KR101970562B1 - Device for correcting depth map of three dimensional image and method for correcting the same - Google Patents

Abstract

The present invention relates to an apparatus and method for correcting a depth map of a three-dimensional image that can remove noise in a depth map to improve the quality of a stereoscopic image. The apparatus includes a first depth- And a depth map correction unit for setting the threshold frequency value based on the first histogram information and correcting the depth map based on the threshold frequency value.

Description

TECHNICAL FIELD [0001] The present invention relates to a device for correcting a depth map of a three-dimensional image and a method of correcting the depth map,

The present invention relates to a depth map correction apparatus for a three-dimensional image, and more particularly, to a depth map correction apparatus and a correction method for a three-dimensional image capable of improving the quality of a stereoscopic image by removing noise in a depth map .

3D stereoscopic images are generated by combining the two-dimensional images, the left eye image and the right eye image. In order to generate such a stereoscopic image, information on the stereoscopic distance difference (depth difference) between the objects included in the left eye image and the right eye image is required, and this information is contained in a depth map.

Since the image quality of the three-dimensional image is influenced by the accuracy of the depth map, it is important to eliminate errors and noise in the depth map in order to improve the image quality of the three-dimensional image.

However, since errors and noise in the depth map can not be precisely removed in the past, the problem that the image quality of a three-dimensional image is deteriorated can not be avoided.

SUMMARY OF THE INVENTION The present invention has been made in order to solve the above-mentioned problems, and it is an object of the present invention to provide an image processing apparatus and method capable of improving image quality of a three-dimensional image by removing errors and noise in a depth map based on a threshold frequency value in histogram information, And an object of the present invention is to provide a map correction apparatus and a correction method.

According to another aspect of the present invention, there is provided an apparatus for correcting a depth map of a three-dimensional image, the apparatus comprising: a first histogram generating unit generating first histogram information on a depth map, And a depth map correction unit for correcting the depth map based on the threshold frequency value.

And the depth map corrector corrects the depth map by resetting the tone values appearing at frequencies lower than the threshold frequency value to different values.

Wherein the depth map corrector comprises: a first histogram generator for generating first histogram information for the depth map; A threshold frequency setting unit for setting a threshold frequency value based on first histogram information from the first histogram generation unit; The histogram information included in the first histogram information based on the first histogram information from the first histogram generation unit and the threshold frequency value from the threshold frequency setting unit to the high frequency tone values, the low frequency tone values, A high frequency tone value, a maximum high frequency tone value, a low region tone value, an intermediate region tone value, and a high region tone value; The gradation value of the depth map corresponding to the low region gradation values of the first histogram information is replaced with the minimum high frequency gradation value by referring to the classification result from the gradation value classifying portion, A first correcting unit for first correcting the depth map by replacing all the gray level values of the depth map corresponding to the high gray level values of the information with the maximum high frequency gray value; And a gradation value classifying unit for classifying the gradation values of the primary correction depth map corresponding to the intermediate gradation values of the first histogram information with gradation values spatially adjacent to the gradation values of the first gradation value information, And a second secondary control unit for performing secondary correction.

Wherein the tone value classifying unit classifies the tone values appearing at a frequency lower than the threshold frequency value into low frequency tone values based on the first histogram and the threshold frequency value, Frequency tone gradation values to classify the lowest gradation value among the high frequency gradation values into a minimum high frequency gradation value and to classify the largest gradation value among the high frequency gradation values as a maximum high frequency gradation value Classifying the low frequency tone values smaller than the minimum high frequency tone value into low tone tone values, classifying the low frequency tone tone values larger than the maximum high frequency tone value into high tone tone values, The low frequency tone values located between the minimum high frequency tone value and the maximum high frequency tone value are classified into intermediate tone values.

When the second gradation value corrects one gradation value in the primary correction depth map corresponding to any one of the intermediate gradation values, the secondary gradation value correcting unit corrects the gradation value of the primary correction depth map, which is located spatially close to the one gradation value, And the one tone value is replaced with the calculated average value. In this case, the one tone value is replaced with the calculated average value.

Wherein the threshold frequency setting unit generates cumulative histogram information based on the first histogram information and sets the threshold frequency value based on the cumulative histogram information.

