KR20150037194A - Method to improve the image quality at night - Google Patents

Abstract

According to the present invention, a method for improving image quality at night comprises the steps of: setting one patch around one pixel from an inputted image, and estimating a biggest value of red, green and blue (RGB) color channels of the set patch as an illumination of the patch; applying a Gaussian filter to the illumination component of the patch to reduce distortion due to the shape of the patch; calculating the difference between the illumination component of the patch to which the Gaussian filter is applied and the maximum illumination component which is possible to represent in the inputted image, and outputting the difference as a preliminary correction value needed in a dim portion of the patch; applying a Sigmoid function to the preliminary correction value to determine the applied correction value; and applying a logarithmic (LOG) function to the original image and the applied correction value to calculate the sum of the resultant values and then applying an exponential function to the sum to obtain an improved image.

Description

[0001] The present invention relates to a method for improving image quality at night,

The present invention relates to a nighttime image quality improvement method, and more particularly, to a nighttime image quality improvement method for improving a nighttime image quality by correcting a dark part by estimating an illumination component of an image from an input image .

About half of the thefts committed during the year were about 240,000, of which 61.7% were committed in the afternoon and at night when the person was absent or sleeping. Especially from 10 pm to 4 am, at night. Surveillance cameras are expected to be able to secure security and reduce labor costs even when the person is defenseless. Surveillance camera technology has been developed so that it is possible not only to record images through digitalization but also to transfer them to the control center quickly by using a local area network. Recently, in Japan, Komatsuzaki, It is possible to distinguish between employees and outsiders through face recognition.

In Korea, a high-power infrared light emitter with a visible range of about 200m is being developed and marketed at ANA, a CCTV company, at night. However, when using infrared equipment, there is a disadvantage of losing color information and waste of power and cost for infrared light emission. There are histogram smoothing and gamma correction as image correction without using a light emitting device. In case of histogram smoothing, performance is poor when there is a lot of noise, and gamma correction has a problem that a user has to adjust parameters.

The present invention discloses a method for improving the quality of a night image for improving the nighttime monitoring performance without greatly changing the equipment currently developed by correcting the dark portion and improving the image by estimating the illuminance of the image.

It is therefore an object of the present invention to provide an apparatus and a method for improving the image quality of a night image photographed by a camera.

Another object of the present invention is to improve the image quality of an image by estimating an illumination in a peripheral region of each pixel and using a Gaussian filter and a Sigmoid function, And to provide a method and an apparatus for the same.

A patch is set from one pixel on the basis of the input image and the patch image is set to the largest value among the color channels of the set patches (R, G, B) Determining an application correction value for illumination correction of the patch based on the estimated illumination component and applying the application correction value to the patch to correct the illumination And acquiring the acquired image.

Wherein the determining the application correction value comprises:

Applying a Gaussian filter to the illumination component of the patch to reduce distortion due to the shape of the patch, adjusting the illumination component of the patch to which the Gaussian filter is applied, Calculating a difference from the maximum illumination component and calculating the difference as a preliminary correction value required in the dark portion of the patch, and applying a sigmoid function to the calculated preliminary correction value And determining the applicable correction value.

Wherein the step of acquiring the corrected image comprises:

Calculating a log to the original image and the applied correction value, summing the sum and computing an exponential function on the sum, and obtaining an improved image.

According to the method of improving night image quality according to the present invention, an illumination component of a source night image is estimated, and an application correction value required in the source night image is determined based on the illumination component, It is possible to provide an improved night image.

1 is a block diagram illustrating an apparatus for improving night image quality according to an embodiment of the present invention.
2 is a flowchart illustrating a night image quality improvement method according to an embodiment of the present invention.
Fig. 3 is a view showing the step of detecting the illumination component of the patch. Fig.
FIG. 4 is a view showing steps of improving the image quality in the night image quality improving method according to the present invention.
FIG. 5 is a view showing a comparison between a night image quality improvement method and another night image quality improvement method according to the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description, when an element is described as being connected to another element, it may be directly connected to another element, but a third element may be interposed therebetween. In addition, the structure and size of each constituent element in the drawings are exaggerated for convenience and clarity of description, and a part which is not related to the explanation is omitted. Wherein like reference numerals refer to like elements throughout. It is to be understood that the terminology used is for the purpose of describing the present invention only and is not used to limit the scope of the present invention.

1 is a block diagram illustrating an apparatus for improving night image quality according to an embodiment of the present invention.

Referring to Figure 1,

The night image quality improving apparatus 1 may include a camera sensor 110, a control unit 100, and a display unit 180.

The camera sensor 110 may capture an external image and convert an optical signal sensed during image capturing into an electrical digital signal. The camera sensor 110 may include a charge coupled device (CCD) module or a complementary metal oxide semiconductor (CMOS) module. The image input to the camera sensor 110 is converted into a video signal, which is a digital signal, and is transmitted to the illumination component estimator 120 and the memory 140. The image input to the camera sensor 110 may be the original night image photographed at night.

