remove noise from an input 3D image using the double linear noise removal model shown in Equation (5). The detail enhancement module 403 may improve the contrast ratio of the 3D image after noise removal. Specifically, the detail enhancement module 403 classifies the 3D image after noise removal into a high-frequency component and a low-frequency component, and weights and synthesizes the high-frequency component and the low-frequency component, thereby improving the contrast ratio of the 3D image after noise removal. For example, the detail enhancement module 403 may improve the contrast ratio of the 3D image after noise removal based on Equation (6).

The method of improving the quality of a 3D image according to an embodiment of the present invention can be realized by a computer-readable code or a computer program of a computer-readable recording medium. The computer-readable recording medium is an arbitrary data storage device capable of storing the data and reading data through a computer system.

The method and apparatus for improving the quality of a 3D image according to an embodiment of the present invention has a fast calculation speed, a strong ability to remove noise and improve detail, and is effective in improving image quality by directly applying it to a commercial OCT system. It is a means. In addition, the double-linear noise removal module proposed in the present invention has nonidentity in the image, and can be effectively applied to a medical imaging system using point scanning or line scanning.

Although the present invention has been specifically illustrated and described with reference to exemplary embodiments, the skilled in the art, in a situation that does not exceed the spirit and scope of the present invention defined in the claims, forms and detailed various changes. You should understand that you can proceed.

401: image classification module
402: noise cancellation module
403: detail enhancement module

Claims (10)

(a) classifying the input 3D image into a plurality of sub-regions based on noise characteristics;
(b) According to the noise characteristics of each of the classified plurality of sub-regions, noise in each sub-region of the input 3D image is removed using different noise removal methods, and a 3D image after noise removal is obtained. Step to do; And
(c) improving the contrast ratio of the 3D image after removing the noise; Including,
Step (a),
Receiving the input 3D image;
Sequentially selecting a target image from the first frame to the last frame of the input 3D image; And
And classifying the selected target image into a foreground region and a background region based on the noise characteristic. delete delete The method according to claim 1,
The quality of a 3D image, characterized in that pixels having brightness exceeding a predetermined threshold in the input 3D image are classified as a foreground area, and pixels in the input 3D image having a brightness below a predetermined threshold are classified as a background area. How to improve. The method according to claim 1,
In step (b), noise of the input 3D image is removed using a double-linear noise removal model represented by Equation 5 below;
[Equation 5]

In Equation 5, f _i is the brightness of the pixel i included in the 3D image after noise removal, g _i is the brightness of the pixel i included in the input 3D image, and g _i is the background of the input 3D image Average brightness of all pixels in the area

, Wow

Represents the classification parameters of the foreground area and the background area of the input 3D image, respectively,
According to Equation 4 below, a _{1 i} and b _{1 i} are calculated in the 3D image object of the input 3D image centering on the pixel i, and the plane area of the input 3D image centering on the pixel i Calculate a _{2 i} and b _{2 i} in:
[Equation 4]

In Equation 4,

Is the 3D image object or the flat area

The number of pixels included in, g _k is the 3D image object or the plane area

The brightness of the pixel k included in, g _{w i} is the 3D image object or the plane area

The average value of the brightness of all pixels included in

Represents the variance of the brightness of each pixel of the input 3D image,
In Equation 5 above

Wow

_Represents the average value of a _{1 i} and b _{1 i} of each pixel included in the 3D image, respectively,

Wow

Respectively displaying an average value of a _{2 i} and b _{2 i} of each pixel included in the planar region. The method of claim 5,
The 3D image quality is a 3D cube of 8×8×10 pixels centered on the pixel i, and the plane area is a flat area of 8×8 pixels centered on the pixel i. How to improve it. The method of claim 5,
Based on Equation 3 below,

And remind

To determine,
[Equation 3]

From here,

Is the plane area centered on the pixel i

The average value of the brightness of all pixels included in

Is a predetermined threshold value, and displays an average value of brightness of pixels included in the background area of the input 3D image. The method of claim 1, wherein step (c) is:
Classifying the 3D image after removing the noise into a high frequency component and a low frequency component; And improving the contrast ratio of the 3D image after removing the noise by weighting synthesizing the high frequency component and the low frequency component. The method according to claim 1,
In step (c), the contrast ratio of the 3D image after noise removal is improved based on Equation 6 below:
[Equation 6]

From here,

Is the brightness of the pixel i in the 3D image after enhancement,

Denotes the brightness of the pixel i in the 3D image after noise removal, k denotes a high frequency component, and k denotes a real number greater than 1,
Among them, included in the plane area of the 3D image after noise removal centered on the pixel i according to Equation 4 below

Wow

To calculate:
[Equation 7]

In Equation 7,

Is the above flat area

The number of pixels included in, g _k is the plane area

The brightness of the pixel k contained in, g _{w i} is the plane area

The average value of the brightness of all pixels included in

Represents the variance of the brightness of each pixel of the 3D image after noise removal,

Wow

_Represents an average value of a _if and b _if calculated for each pixel in the planar area, respectively. An image classification module for classifying the input 3D image into a plurality of sub-regions based on noise characteristics;
A noise removal module for obtaining a 3D image after noise removal by removing noise in each sub-region of the input 3D image by using different noise removal strategies according to noise characteristics of each of the plurality of sub-regions; And
It includes a detail enhancement module for improving the contrast ratio of the 3D image after the noise is removed,
The image classification module,
Receive the input 3D image,
Select a target image sequentially from the first frame to the last frame of the input 3D image,
And classifying the selected target image into a foreground region and a background region based on the noise characteristic.

Images