Wherein the threshold frequency setting unit first differentiates the cumulative curve included in the cumulative histogram information by a gradation value interval from a lowest gradation value to a highest gradation value to calculate a slope for each gradation period; Selects one of the gradation values included in the gradation period having the sharpest slope among the gradations; From the grayscale values in the first histogram information, a grayscale value equal to the selected grayscale value; The frequency value in the first histogram information corresponding to the found tone value is set as the threshold frequency value.

Wherein the threshold frequency setting unit sets a frequency value located at a lower k% (k is a natural number smaller than 100) among the frequency values included in the accumulated histogram information as a reference frequency value; Calculating a slope for each gradation section by first differentiating an accumulation curve included in the cumulative histogram information by a gradation value interval from a lowest gradation value to a highest gradation value; Selects one of the gradation values included in the gradation period having the sharpest slope among the gradations; Comparing the frequency value corresponding to the selected gradation with the reference frequency value; And if the frequency value is greater than the reference frequency value, searching for the same tone value as the selected tone value from the tone values in the first histogram information; The frequency value in the first histogram information corresponding to the found tone value is set as the threshold frequency value.

Wherein the threshold frequency setting unit sets the threshold frequency of the other one of the plurality of pixels included in the gradation section having the sharpest slope to a value of the other frequency value larger than the reference frequency value when the frequency value is not greater than the reference frequency value, And the reference frequency value is compared with the frequency value corresponding to one of the tone values included in the gradation section having a sharp slope and then compared with the reference frequency value .

And the reference frequency value is a frequency value located in the lower 1% of the frequency values included in the accumulated histogram information.

Wherein the depth map corrector comprises: a second histogram generator for generating a second histogram based on a secondary correction depth map provided from the secondary controller; And correcting the second histogram information based on a preset lower limit cutoff value and an upper limit cutoff value, and based on the corrected second histogram information and the critical frequency value from the critical frequency setting unit, And a third-order bending unit for correcting the third-order bending unit.

The lower limit cutoff value is a gradation value located at a lower i% (i is a natural number smaller than 100) among the gradation values included in the second histogram information, and the upper limit cutoff value is a gradation value contained in the second histogram information (Where j is a natural number smaller than 100) among the upper j (j is a natural number smaller than 100).

Wherein the third beam forming section classifies the tone values smaller than the lower limit blocking value in the second histogram into lower limit tone values; Classify the gradation values larger than the upper limit cutoff value in the second histogram into the upper limit excess gradation values; The minimum high frequency tone value and the maximum high frequency tone value are reset based on the corrected second histogram information in which the lower limit tone values and the upper limit tone values are removed and the threshold frequency value from the threshold frequency setting unit ; Replacing all the tone values of the secondary correction depth map corresponding to the lower-limit tone values of the second histogram information with the reset minimum high-frequency tone value; The third correction depth map is generated by replacing all of the tone values of the secondary correction depth map corresponding to the upper-limit excess tone values of the second histogram information with the reset maximum high frequency tone value.

According to another aspect of the present invention, there is provided a method for correcting a depth map of a three-dimensional image, the method comprising: generating first histogram information for a depth map; generating a first histogram information based on the first histogram information; And a step of correcting the depth map based on the threshold frequency value.

In the first step, the depth map is corrected by resetting the tone values appearing at frequencies lower than the threshold frequency value to different values.

The first step may include: 1-A step of generating first histogram information on the depth map; A 1-B step of setting a threshold frequency value based on the first histogram information generated from the step 1-A; Wherein the gray level values included in the first histogram information are calculated based on the first histogram information generated from the 1-A step and the threshold frequency value generated from the 1-B step to a minimum high frequency tone value, a maximum high frequency tone value, A 1-C step of classifying the image data into low region tone values, intermediate region tone values, and high region tone values; The gradation values of the depth map corresponding to the low region gradation values of the first histogram information are all replaced with the minimum high frequency gradation value by referring to the classification result from the 1-C step, and the first histogram A 1-D step of first adjusting the depth map by replacing all the gray level values of the depth map corresponding to the high gray level values of the information with the maximum high frequency gray level value; And comparing the gradation values of the primary correction depth map corresponding to the intermediate region gradation values of the first histogram information with the gradation values spatially adjacent to each other with reference to the classification result from the 1-C step And a 1-E step to be secondarily determined.