The control unit 100 may include an illumination component estimating unit 120, a filter unit 130, a memory 140, a preliminary correction value calculating unit 150, an applied correction value determining unit 160, and a calculating unit 170 have. The control unit 100 may be implemented on a single chip, a plurality of chips, or a plurality of electrical components. Various architectures may be used for the control unit 10, including, for example, a dedicated or embedded processor, a single purpose processor, a controller, an ASIC,

The illumination component estimating unit 120 may detect an illumination component of the image signal received from the camera sensor 110.

According to Retinex theory, an image can be expressed as [Equation 1] as a product of an illumination component and a reflection component.

I (x, y) is the brightness of the image, L (x, y) is the illumination component, and R (x, y) is the reflection component.

The illumination component estimator 120 may detect the illumination component of the image signal and may transmit the illumination component of the detected image signal to the filter unit 130. [

The filter unit 130 may apply a Gaussian filter that softens the image to prevent distortion that may occur during the operation of the image signal. The image signal filtered by the Gaussian filter may be transmitted to the preliminary correction value calculator 150 for calculating a correction value.

The memory 140 may store a video signal input from the camera sensor 110. [ The stored video signal may be a digital signal converted by the camera sensor 110. The image signal stored in the memory 140 may be transmitted to the preliminary correction value calculator 150 for calculating a preliminary correction value.

The preliminary correction value calculator 150 calculates the difference between the maximum illumination component value that can be expressed in the original image stored in the memory 110 and the illumination component of the image signal filtered by the filter unit 130, It is possible to obtain a preliminary correction value required in the dark portion of the image. The preliminary correction value calculated by the preliminary correction value calculator 150 may be transmitted to the application correction value determiner 160 to determine an actual correction value required for the original night image.

The application correction value determination unit 160 determines that the preliminary correction value calculated by the preliminary correction value calculation unit 150 does not match the amount of the illumination component (Illumination) that is actually needed and therefore the sigmoid function The preliminary correction value may be adjusted to determine the applied correction value. The application correction value determination unit 160 may transmit the determined application correction value to the operation unit 170. [

The operation unit 170 may perform an operation based on the application correction value determined by the application correction value determination unit 160 to correct the input image. The calculation may be performed by adding a log to the original image and the applied correction value, and then calculating an exponential function. The operation unit 170 may transmit the improved image to the display unit 180 through the operation.

The display unit 180 may output the received video signal to the screen.

The display unit 180 may be a cathode ray tube (CRT), a light emitting diode (LED), a liquid crystal display (LCD), a plasma display panel (PDP), an organic electroluminescence display Device, but is not limited thereto.

2 is a flowchart illustrating a night image quality improvement method according to an embodiment of the present invention.

Referring to FIG. 2,

A patch is set based on one pixel from the input image and the largest value among the color information channels of the set patches is set to the illumination of the patch (S100) of estimating a component (Illumination);

Applying (S110) a Gaussian filter to the illumination of the patch to reduce distortion due to the shape of the patch;

A difference between an illumination component of a patch to which the Gaussian filter is applied and a maximum illumination component that can be expressed in the input image is calculated and the difference is calculated as the darkness of the patch A step (S120) of calculating a preliminary correction value necessary for the part;

A step (S130) of applying the sigmoid function to the calculated preliminary correction value to determine the applied correction value; And

A step S140 of computing an exponential function by summing up the original image and the applied correction value with log, and obtaining an improved image by summing the exponential function;

And an image quality improvement method for a night image.

Hereinafter, a method for improving night image quality according to the present invention will be described in detail.

In the night image quality improvement method according to the present invention,

First, in step S100, one patch is extracted from the input image with one pixel as a center, and a color channel of the extracted patches (R, G, B) And detecting the illumination component of the patch with the largest value among the patches.

(R, G, B) color channel among the patches centering on one pixel of the image as shown in FIG. 3, which is expressed by Equation (2).

는 연산된 결과로 얻은 상기 조명성분(Illumination)을 의미하고 J^c는 원본이미지의 색상 채널(Channel), x는 픽셀(Pixel), y는 x가 속해 있는 패치(Patch)안의 한 픽셀(Pixel)을 의미한다. 상기 연산에서 검출된 조명성분(Illumination)이 큰 값을 가지는 경우는 주변이 밝거나, 물체가 흰색일 때가 있다. 원본 영상의 일정 부분에서 가장 큰 값이 해당 부분 전체의 값이 되므로 저주파의 특성이 있고, 물체의 조명성분(Illumination)과 비슷한 특성을 가지게 된다.
In Equation (2)

Is a pixel (Pixel) in sense the illumination (Illumination) obtained by the calculated result and J ^c is the color channel of the original image (Channel), x is the pixel (Pixel), y is the patch (Patch) which x belongs . When the illumination component detected in the operation has a large value, the surrounding may be bright or the object may be white. Since the largest value in a certain portion of the original image is the value of the entire portion, the low frequency characteristic is obtained and the characteristic similar to the illumination component of the object is obtained.

Next, step S110 is a step of applying a Gaussian filter to the detected illumination component to reduce distortion due to the shape of the patch in step S100.

