WO2014088259A1

WO2014088259A1 - Apparatus and method for generating hdr image using single image

Info

Publication number: WO2014088259A1
Application number: PCT/KR2013/010921
Authority: WO
Inventors: 이상근
Original assignee: 중앙대학교 산학협력단
Priority date: 2012-12-05
Filing date: 2013-11-28
Publication date: 2014-06-12
Also published as: KR20140072980A; KR101437898B1

Abstract

Disclosed is an apparatus for generating an HDR image using a single image. An exposure image generation unit generates a plurality of exposure images having brightness values which differ from the brightness value of an inputted image with respect to the inputted single image. A detail estimation unit estimates a detail region as a region to be synthesized by calculating a standard deviation with respect to each exposed image. An alpha matte generation unit generates a plurality of alpha mattes on the basis of a region having a maximum standard deviation value by comparing estimated details. A result image generation unit generates one result image by synthesizing the plurality of generated alpha mattes. According to the present invention, since a digital device such as a camera has a narrow dynamic range, which differs from a human visual system (HVS), an improvement of an image is required and thus the digital device can generate and merge a plurality of images having different exposures from one single image.

Description

【Specification】

[Name of invention]

HDR image generating device and method using a single image

Technical Field

The present invention relates to an apparatus and method for generating an HDR image using a single image, and more particularly, to an apparatus and method for generating an HDR image by generating a plurality of images having different exposures from a single image and fusing them. will be.

Background Art

In general, the dynamic range of luminance for a scene viewed by the human eye is very wide, whereas the dynamic range of a device representing the scene is narrow. Therefore, when a high dynamic range (HDR) image is generated by fusing images having various exposures, an image similar to that having a wide dynamic range may be represented on the device. Digital cameras are growing rapidly with interchangeable-lens digital single-lens reflex cameras and mirrorless single-lens cameras, and the market for mirrorless single-lens cameras is expanding rapidly in Japan and Southeast Asia. The market for mirrorless single-lens cameras is expected to grow rapidly from an annual average of 6 to 1.5 million units in 2010, to 930 million units in 2013, and 15 million units in 2015. In 2013, it is expected to surpass the DSLR market (7.5 million units) in size. In 2011, the digital single-lens reflex camera market is expected to grow by 42.1% year-on-year to 1,474 million units, after which growth will slow down due to competition with mirrorless single-lens cameras. Higher functionalization is expected to proceed further.

Looking at the related art in detail, Korean Laid-Open Publication No. 2011-0084025 (name of the invention: an image synthesizing apparatus and method for synthesizing a multi-exposure image) includes an image synthesis for synthesizing a multi-exposure image obtained using an image sensor. Apparatus and method are disclosed. In this way, high-definition HDR images can be generated while expressing details of low and high luminance regions without degrading contrast.

In addition, Korean Laid-Open Patent Publication No. 2012-0112709 (name of invention: high dynamic range image generation and rendering) discloses an algorithm for generating and rendering a high dynamic range digital image.

[Detailed Description of the Invention]

[Technical problem]

The technical problem to be achieved by the present invention is different from the Human Visual System (HVS) Digital devices such as cameras require image enhancement with H with a narrow dynamic range, so HDR can be achieved using a single image that can fuse images by creating multiple images with different exposures from a single image. It is to provide an image generating apparatus and method.

Another technical problem to be solved by the present invention is that unlike a human visual system (HVS), a digital device such as a camera has a narrow dynamic range and thus needs to be improved. An HDR image generation method using a single image that can generate multiple images and fuse the images is to provide a computer-readable recording medium that records a program to be executed on a computer.

Technical Solution

In accordance with an aspect of the present invention, there is provided an HDR image generating apparatus using a single image, and an exposure image generating a plurality of exposure images having different brightness values from the input image. Generation unit; A detail estimator for estimating a detail, which is an area to be synthesized, by calculating a standard deviation for each of the exposed images; An alpha matte generation unit configured to generate a plurality of alpha mattes based on a region having a maximum standard deviation value by comparing the estimated details; And the generated ^'; and a non-sphere; final image generation unit for generating one final image by synthesizing a plurality of the alpha matte.

In accordance with another aspect of the present invention, there is provided a method for generating an HDR image using a single image, in the HDR image generating method performed by the HDR image generating apparatus. An exposure image generation step of generating a plurality of exposure images having different brightness values from the input image; (b) a detail estimating step of estimating a detail, which is an area to be synthesized by calculating a standard deviation for each of the exposed images; (c) generating an alpha matte by comparing the estimated detail and generating a plurality of alpha mattes based on an area having a maximum standard deviation value; And (d) generating a result image by synthesizing the generated plurality of alpha mattes-to generate one result image.

The computer-readable medium according to the present invention for achieving the above another technical problem records a program for executing any one of the above methods in a computer.

Advantageous Effects According to the HDR image generating apparatus and method using a single image according to the present invention, unlike a HVS (Human Visual Syste), a digital device such as a camera has a large dynamic range, so it is necessary to improve the image, Multiple images from different exposures can be generated from a single image to fuse the images.