The 1-C step divides the tone values appearing at a frequency lower than the threshold frequency value into low frequency tone values based on the first histogram and the threshold frequency value, Frequency tone gradation values, classifies the lowest gradation value among the high frequency gradation values into a minimum high frequency gradation value, and outputs the largest gradation value among the high frequency gradation values as a maximum gradation value Frequency gradation values smaller than the minimum high frequency gradation value into low region gradation values and classifying low frequency gradation values larger than the maximum high frequency gradation value into high region gradation values And classifies the low frequency tone values located between the minimum high frequency tone value and the maximum high frequency tone value as intermediate tone values.

In the step 1-E, when correcting any one of the intermediate tone values, an average value of the total sum of the tone values of any one of the intermediate tone values and the neighboring tone values spatially adjacent thereto is calculated, And the one intermediate area tone value is replaced with the calculated average value.

In the step 1-B, cumulative histogram information is generated based on the first histogram information, and the threshold frequency value is set based on the accumulated histogram information.

The 1-B step includes a 1-B1 step of calculating a slope for each gradation section by first differentiating an accumulation curve included in the cumulative histogram information by a gradation value interval from a lowest gradation value to a highest gradation value; A 1-B2 step of selecting one of the grayscale values included in the grayscale section having the sharpest slope among the slopes; 1-B3 step of finding the same tone value as the selected tone value from the tone values in the first histogram information; And a 1-B4 step of setting a frequency value in the first histogram information corresponding to the found tone value as the threshold frequency value.

1-B1 step of setting a frequency value located at a lower k% (k is a natural number smaller than 100) among the frequency values included in the accumulated histogram information as a reference frequency value; A 1-B2 step of firstly differentiating an accumulated curve included in the cumulative histogram information from a lowest gradation value to a highest gradation value according to a gradation value interval to calculate a slope for each gradation interval; A 1-B3 step of selecting one of the grayscale values included in the grayscale having the sharpest slope among the slopes; A 1-B4 step of comparing the frequency value corresponding to the selected gradation with the reference frequency value; A 1-B5 step of finding, from the tone values in the first histogram information, the same tone value as the selected tone value when the frequency value is greater than the reference frequency value; And a 1-B6 step of setting a frequency value in the first histogram information corresponding to the found tone value as the threshold frequency value.

Wherein the step 1-B is a step of comparing the frequency value with the reference frequency value in order to find another frequency value having a value larger than the reference frequency value as a result of the comparison from the step 1-B4, A frequency value corresponding to one of the tone values included in the gradation section is compared with the reference frequency value, or a frequency value corresponding to one of the tone values included in the gradation section having a sharp gradient, And comparing the frequency values with each other.

And the reference frequency value is a frequency value located in the lower 1% of the frequency values included in the accumulated histogram information.

The first step may include a 1-F step of generating a second histogram based on the secondary correction depth map generated from the 1-E step; And correcting the second histogram information based on the preset lower limit cutoff value and the lower limit cutoff value, and based on the corrected second histogram information and the critical frequency value from the 1-B step, 1 < / RTI >

The lower limit cutoff value is a gradation value located at a lower i% (i is a natural number smaller than 100) among the gradation values included in the second histogram information, and the upper limit cutoff value is a gradation value contained in the second histogram information (Where j is a natural number smaller than 100) among the upper j (j is a natural number smaller than 100).

1-G1 step of classifying the tone values smaller than the lower limit cut-off value in the second histogram into lower-limit tone values; 1-G2 step of classifying the tone values larger than the upper limit cut-off value in the second histogram into the upper limit tone values; The minimum high frequency tone value and the maximum high frequency tone value are reset based on the corrected second histogram information in which the lower-limit tone values and the upper-limit excess tone values are removed and the threshold frequency value from the 1-B step 1-G3 step; 1-G4 step of replacing all the tone values of the secondary correction depth map corresponding to the lower-limit tone values of the second histogram information with the reset minimum-high-frequency tone values; The 1-G5 step of generating a tertiary correction depth map by replacing all the tone values of the secondary correction depth map corresponding to the upper-limit excess tone values of the second histogram information with the reset maximum high frequency tone value .

The apparatus and method for correcting a depth map of a three-dimensional image according to the present invention have the following effects.

First, in the present invention, any one frequency value distributed on the Y axis of the histogram information is set as a threshold frequency value, and the depth map is corrected on the basis of the threshold frequency value. Therefore, a frequency that does not reach the threshold frequency value It is possible to find all of the gray level values and correct them all.