Since the illumination component detected in the operation of step S100 is affected by the shape of the patch, a Gaussian filter is applied to smooth it.

Next, in step S120, a difference between an illumination component that is Gaussian filtered by the Gaussian filter and a maximum illumination component that can be expressed in the original image is calculated in step S110, The preliminary correction value is calculated.

Next, since the preliminary correction value obtained in step S120 is not matched with the actually required application correction value, step S130 is a step of determining an applied correction value using a sigmoid function .

In Equation (3), y is a difference between a maximum illumination component (Illumination) that can be expressed in an image and a preliminary correction value filtered through a Gaussian filter, and y 'is an application correction value obtained by correcting the difference. λ and m are parameters which are used to determine the applied correction value necessary to correct the dark part of the original image.

Finally, step S140 is an operation for correcting the original image based on the application correction value determined in step S130.

The operation may be performed by adding logarithm to the original image and the applied correction value, and then outputting an improved image by calculating an exponential function.

The above calculation is expressed by equation (4).

I (x, y) is a function of the improved image, L '(x, y) is the applied correction value function, and I (x, y) is the function of the original image.

Fig. 3 is a view showing the step of detecting the illumination component of the patch. Fig.

Referring to Figure 3,

A set of patches is set around one pixel from the image input in the above-mentioned step S100 and a color channel of the extracted patches (R, G, B) And estimating the illumination of the patch 300 with the largest value among the plurality of patches.

FIG. 3 is a diagram illustrating the red color R of the patch 300 in the case of the image 310, the green color G of the patch 300 in the case of the image 420 and the blue color G of the patch 300 in the case of the image 430, (B), and it can be seen that the original image 440 is estimated as the illumination component through the step S100 and is the image 450 representing the illumination component.

FIG. 4 is a view showing steps of improving the image quality in the night image quality improving method according to the present invention.

Referring to Figure 4,

The image 410 indicates a state in which the image quality recognized by the camera, which is generally used as the original night image recognized by the camera, is not improved.

The image 420 is an image that estimates the illumination component through the method of step S100 and displays the image.

The image 430 is an image that determines the applied correction value by applying the image 420 from step S110 to step S130 and displays the applied correction value.

The image 440 represents an improved image by applying the step S 140 to the original night image or the image 430. The illumination component of the dark portion of the image 440 increases and it can be seen that it is clearer than the original image.

FIG. 5 is a view showing a comparison between a night image quality improvement method and another night image quality improvement method according to the present invention.

Referring to Figure 5,

The image 510 is the original night image input through the camera sensor. In the case of the original night image, it can be seen that there is a black background and an unrecognizable building.

The image 520 shows a sharpened image by applying histogram equalization to the original night image 510. However, in the case of a black background, it can be seen that the color becomes too bright and the color distortion becomes worse.

The image 530 shows a sharpened image by applying gamma correction to the original night image 510.

The image 540 applies a method of improving night image quality according to the present invention to the original night image 510 to find that the image is sharp and has less noise than the image 510, .

The method of improving the night image quality according to the present invention has a good performance in terms of visual clarity and distinctness as a result of comparing the result images with the conventional methods. This is especially a result of calculating and correcting the illumination component of the night image.

Meanwhile, the present invention can be embodied in computer readable code on a computer readable recording medium.

A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored. Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device and the like, and also a carrier wave (for example, transmission via the Internet) . In addition, the computer-readable recording medium may be distributed over network-connected computer systems so that computer readable codes can be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the present invention can be easily deduced by programmers skilled in the art to which the present invention belongs.

The present invention has been described above with reference to preferred embodiments. It will be understood by those skilled in the art that the present invention may be embodied in various other forms without departing from the spirit or essential characteristics thereof. Therefore, the above-described embodiments should be considered in an illustrative rather than a restrictive sense. The scope of the present invention is indicated by the appended claims rather than by the foregoing description, and all differences within the scope of equivalents thereof should be construed as being included in the present invention.

100:
110: camera sensor
120:
130:
140: Memory
150: preliminary correction value calculating unit
160: Application correction value determination unit
170:
180:

Claims (3)

A patch is set from one pixel on the basis of the input image and the patch image is set to the largest value among the color channels of the set patches (R, G, B) Estimating a component (Illumination);
Determining an application correction value for illumination component correction of the patch based on the estimated illumination component; And
Applying the application correction value to the patch to obtain a corrected image; The method comprising the steps of:
The method according to claim 1,
Wherein the determining the application correction value comprises:
Applying a Gaussian filter to the illumination component of the patch to reduce distortion due to the shape of the patch;
Calculating a difference between an illumination component of a patch to which the Gaussian filter is applied and a maximum illumination component that can be expressed in the input image, and calculating the difference as a preliminary correction value required in a dark portion of the patch; And
Applying the sigmoid function to the calculated preliminary correction value to determine the applied correction value; The method comprising the steps of:
The method according to claim 1,
Wherein the step of acquiring the corrected image comprises:
And obtaining an improved image using the following equation: < EMI ID = 1.0 >

(X, y) is a function of the improved image, L '(x, y) is the applied correction value function, and I (x, y)

Images