[Brief Description of Drawings]

1 is a diagram illustrating an image processing method of tone mapping.

2 is a diagram illustrating a dynamic range of a device representing an image;

3 is a diagram illustrating a process of generating an existing HDR image;

4 is a diagram illustrating a process of obtaining a resultant image through the HDR image generating apparatus according to the present invention.

5 is a block diagram showing the configuration of an HDR image generating apparatus according to the present invention, FIG. 6 is a view showing an embodiment of generating a plurality of images having different exposures from a single image;

7 is a diagram illustrating an embodiment of normalizing brightness values for a plurality of images having different exposures;

8 is a diagram illustrating an embodiment of calculating a standard deviation for each exposure image.

9 illustrates an embodiment of estimating detail of a plurality of images having different exposures;

Figure ^"10" alpha matte is a diagram illustrating one embodiment of generating a plurality of,

FIG. 11 is a diagram illustrating an example of blurring by applying a Gaussian function to an alpha matte.

12 to 19 illustrate a resultant image obtained through a tone mapping process from a plurality of exposed images generated by applying the HDR image generating method according to the present invention.

[Form for implementation of invention]

Hereinafter, exemplary embodiments of an HDR image generating apparatus and method using a single image according to the present invention will be described in detail with reference to the accompanying drawings.

In general, referring to HDRC High Dynamic Range, an image processing technique of tone mapping a plurality of Low Dynamic Range (LDR) images into one HDRC High Dynamic Range (LDR) is illustrated in FIG. 1. The HDR technique uses HVS (Hufflan) as shown in FIG. This technique is necessary when the dynamic range of devices representing images is narrow compared to the dynamic range of Visual System. Conventionally, as shown in FIG. 3, not only HDR images are generated using a plurality of images having different exposures, but also a method of selecting a region to be fused using a gradient of each image. Because we got all of the gradients of, G, and B, the progress was slow. Therefore, in the present invention, as shown in FIG. 4, an HDR image is generated through a tone mapping process by generating a plurality of images having different exposures from a single image.

5 is a block diagram illustrating an HDR image generating apparatus 100 using a single image according to the present invention. Referring to FIG. 5, the HDR image generating apparatus 100 (hereinafter, referred to as an 'HDR image generating apparatus') using a single image according to the present invention may include an exposure image generating unit 110, a brightness normalizing unit 120, and a detail. An estimator 130, an alpha matte generator 140, and a resultant image generator 150.

The exposure image generator 110 generates a plurality of exposure images having different brightness values from the input image from the input single image. In this case, two or more exposed images are generated. A plurality of images having different exposures from a single image may be generated from the single image by applying weight values (ω) of n (where n is a positive number of 2 or more) to the input single image.

[Equation 1]

L _n ^c = f _n ^{C fo} r example, f „(I

Where ^ is a single image (c ^'e {R, G, B}) input, ω is the weight value, a large positive than 0, L _n ^c is the resulting exposure of different exposure images (generated by n), And n is a positive number of two or more. Referring to FIG. 6, it can be seen that a plurality of images having different exposures generated from the input single image are generated according to the weight value (ω).

In this case, since a plurality of generated exposure images are different from each other (exposure), the brightness difference is so severe that inaccurate detail (detail) can be estimated. Therefore, in order to reduce the influence of brightness and estimate the detail more accurately, it is desirable to perform a process of normalizing the intensity value. The brightness normalizer 120 performs an intensity normalization process on each exposed image using Equation 2 below as shown in Equation 3 below. [Equation 2]

7 „=

Here, L _n ^c is generated exposure images (n number of times, ce {R, G, B}), and Y _n is grayscale images. [Equation 3]

Where is the brightness normalized image, Y _in is the Y channel of the input image, μ _ίη is the average brightness of the input image, o _in is the brightness standard deviation of the input image, ii _ref is the average brightness of the reference image, and o _rei is the brightness standard deviation of the reference image, where „is created as an odd number so that the middle image can be used as the reference image.

Referring to FIG. 7, after performing an Intensity Normalization process on the Y channel representing the brightness value of the image, it may be confirmed that the brightness difference is similarly changed between each exposed image compared to before normalization. EV-2.0 and EV + 2.0 were normalized to EV + 0 brightness based on the EV + 0 image. In other words, the middle image is used as a reference image.

As described above, since the brightness value is normalized for each of the exposed images, the detail estimator 130 may more accurately estimate a detail, which is an area to be synthesized, using a standard deviation. For this purpose, the standard deviation was calculated for each exposure image by the following equation. As shown in FIG. 8, the standard deviation was calculated once for each exposure image.

[Equation 4]

Where is the detail region of the γ channel, ^ is the number of mask pixels, and W 201

6

^ * Means the width and height of the mask, and, means the average of the masks.

A large standard deviation in the mask means that there are a lot of details in the mask, and the estimated details vary with the exposure of each image. Referring to FIG. 9, when the estimated detail image is compared with the above image, the estimated detail may be determined to estimate the most optimal detail among the three images.