Second, more effective data compensation is possible by correcting the gray level values in the depth map corresponding to the intermediate gray level values among the gray level values that do not reach the critical frequency value, based on the neighboring gray level values spatially close to each other.

Third, since the threshold frequency value is set based on the histogram information and the accumulated histogram information, the image characteristic can be reflected in the threshold frequency value.

Fourth, it is possible to remove noise due to image characteristics by removing an extremely high gray level value or an extremely low gray level value by using the lower limit blocking value and the upper limit blocking value, and compensating for the different values.

1 is a block diagram of a three-dimensional image generating apparatus according to an embodiment of the present invention.
FIG. 2 is a view for explaining a process of generating a left-eye depth map and a right-eye depth map, which are generated from the depth-
FIG. 3 is a block diagram of the left-eye depth map correcting section of FIG.
4 is a view for explaining a depth map generated from the depth map generating unit;
5 is a diagram showing first histogram information
6 is a diagram for explaining a method of classifying gray level values in first histogram information
7 is a diagram for explaining a method of first-order-correcting a depth map
8 is a diagram for explaining a method of secondary correction of the primary correction depth map
9 is a diagram showing second histogram information
FIG. 10 is a diagram for explaining a third-order correction method for a secondary correction depth map that does not include the lower-limit gradation value and the upper-limit excess gradation values at all
11 is a diagram showing a final correction depth map
12 to 14 are diagrams for explaining a third correction method for a secondary correction depth map including a lower limit gradation value
FIG. 15 is a diagram showing cumulative histogram information for the first histogram information of FIG. 5; FIG.
16 is a diagram for explaining a method of setting a threshold frequency value from the first histogram information
17 to 19 are diagrams for explaining the effect of the present invention

1 is a block diagram of a three-dimensional image generating apparatus according to an embodiment of the present invention.

1, a 3D image generating apparatus according to an embodiment of the present invention includes a depth map generating unit (DMG), a depth map correcting device (DM-CRU), and a stereoscopic image generating unit (DVG) .

The depth map generating unit DMG receives the left eye image data img_L and the right eye image data img_R and generates a left eye depth map d-mp_L and a right eye depth map d-mp_R. Here, the left eye image data img_L and the right eye image data img_R are two-dimensional image data for images (left eye image, right eye image) obtained by photographing the same image at different angles. The left eye image data img_L and the right eye image data img_R are all image data of one frame. The left-eye depth map (d-mp_L) and the right-eye depth map (d-mp_R) are maps for expressing relative distances between a plurality of subjects included in an image of one frame in terms of a gradation difference. The left eye depth map d-mp_L represents the relative distance between the objects based on the left eye image data img_L and the right eye depth map d-mp_R represents the relative distance between the objects based on the left eye image data img_L, And the relative distance between the two.

Referring to FIG. 2, the left-eye depth map d-mp_L and the right-eye depth map d-mp_R will be described in more detail.

FIG. 2 is a diagram for explaining a process of generating a left-eye depth map d-mp_L and a right-eye depth map d-mp_R generated from the depth map generating unit DMG of FIG. (D-mp_L), and FIG. 2 (b) shows the process of generating the depth map of the right eye (d-mp_R). 2 (a) and 2 (b) are referred to as " Fig. 2 (b) " Explanation of (a) and (b) in FIG. 8 and FIG. 12 will be described by attaching an English letter to the corresponding numbers as described above.

As shown in FIG. 2 (a), both the left eye image IMG_L and the right eye image IMG_R include a circle CC and a triangle TA. Assume here that the circle CC is projected further forward than the triangle TA and that the triangle TA is located considerably behind the circle CC. Generally, as the subject further protrudes further, the amount of positional change between the subject in the left eye image IMG_L and the subject in the right eye image IMG_R increases further. On the other hand, as the subject is located further behind, the amount of positional change of the subject decreases. For example, as shown in FIG. 2 (a), it is known that the position of the circle CC included in the left eye image IMG_L and the position of the circle CC included in the right eye image IMG_R are significantly different . On the other hand, the position of the triangle TA included in the left eye image IMG_L and the position of the triangle TA included in the right eye image IMG_R are substantially the same.

