KR101769611B1 - Depth information acquiring device and method thereof - Google Patents

Abstract

A device for obtaining depth information by using an input color image and an input black and white image which are stereo captured according to an embodiment of the present invention comprises: a candidate bleaching parameter set selecting part selecting at least one candidate bleaching parameter set with respect to the input color image from a plurality of random bleaching parameter sets; a bleaching image generating part generating at least one bleaching image corresponding to the input color image by applying the candidate bleaching parameter set; and a first disparity map obtaining part obtaining a first disparity map by stereo matching the at least one bleaching image and the input black and white image.

Description

TECHNICAL FIELD [0001] The present invention relates to a depth information acquisition apparatus,

The present invention relates to an apparatus and method for acquiring depth information.

Recently, smartphone makers are launching smartphone models equipped with stereo cameras. The depth information of the image extracted from the stereo camera can give interesting special effects to the image and can be used to improve the recognition performance.

However, conventional stereo cameras may cause serious errors in depth information measurement due to noise problems amplified in low-light environments. Therefore, a method of overcoming the low light environment is also used by using a separate light source such as a flash or an infrared device.

A technical problem to be solved is to provide an apparatus and method for acquiring depth information robust to a low-illuminance environment without a separate light source.

An apparatus for acquiring depth information using a stereoscopic input color image and an input monochrome image according to an aspect of the present invention includes at least one of a set of a plurality of discoloration parameters and a set of at least one candidate discoloration parameter for the input color image A candidate decoloring parameter set selection unit for deciding a candidate decoloring parameter set by using the candidate decoloring parameter set to generate at least one decolored image corresponding to the input color image; and a controller for performing stereo matching on the at least one decolored image and the input monochrome image And a first disparity map obtaining unit for obtaining a first disparity map.

Wherein the depth information obtaining apparatus includes a gain map for obtaining a gain map and a second disparity for each of the plurality of unit areas of the input monochrome image and a gain for obtaining a second disparity map using the first disparity map And a second disparity map obtaining unit.

The apparatus of claim 1, wherein the depth information obtaining apparatus comprises: a color restoration unit that uses the second disparity map to propagate chrominance information of the input color image to luminacity information of the input monochrome image to generate a color restoration image And the like.

Wherein the candidate decoloring parameter set selection unit selects one of the plurality of decoloring parameter sets so that the difference between the restored color image recovered to the contrast of the anticipated decoloring image to which the decoloring parameter set is applied and the self- Selecting a first set of color decay parameters and selecting at least one second set of color decay parameters having a normalized sparsity of the plurality of sets of color decay parameters less than a predetermined criterion; The at least one candidate discoloration parameter set may be selected as the intersection of at least one second discoloration parameter set.

Wherein the first disparity map obtaining unit obtains the difference between pixel intensity and edge information of the input monochrome image and pixel intensity and edge information of the at least one discoloration image to a minimum value for each pixel, And for each pixel, if the difference between the disparity from the input monochrome image and the disparity from the at least one decolorized image is equal to or greater than a certain reference, a corresponding corresponding disparity Pixels from the first disparity map.

Wherein the gain map and the second disparity map obtaining unit obtains, for each of the unit areas, the gain in which a difference between a pixel intensity of a corresponding point in the unit area and a corresponding pixel intensity of the at least one discoloration image is minimum, A second disparity can be obtained.

Wherein the color depth restoration image has a U channel and a V channel which are Y channel and chrominance channel which are luminance channels and the information of the input monochrome image constitutes the Y channel, The information may constitute the U channel and the V channel.

A method for acquiring depth information using a stereoscopically input color image and an input monochrome image according to another aspect of the invention includes selecting at least one candidate discoloration parameter set for the input color image from among a plurality of sets of discoloration parameters The method comprising: generating at least one discoloration image corresponding to the input color image by applying the candidate discoloration parameter set; and stereo matching the at least one discoloration image and the input monochrome image to obtain a first disparity map Step < / RTI >

The method of acquiring depth information includes obtaining a gain map and a second disparity map including a gain and a second disparity for each of a plurality of unit areas of the input monochrome image using the first disparity map As shown in FIG.