When comparing the estimated detail value for each exposure image, the mask with the largest value means that it has the best detail for generating the resulting image. In consideration of this, the alpha matte generation unit 140 generates a plurality of alpha mattes. An alpha matte is generated using Equation 5 below to determine the part to be synthesized into the resultant image. That is, referring to FIG. 10, it can be seen that three alpha mattes are generated.

(Equation 5)-^ D ^ L))

Also, for natural synthesis, the alpha matte is blurred using a Gaussian filter. That is, as shown in FIG. 11, the Al matte shown in FIG. 10 is blurred using a ginsian filter for natural image synthesis. Downsample the matte, perform a convolution with a Gaussian filter, and upsample to the original image size.

[Equation 6]

Where ^B "is the blurred alpha matte, ^Ai Shae alpha matte, and G is

2?

Gaussian function, ^G O ⁼ e.

The result image generator 150 synthesizes n alpha mattes to generate one result image. In this case, as shown in Equation 7 below, multiply the blurred alpha matte by each color channel of the corresponding exposure image, In addition, the resulting image is generated. [Equation 7]

12 to 19 illustrate a resultant image obtained through a tone mapping process from a plurality of exposed images generated by applying the HDR image generating method according to the present invention. 12 to 14 illustrate an embodiment in which the final image is obtained by generating three exposure images having different brightness values from the single input image. 15 to 16 illustrate an embodiment in which six final exposure images having different brightness values are generated from a single input image to obtain a final image. Further, referring to FIGS. 17 to 19, when generating an exposure image having ⁷ different brightness values from a single input image, not only the input image is brightened but also the exposure image that is made dark is also generated to derive the result image. That is, as a result of applying the HDR image generation method using the single image according to the present invention, it can be seen that the image generates the HDR image more efficiently than the conventional method.

The invention can also be embodied as computer readable code on a computer readable recording medium. Computer-readable recording media include all kinds of recording devices that store data that can be read by a computer system. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disks, optical data storage devices, and the like, which may also be implemented in the form of carrier waves (transfer over the Internet). . The computer-readable recording medium may also be distributed over computer systems connected through wired and wireless communication networks so that the computer-readable code may be stored and executed in a distributed manner.

Although the preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the above-described specific preferred embodiments, and the present invention belongs to the present invention without departing from the gist of the present invention as claimed in the claims. Anyone of ordinary skill in the art can make various modifications, as well as: TL, such changes are within the scope of the claims.

Claims

[Range of request]

[Claim 1]

An exposure image generator configured to generate a plurality of exposure images having a brightness value different from the input image with respect to the input single image;

A detail estimator for estimating a detail, which is an area to be synthesized by calculating a standard deviation for each of the exposed images;

An alpha matte generation unit configured to generate a plurality of alpha mattes based on a region having a maximum standard deviation value by comparing the estimated details; And

And a result image generator for synthesizing the generated plurality of alpha mats to generate one result image.

[Claim 2]

The method of claim 1,

And a brightness normalization unit for normalizing brightness values for each of the generated exposure images.

And the detail estimating unit calculates a standard deviation after normalizing brightness values for each of the exposed images.

[Claim 3]

The HDR image generating apparatus according to claim 1 or 2, wherein the exposure image generating unit sets η weight values to generate η images having different exposures.

[Claim 41]

The HDR image generating apparatus of claim 3, wherein the weight value (° ^) is a positive number greater than zero.

[Claim 5]

The method according to claim 1 or 2,

And the resultant image generator is configured to apply a blurring filter to each of the plurality of alpha mattes and then apply them to each color channel of the exposed image.

[Claim 6]

The method of claim 5, And the recalling filter performs a convolution with the alpha matte using a Gaussian filter.

[Claim 7]-

In the HDR image generating method performed by the HDR image generating apparatus,

(a) an exposure image generation step of generating a plurality of exposure images having different brightness values from the input image with respect to the input single image;

(b) a detail estimating step of estimating a detail, which is an area to be synthesized by calculating a standard deviation for each of the exposed images;

(c) generating a plurality of alpha mattes based on a region having a maximum standard deviation value by comparing the estimated details; And

An HDR image generation method comprising: (d) a result image generation step of generating one resultant image by synthesizing the generated plurality of alpha mattes;

[Claim 8]

The method of claim 7,

The step (b) further includes a brightness normalization step of normalizing the brightness value for each of the generated exposure image;

^'HDR image generation method, characterized in that for calculating the deviation pyochun after normalizing the luminance value for each of said exposure image.

[Claim 9]

9. The method of claim 8, wherein the step (a) sets η weight values ( ^ω ) to generate η images having different exposures.

[Claim 10]

10. The method of claim 9, wherein the weight value ^ is a positive number greater than zero.

[Claim 11]

The method of claim 8,

Step (d) is a blurring filter for each of the plurality of alpha mattes. HDR image generation method characterized in that to apply to each color channel of the exposure image after applying.

[Claim 12]

The method of claim 11,

The blurring filter performs a convolution with the alpha matte using a Gaussian filter.

[Claim 13]

A computer-readable recording medium having recorded thereon a program for executing the HDR image generating method according to any one of claims 7 to 12 on a computer.