The depth map generation unit DMG compares the left eye image data corresponding to the left eye image IMG_L with the right eye image data corresponding to the right eye image IMG_R to determine whether the circle and the triangle TA included in the left eye image And calculates how far it has moved from the position where the included circle and the triangle TA are located. Then, based on the calculated result, the tone values of the pixels constituting the circle CC in the left eye image IMG_L and the tone values of the pixels constituting the triangle TA are set, .

As shown in the left eye depth map d-mp_L in Fig. 2 (a), the circle CC, which is a relatively moving subject, is displayed with bright tone values, while a triangle (TA) is represented by dark gray scale values. As a result, subjects that are more projected in the future will be displayed with bright gradation values, while those located behind will be displayed with darker gradation values. Similarly, the right eye depth map d-mp_R is generated through comparison between the right eye image IMG_R and the left eye image IMG_L based on the right eye image IMG_R as shown in FIG. 2 (b) .

The depth map correction unit DM-CRU receives a left-eye depth map d-mp_L and a right-eye depth map d-mp_R from the depth map generating unit DMG, ) And a right eye correction depth map (d-mp_R_C). To this end, the depth map correction apparatus DM-CRU includes a left-eye depth map correction unit DMC_L for generating a left-eye correction depth map d-mp_L_C based on a left-eye depth map d- And a right eye depth map correction unit (DMC_R) that generates a right eye correction depth map (d-mp_R_C) based on the depth map (d-mp_R). The left eye depth map correcting unit DMC_L corrects errors or noise in the left eye depth map d-mp_L and the right eye depth correcting unit DMC_R corrects the right eye depth map d- Such as an error or noise in the image.

The stereoscopic image generating unit DVG generates a stereoscopic image based on the left eye correction depth map d-mp_L_C provided from the left eye depth map correction unit DMC_L and the right eye correction depth map d-mp_R_C provided from the right eye depth correction unit DMC_R Dimensional stereoscopic image data by rendering the left eye image data img_L and the right eye image data img_R with the left eye image data img_L and the right eye image data img_R.

Hereinafter, the left-eye depth map correcting unit and the right-eye depth map correcting unit of FIG. 1 will be described in more detail with reference to FIG.

FIG. 3 is a block diagram of the left-eye depth map correcting unit of FIG. 1. FIG.

First, the left eye depth map correction unit DMC_L and the right eye depth correction unit DMC_R are different from each other in that they receive different image data (left eye image data img_L or right eye image data img_R) Only the left eye depth map correction unit DMC_L will be described as an example. Unless otherwise specified, the depth map correction unit, the image data, the depth map, and the correction depth map, which will be described later, are each composed of the left-eye depth map correction unit DMC_L, the left- eye image data img_L, (d-mp_L) and the left-eye correction depth map (d-mp_L_C).

The depth map correcting unit DMC_L of the present invention receives the depth map from the depth map generating unit DMG and outputs the first histogram information hgrm_1 for the depth map, And sets the threshold frequency value T_frq based on the first histogram information hgrm_1 and corrects the depth map d_mp_L based on the threshold frequency value T_frq. For example, in the present invention, the depth map correcting unit DMC_L resets the gray level values (gray level values in the depth map) appearing at a frequency lower than the threshold frequency value T_frq described above to a different value, (d-mp_L) can be corrected.

As described above, the depth map generating unit DMG of the present invention sets one frequency value distributed on the Y axis of the histogram information as the threshold frequency value T_frq, and based on the threshold frequency value T_frq as a reference Since the depth map d-mp_L is corrected, it is possible to find all of the tone values that do not reach the threshold frequency value T_frq and to correct them all.

3, the depth map (d-mp_L) correcting unit DMC_L in the present invention includes a first histogram generating unit HGG1, a threshold frequency setting unit TFS, And may include a value classifying section GLC, a first order generating section CRU1, a second order generating section CRU2, a second histogram generating section HGG2 and a third order generating section CRU3, As follows.

The first histogram generator ( HGG1 )

The first histogram generator HGG1 receives the depth map d-mp_L from the depth map generator DMG and generates first histogram information for the depth map d-mp_L. This will be described in more detail with reference to FIGS. 4 and 5. FIG.

FIG. 4 is a view for explaining a depth map d-mp_L generated from the depth map generating unit DMG, and FIG. 5 is a diagram showing the first histogram information hgrm_1.

4 (a) shows the left eye image, and FIG. 4 (b) shows the right eye image data img_L constituting the left eye image and the right eye image data (d-mp_L) based on the img_R.