The method of acquiring depth information may further include generating chrominance reconstructed image by propagating chrominance information of the input color image to luminance information of the input monochrome image using the second disparity map.

Wherein the step of selecting the at least one candidate decolorization parameter set further comprises the step of determining whether the difference between the restored color image recovered to the contrast of the anticipated decoloring image to which the decoloring parameter set is applied and the self- Selecting a set of at least one first discoloration parameter that is less than or equal to a normalized sparsity of the plurality of sets of discoloration parameters; And selecting the at least one candidate discoloration parameter set as an intersection of the set of discoloration parameters and the at least one second discoloration parameter set.

Wherein the obtaining of the first disparity map comprises, for each pixel, disparity so that the difference between the pixel intensity and edge information of the input monochrome image and the pixel intensity and edge information of the at least one discoloration image is at a minimum And for each pixel, if the difference between the disparity from the input monochrome image and the disparity from the at least one decolorized image is greater than or equal to a certain reference, Removing from the first disparity map.

Wherein in obtaining the gain map and the second disparity map, for each of the unit areas, a difference between a pixel intensity of a corresponding point inside the unit area and a corresponding pixel intensity of the at least one discoloration image is minimized And obtain the gain and the second disparity.

Wherein the color restored image has a Y channel as a luminance channel and a U channel and a V channel as chrominance channels, the information of the input monochrome image constitutes the Y channel, V channel.

According to the depth information obtaining apparatus and method according to embodiments, robust depth information can be estimated even in a low-illuminance environment without a separate light source.

FIG. 1 is a view for explaining a depth information obtaining apparatus according to an embodiment of the present invention.
2 is a diagram for explaining an input color image.
3 is a diagram for explaining an input monochrome image.
4 is a diagram for explaining a second disparity map.
5 is a diagram for explaining a color restored image.
6 is a view for explaining a refocused image according to an application application.
7 is a diagram for explaining a stylized image according to an application application.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. The embodiments may be implemented in various different forms and are not limited to the embodiments described herein.

In order to clearly illustrate the embodiments, parts not related to the description are omitted, and the same reference numerals are used for the same or similar components throughout the specification. Therefore, reference numerals of the components used in the previous drawings can be used in the following drawings.

FIG. 1 is a view for explaining a depth information obtaining apparatus according to an embodiment of the present invention.

1, the depth information obtaining apparatus 10 according to an exemplary embodiment of the present invention includes a candidate decoloring parameter set selection unit 11, a decoloring image generation unit 12, a first disparity map obtaining unit 13, A gain map and second disparity map acquiring unit 14, and a color restoring unit 15. [

The depth information obtaining apparatus 10 is an apparatus for obtaining depth information by using a stereoscopically input input color image 200 and an input monochrome image 300 (see FIGS. 2 and 3). In the case of the input monochrome image 300, the loss of the light amount is relatively small because the input color image 200 is photographed by applying a color filter to the monochromatic camera. By comparing the enlarged portion of the input color image 200 of FIG. 2 with the enlarged portion of the input black image 300 of FIG. 3, it can be seen that the input black and white image 300 is clearer. In addition, demosaicing for creating a color image in a raw image has a problem of lowering the resolution of a color image. According to the present invention, a vivid color restored image can be generated using a method of transmitting chrominance information of the input color image 200 to the input monochrome image 300.

The depth information acquisition apparatus 10 may be a computing device capable of operating a program such as a desktop, a notebook, and a server. The depth information obtaining apparatus 10 may include at least one processor and a memory. The depth information acquisition method according to an embodiment of the present invention may be pre-programmed in the memory. The processor can drive such a program. On the other hand, such a program may be imported remotely from an external server or the like. On the other hand, the distinction between the processor and the memory may be ambiguous in the depth information acquiring apparatus 10, and it is possible for a person skilled in the art to determine the proportion of software and hardware for implementing the depth information acquiring method.

Although the depth information obtaining apparatus 10 may implement the depth information obtaining method of the present invention by one processor and one memory, the depth information obtaining apparatus 10 may use a plurality of processors and a plurality of memories, or a combination of various software and hardware, Can be implemented. 1, the depth information obtaining apparatus 10 includes a plurality of function units 11, 12, 13, 14, 15 logically divided according to the depth information obtaining method do. It is optional for those skilled in the art to integrate or separate each functional unit into one or a specific number and implement it in hardware and software.

First, the candidate decoloring parameter set selector 11 selects at least one candidate decoloring parameter set for the input color image 200 among the plurality of decoloring parameter sets.

At this time, the candidate decoloring parameter set selection unit 11 selects the candidate decoloring parameter set selector 11 from among the plurality of decoloring parameter sets, the difference between the restored color image recovered to the contrast of the anticipated decoloring image to which the decoloring parameter set is applied and the self- Selecting at least one second set of discoloration parameter sets having a normalized sparsity of less than a certain standard among the plurality of set of discolored parameter sets; And selecting at least one candidate discoloration parameter set as an intersection of the at least one second discoloration parameter set.

First, a method of selecting the first set of discoloration parameters will be described. This is to maintain as much information as possible on the decolorized image despite the decolorizing treatment.

라고 하는 경우, 탈색 이미지

는, 아래 수학식 1과 같이, 입력 컬러 이미지(200)의 적색 컬러 채널

, 녹색 컬러 채널

, 및 청색 컬러 채널

에 탈색 파라미터

,

,

가 각각 적용된 합으로 표현될 수 있다.Decolorized image

When it is called, discoloration image

Of the input color image 200, as shown in Equation (1) below,

, Green color channel

, And a blue color channel

The decolorization parameter

,

,

Can be expressed as a sum applied to each of them.

[Equation 1]

,

,

를 포함한다. 이러한 탈색 파라미터 세트는 탈색 파라미터의 임의 조합에 따라 무한정으로 존재할 수 있다. 따라서 복수의 탈색 파라미터 세트는 그 개수에 제한이 없다. 한 실시예로 각각의 탈색 파라미터는 0.1 간격(interval)을 갖도록 할 수 있다.The color fading parameter set of this embodiment is a color fading parameter

,

,

. This set of bleaching parameters may be present indefinitely depending on any combination of bleaching parameters. Therefore, the number of sets of the plurality of discoloration parameters is not limited. In one embodiment, each of the decolorization parameters may have an interval of 0.1.

를 임의의 탈색 파라미터 세트가 적용된 예상 탈색 이미지

의 콘트라스트(contrast)로 유도함으로써, 복구 컬러 이미지

를 획득할 수 있다.The candidate color fading parameter set selector 11 selects the input color image 200

≪ RTI ID = 0.0 > a < / RTI > expected discoloration image

To the contrast of the restored color image < RTI ID = 0.0 >

Can be obtained.

가 이용될 수 있다(K. He, J. Sun, and X. Tang. Guided image filtering. IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 35(6):1397-1409, 2013.). 참조).

는 각각의 픽셀 i에 대해서 아래 수학식 2와 같이 표현될 수 있다.

는 픽셀 i에 대해서

의 콘트라스트로

를 유도하여

를 출력한다.In this case, the guided filter

(K. He, J. Sun, and X. Tang, Guided image filtering, IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 35 (6): 1397-1409, 2013.). Reference).

Can be expressed by Equation (2) below for each pixel i.

For pixel i

In contrast

To induce

.

&Quot; (2) "

는 픽셀 k를 중심으로한 윈도우(window)이고,

는 윈도우

에 포함된 픽셀의 개수이다.

는 정규화 파라미터(regularization parmeter)이다.

및

는 각각, 윈도우

내의

의 평균(mean) 및 표준 편차(standard deviation)이다.At this time,

Is a window centered on the pixel k,

Windows

Lt; / RTI >

Is a regularization parmeter.