5 shows the first histogram information hgrm_1 generated based on the depth map d-mp_L of Fig. 4. The first histogram generator HGG1 generates the depth map d- and generates first histogram information hgrm_1 as shown in FIG. The X axis of the first histogram information hgrm_1 indicates the gradation, and the Y axis indicates the frequency. The first histogram information hgrm_1 in FIG. 5 shows the individual frequency numbers for each of the gray-level values included in the depth map d-mp_L in FIG. 4 (b). For example, each of the grayscale values included in the depth map d-mp_L of FIG. 4B is a grayscale value of 256 grayscale values from the darkest grayscale value 0 to the brightest grayscale value 255 It can be either. Referring to FIG. 5, it can be seen that within the depth map d-mp_L, the tone values near about 100 and the tone values near 125 are relatively more distributed than the other tone values.

The threshold frequency setting unit TFS )

The threshold frequency setting unit TFS receives the first histogram information hgrm_1 described above from the first histogram generation unit HGG1 and sets the threshold frequency value T_frq based on the first histogram information hgrm_1 do. Hereinafter, a method of setting the threshold frequency value T_frq will be described in detail.

A gradation value classifying section GLC )

The tone value classifying unit GLC receives the first histogram information hgrm_1 from the first histogram generating unit HGG1 and receives the threshold frequency value T_frq from the threshold frequency setting unit TFS. Based on the received first histogram information hgrm_1 and the threshold frequency value T_frq, the tone values included in the first histogram information hgrm_1 are converted into high frequency tone values, low frequency tone values, The gradation value, the maximum high frequency gradation value, the low region gradation values, the intermediate region gradation values, and the high region gradation values. This will be described in more detail with reference to FIG.

6 is a diagram for explaining a method of classifying gray level values in the first histogram information hgrm_1.

Referring to FIG. 6, the tone values in the first histogram information hgrm_1 are classified into high frequency tone values H1 and low frequency tone values H2 based on the threshold frequency value T_frq. That is, the tone values appearing at a frequency lower than the threshold frequency value T_frq are classified into the low frequency tone values H2, while the tone values appearing at a frequency higher than or equal to the threshold frequency value T_frq are classified into the high frequency tone Values H1. Here, the high frequency tone values H1 can be regarded as meaningful data sets necessary for the actual depth map d-mp_L, while the low frequency tone values H2 can be regarded as meaningful You can see it as a missing dataset. The high frequency tone values H1 and the low frequency tone values H2 can be expressed by Equation 1 below.

[ Equation 1 ]

H1 = {x | h (x) T_frg}

H2 = {x | h (x) < T_frg}

In Equation (1), x is a tone value and h (x) is a function representing a frequency value corresponding to the tone value.

On the other hand, the smallest tone value among the above-mentioned high frequency tone values H1 is classified as a minimum high frequency tone value G_min. That is, as shown in FIG. 6, the minimum high frequency gradation value G_min is a value obtained by subtracting the gradation value (T_frq) from the gradation value (T_frq) when the virtual horizontal line is parallel to the X axis The gradation value of the gradation bar having the smallest gradation value among the gradation bars of the high frequency gradation value) is determined as the minimum high frequency gradation value G_min. In other words, as shown in FIG. 6, the X-axis value of the coordinate P1 is defined as a minimum high frequency tone value (G_min).

Also, the highest tone value among the above-described high frequency tone values H1 is classified into a maximum high frequency tone value G_max. That is, as shown in FIG. 6, the maximum high frequency gradation value (G_max) is the maximum of the gradation bars (the gradation bars of the high frequency tone value) meeting with the imaginary horizontal line when there is the above described virtual horizontal line The gradation value of the gradation bar having the large gradation value is directly determined as the maximum high frequency gradation value (G_max). In other words, as shown in FIG. 6, the X-axis value of the coordinate P2 is defined as a minimum high frequency tone value (G_min).

6, the low frequency tone values smaller than the minimum high frequency tone value G_min are classified into the low region tone values G_L and are compared with the maximum high frequency tone value G_max described above The low frequency tone values are classified into the high region tone values G_H and the low frequency tone values located between the minimum high frequency tone value G_min and the maximum high frequency tone value G_max are classified into the middle region gradation value G_H, Values (G_M).