And

Respectively,

undergarment

Mean and standard deviation of the mean.

를 이용하여 입력 컬러 이미지

를 자기유도할 수 있다. 이때 출력 값은

이다.Similarly, the candidate decoloring parameter set selection section 11 selects the candidate decoloring parameter set

Input color image using

Can be magnetically induced. At this time,

to be.

와 복구 컬러 이미지

의 차이가 일정 기준 이하인 적어도 하나의 제1 탈색 파라미터 세트를 선택할 수 있다.The candidate decoloring parameter set selection section 11 selects, according to the following equation (3), the self-induced input color image

Color Image Recovery

At least one first discoloration parameter set whose difference is less than or equal to a certain standard value can be selected.

&Quot; (3) "

는 임의의 탈색 파라미터 세트를 의미하고,

는 복수의 탈색 파라미터 세트를 의미한다.

은 콘트라스트 보존 비용(contrast preserviing cost)를 의미한다.

Quot; means any set of bleaching parameters,

Means a plurality of sets of discoloration parameters.

Means the contrast preserving cost.

을 계산하고, 이산적인 데이터들을 선형적으로 보간(interpolating)시킨 다음, 국소적인 최소점들을 검출함으로써 적어도 하나의 제1 탈색 파라미터 세트인

로 선택할 수 있다.The candidate decoloring parameter set selector 11 selects a candidate decoloring parameter set

, Linearly interpolating the discrete data, and then detecting local minimum points to determine at least one set of first discoloration parameters < RTI ID = 0.0 >

.

Next, a method of selecting the second color decoloring parameter set will be described. This is to prevent noise amplification occurring in a low-light environment.

The candidate color fading parameter set selector 11 can select at least one second color fading parameter set among the plurality of color fading parameter sets whose normalized sparsity is below a certain standard.

는 아래 수학식 4와 같이 표현될 수 있다.Normalized sparse

Can be expressed as Equation (4) below.

&Quot; (4) "

는 탈색 파라미터

로 탈색된 탈색 이미지이고,

는 각각 x, y 방향에 대한 그래디언트 크기(gradient magnitude)이다.

및

는 각각 L₁-놈(norm) 및 L₂-놈을 의미한다.

The decoloring parameter

≪ / RTI > is a decolorized image,

Are the gradient magnitudes for the x and y directions, respectively.

And

Denote L ₁ -nom and L ₂ -nom, respectively.

로 선택할 수 있다. 20% 등 퍼센티지는 당업자의 선택에 의해 적절히 선택될 수 있다.The candidate decoloring parameter set selection unit 11 sets the decoloring parameter set for the values corresponding to 20% of the obtained full normalization dilution value to the second decoloring parameter set

. 20%, etc. may be appropriately selected by a person skilled in the art.

및적어도 하나의 제2 탈색 파라미터 세트

의 교집합

으로 적어도 하나의 후보 탈색 파라미터 세트

를 선택할 수 있다.Finally, the candidate decoloring parameter set selector 11 selects at least one first decoloring parameter set < RTI ID = 0.0 >

And at least one second discoloration parameter set

Intersection of

At least one candidate discoloration parameter set

Can be selected.

를 적용하여 입력 컬러 이미지(200)에 대응하는 적어도 하나의 탈색 이미지

를 생성한다. 탈색 이미지

생성에 대해서는 수학식 1을 다시 참조한다.The discoloration image generating unit 12 generates a discoloration image parameter set

To obtain at least one discoloration image 200 corresponding to the input color image 200

. Discoloration image

For the generation, refer to Equation 1 again.

Next, the first disparity map obtaining unit 13 stereo-matches at least one discoloration image and the input monochrome image 300 to obtain a first disparity map.

At this time, the first disparity map acquiring unit 13 acquires the difference between the pixel intensity and edge information of the input monochrome image 300 and the pixel intensity and edge information of the decoloring image, for each pixel, So as to construct a first disparity map.

에 대한 비용 부피(cost volume)

를 픽셀 강도 차이에 대한 비용 부피

와 엣지 정보 차이에 대한 비용 부피

의 합으로 나타낸다.

은 비용 레이블(cost label)이고,

는 밸런싱 파라미터(balancing parameter)이다(비용 부피에 대해서는 논문 C. Rhemann, A. Hosni, M. Bleyer, C. Rother, and M. Gelautz. Fast cost-volume filtering for visual correspon-dence and beyond. In Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), 2011. 참조).In Equation (5) below,

The cost volume for the < RTI ID = 0.0 >

The cost volume for the pixel intensity difference

And the cost of the difference in edge information

.

Is a cost label,

Is the balancing parameter (see Costs Volume by C. Rhemann, A. Hosni, M. Bleyer, C. Rother, and M. Gelautz, Fast Cost-Volume Filtering for Visual Correction- IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), 2011.).

&Quot; (5) "

The sum of absolute differences (SAD) is used to measure brightness consistency, and the sum of informative edges (SIE) is used to measure edge similarity.

는 아래 수학식 6과 같이 표현될 수 있다.

Can be expressed as Equation (6) below.

&Quot; (6) "

는 픽셀

를 중심으로 하는 지지 윈도우(supporting window)이고,

은 입력 흑백 이미지(300)이고,

는 탈색 이미지이고,

는 디스패리티이고,

는 절단값(truncation value)이다.

Gt;

And a supporting window,

Is an input monochrome image 300,

Is a discoloration image,

Is a disparity,

Is a truncation value.

는 아래 수학식 7과 같이 표현될 수 있다.

Can be expressed by Equation (7) below.

&Quot; (7) "

는 절단값이고, 함수

는 유익한 엣지(informative edge)를 판단하는 데 사용되는 기준(criterion)으로 이용된다(L. Xu and J. Jia. Two-phase kernel estimation for robust motion deblurring. In Proceedings of European Conference on Computer Vision (ECCV). 2010. 참조).

Is the cutoff value,

Is used as a criterion used to judge informative edges (L. Xu and J. Jia. Two-phase kernel estimation for robust motion deblurring. In Proceedings of the European Conference on Computer Vision (ECCV) 2010.).

를 최소로 하는 임의의 픽셀

에 대한 디스패리티

를 획득할 수 있다.Therefore, the first disparity map obtaining unit 13 obtains the cost disparity

Lt; RTI ID = 0.0 >

Disparity for

Can be obtained.

Next, for each pixel, the first disparity map acquiring unit 13 acquires the disparity from the input monochrome image 300 and the disparity from the decolorized image, It can be determined as an outlier and removed from the first disparity map. In this left-right consistency check, the disparity of the corresponding point is compared with the disparity of the corresponding point at the right point, and if the inconsistency is not determined, the pixel is determined as the abnormal point.

Next, the gain map and second disparity map acquiring section 14 acquires the gain for each of the plurality of unit areas of the input monochrome image 300 and the gain of the second monochrome image 300 using the first disparity map, A gain map including parity and a second disparity map 400 are obtained. The gain map and the second disparity map obtaining unit 14 obtains, for each unit area, a gain that minimizes the difference between the pixel intensity of the corresponding point inside the unit area and the corresponding pixel intensity of at least one discoloration image, 2 disparity can be obtained.

Specifically, the input monochrome image 300 is divided into a plurality of unit areas, and each unit area has one gain value and one disparity value. For example, each unit area may be composed of 15 * 20 pixels.

The gain and the second disparity can be obtained according to Equation (8) below.

&Quot; (8) "

는 단위 영역의 인덱스(index)이고,

는 단위 영역

내부의 대응점들의 강도 세트(set of intensities)이다.

는

가 해당 디스패리티로

로 정렬된 이미지(aligned image)이다.

는 디스패리티 추정(disparity estimation)과 게인 조정 루프(gain adjustment loop)에 대한 반복 인덱스(iteration index)이다.

는 지시자 함수(indicator function)로서 단위 영역

내부의 대응점의 개수가 적어도 3 개보다 많을 때 활성화된다. 지시자 함수를 활성화 시키는 대응점의 개수는 당업자의 선택에 의해 임의로 설정될 수 있다.

Is an index of a unit area,

Lt; RTI ID =

It is the set of intensities of the corresponding points inside.

The

To the disparity

Is an aligned image.

Is an iteration index for disparity estimation and gain adjustment loop.

As an indicator function,

And is activated when the number of internal corresponding points is greater than at least three. The number of corresponding points for activating the indicator function can be arbitrarily set by a person skilled in the art.

에 전파할 수 있다(A. Levin, D. Lischinski, and Y. Weiss. Colorization using optimization. In ACM Transactions on Graphics (TOG), volume 23, pages 689-694, 2004. 참조).By solving Equation 8, a dense and smooth gain map can be estimated, and a local affinity model can be used to map these gain maps to the entire image

(A. Levin, D. Lischinski, and Y. Weiss, Colorization using optimization, In ACM Transactions on Graphics (TOG), volume 23, pages 689-694, 2004).

Next, the color restoring unit 15 uses the second disparity map to propagate chrominance information of the input color image 200 to the luminacity information of the input monochrome image 300, Thereby generating an image 500. The color restoration image 500 has a Y channel as a luminance channel and a U channel and a V channel as chrominance channels, the information of the input monochrome image 300 constitutes a Y channel, the information of the input color image 200 is U channel and V channel.

Referring to the color restored image 500 of FIG. 5, it can be seen that a clear color image is restored as compared to the input color image 200 of FIG.

6 is a view for explaining a refocused image according to an application application.

Since there is a second disparity map for the color restored image 500, the depth information can be restored and the refocused image 600 can be generated with focus on the desired portion.

7 is a diagram for explaining a stylized image according to an application application.

Similarly, since there is a second disparity map for the color restored image 500, it is possible to generate the stylized image 700 with specific effects according to the depth information.

It is to be understood that both the foregoing general description and the following detailed description of the present invention are illustrative and explanatory only and are intended to be illustrative of the invention and are not to be construed as limiting the scope of the invention as defined by the appended claims. It is not. Therefore, those skilled in the art will appreciate that various modifications and equivalent embodiments are possible without departing from the scope of the present invention. Accordingly, the true scope of the present invention should be determined by the technical idea of the appended claims.

10: Depth information acquisition device
11: Candidate decoloring parameter set selection unit
12: Discoloration image generating unit
13: First disparity map acquisition unit
14: gain map and second disparity map obtaining unit
15: Color restoration unit

Claims (14)

A candidate decoloring parameter set selection unit that selects at least one candidate decoloring parameter set for an input color image from among a plurality of any of the plurality of decoloring parameter sets;
A discoloration image generating unit applying the set of candidate discoloring parameters to generate at least one discolored image corresponding to the input color image; And
And a first disparity map obtaining unit for obtaining a first disparity map by stereo matching the at least one discoloration image and the input monochrome image,
Wherein the input color image and the input monochrome image are stereoscopic images taken by a color camera and a monochromatic camera, the input color image is an image photographed by the color camera, and the input monochrome image is an image captured by the monochromatic camera Image depth information acquisition device. The method according to claim 1,
A gain map for obtaining a gain map including a gain and a second disparity for each of a plurality of unit areas of the input monochrome image and a gain map for obtaining a second disparity map and a second disparity map Further,
Depth information acquisition device. 3. The method of claim 2,
And a color restoration unit for generating chrominance restoration image by propagating chrominance information of the input color image to luminosity information of the input monochrome image using the second disparity map
Depth information acquisition device. The method according to claim 1,
The candidate decoloring parameter set selection unit
Selecting at least one first discoloration parameter set having a difference between the restored color image recovered to the contrast of the anticipated discoloration image to which the discoloration parameter set is applied and the magnetically induced input color image, and,
Selecting at least one second set of discoloration parameters, the normalized sparsity of which is less than or equal to a predetermined criterion,
Selecting the at least one candidate discoloration parameter set as the intersection of the at least one first discoloration parameter set and the at least one second discoloration parameter set,
Depth information acquisition device. The method according to claim 1,
The first disparity map obtaining unit obtains, for each pixel,
Wherein the first disparity map is constructed by obtaining a disparity so that a difference between pixel intensity and edge information of the input monochrome image and pixel intensity and edge information of the at least one discoloration image is a minimum value,
For each pixel, removing a corresponding pixel from the first disparity map when the difference between the disparity from the input monochrome image and the disparity from the at least one discoloration image is greater than a certain criterion,
Depth information acquisition device. 3. The method of claim 2,
The gain map and the second disparity map obtaining unit
Acquiring, for each of the unit areas, the gain and the second disparity, in which a difference between a pixel intensity of a corresponding point inside the unit area and a corresponding pixel intensity of the at least one discoloration image is minimized,
Depth information acquisition device. The method of claim 3,
Wherein the color restoration image has a Y channel which is a luminance channel and a U channel and a V channel which are chrominance channels,
Wherein the information of the input monochrome image constitutes the Y channel,
Wherein the information of the input color image comprises the U channel and the V channel,
Depth information acquisition device. A depth information obtaining method of a depth information obtaining apparatus,
Selecting at least one candidate bleaching parameter set for an input color image from among a plurality of any of the plurality of bleaching parameter sets;
Applying the candidate discoloration parameter set to generate at least one discolored image corresponding to the input color image; And
And stereo matching the input monochrome image with the at least one decolorized image to obtain a first disparity map,
Wherein the input color image and the input monochrome image are stereoscopic images taken by a color camera and a monochromatic camera, the input color image is an image photographed by the color camera, and the input monochrome image is an image captured by the monochromatic camera An image depth acquisition method. 9. The method of claim 8,
And obtaining a gain map and a second disparity map including a gain and a second disparity for each of the plurality of unit areas of the input monochrome image using the first disparity map
Depth information acquisition method. 10. The method of claim 9,
And propagating the chrominance information of the input color image to the luminance information of the input monochrome image using the second disparity map to generate a color restoration image
Depth information acquisition method. 9. The method of claim 8,
Wherein the selecting of the at least one candidate bleaching parameter set comprises:
Selecting at least one first discoloration parameter set in which the difference between the restored color image recovered to the contrast of the anticipated discoloration image to which the discoloration parameter set is applied and the magnetically induced input color image is less than or equal to a certain criterion, among the plurality of discoloration parameter sets;
Selecting at least one second set of discoloration parameters, wherein the normalized sparsity of the plurality of sets of discoloration parameters is less than or equal to a predetermined criterion; And
And selecting the at least one candidate discoloration parameter set as an intersection of the at least one first discoloration parameter set and the at least one second discoloration parameter set.
Depth information acquisition method. 9. The method of claim 8,
Wherein obtaining the first disparity map comprises: for each pixel,
Wherein the first disparity map is constructed by obtaining a disparity so that a difference between pixel intensity and edge information of the input monochrome image and pixel intensity and edge information of the at least one discoloration image is a minimum value,
And removing, for each pixel, a corresponding pixel from the first disparity map if the difference between the disparity from the input monochrome image and the disparity from the at least one discoloration image is greater than a certain criterion,
Depth information acquisition method. 10. The method of claim 9,
In the step of acquiring the gain map and the second disparity map,
Acquiring, for each of the unit areas, the gain and the second disparity, in which a difference between a pixel intensity of a corresponding point inside the unit area and a corresponding pixel intensity of the at least one discoloration image is minimized,
Depth information acquisition method. 11. The method of claim 10,
Wherein the color restoration image has a Y channel which is a luminance channel and a U channel and a V channel which are chrominance channels,
Wherein the information of the input monochrome image constitutes the Y channel,
Wherein the information of the input color image comprises the U channel and the V channel,
Depth information acquisition method.

Images