First Provincial Government ( CRU1 )

The first generating unit CRU1 receives the depth map d-mp_L from the depth map generating unit DMG and outputs the depth map d-mp_L to the gradation value classifying unit GLC To perform primary correction. In other words, the first transmission control unit (CRU1) refers to the tone values (i.e., low frequency tone values) determined to be error or noise by the threshold frequency value T_frq as described above, (Gradation values in the depth map d-mp_L) corresponding to the area and high region gradation values G_L and G_H are removed and the removed gradation values are changed to another meaningful gradation value.

Specifically, the primary correcting unit replaces all of the tone values of the depth map d-mp_L corresponding to the low region tone values G_L of the first histogram information hgrm_1 with the minimum high frequency tone value G_min And the depth map d-mp_L corresponding to the high region gradation values G_H of the first histogram information hgrm_1 is replaced with the maximum high frequency gradation value G_max, lt; / RTI &gt; An example of this will be described with reference to FIG.

Fig. 7 is a diagram for explaining a method of first-order correction of the depth map d-mp_L.

7A and 7B, gray level values for some pixels in the depth map d-mp_L are shown, and the numbers in the respective pixels mean gray level values of the corresponding pixels. Here, FIG. 7A shows some gray-scale values included in the original depth map d-mp_L, and FIG. 7B shows the original depth map d-mp_L shown in FIG. The results of the first calibration are shown.

At this time, assuming that the minimum high frequency tone value G_min is 100 and the maximum high frequency tone value G_max is 140, tone values smaller than 100 of the tone values shown in FIG. 7 (a) (b). On the other hand, gray scale values larger than 140 among the gray scale values shown in FIG. 7 (a) are all changed to 140 as shown in FIG. 7 (b).

In this way, the first beam controller CRU1 replaces all of the tone values (i.e., tone values in the depth map d-mp_L) corresponding to the low tone tone values G_L with the minimum high frequency tone value G_min And the gray level values corresponding to the high gray level values G_H (i.e., the gray level values in the depth map d-mp_L) are all replaced with the maximum high frequency gray level value G_max, A primary correction depth map d-mp_L_c1 is generated.

The second government ( CRU2 )

The second correction unit CRU2 receives the primary correction depth map d-mp_L_c1 from the primary correction unit CRU1 and outputs the primary correction depth map d-mp_L_c1 from the above-mentioned gray value classification unit GLC The second result is referred to. That is, the second transmission control unit (CRU2) refers to the tone values (i.e., low frequency tone values) determined to be error or noise by the threshold frequency value T_frq as described above, (Gradation values in the primary correction depth map d-mp_L_c1) corresponding to the area gradation values G_M and changes the removed gradation values to another meaningful gradation value.

Specifically, the secondary-deflecting unit (CRU2) outputs the gradation values of the primary correction depth map (d-mp_L_c1) corresponding to the intermediate gradation values (G_M) of the first histogram information (hgrm_1) (Gradation values in the primary correction depth map d-mp_L_c1).

As one example, the second correction unit CRU2 corrects one tone value (i.e., a tone value in the primary correction depth map d-mp_L_c1) corresponding to any one of the intermediate tone values G_M (The tone values in the primary correction depth map (d-mp_L_c1)) located spatially close to the one tone value and its surroundings are calculated, and the average value of the sum (I.e., the tone value in the primary correction depth map (d-mp_L_c1)) is replaced with the average value. A specific example of this will be described with reference to FIG.

8 is a diagram for explaining a method of secondary correction of the primary correction depth map d-mp_L_c1.

8A and 8B, gray level values for some pixels in the primary correction depth map d-mp_L_c1 are shown, and the numbers in each pixel mean the gray level values of the corresponding pixels. Here, FIG. 8A shows some gray-scale values included in the primary correction depth map d-mp_L_c1, and FIG. 8B shows the primary correction depth map d -mp_L_c1) obtained by the second correction.

Assuming that the minimum high frequency tone value G_min and the maximum high frequency tone value G_max are 100 and 140 and the intermediate tone values G_M are larger than 100 and smaller than 125, the gray level values corresponding to the intermediate gray level values, that is, the gray level values of 120 and 110 among the gray level values shown in a) are corrected based on the surrounding gray level values. For example, when the first window W1 having a size of 3 * 3 is set around the grayscale value 120, the grayscale values 120 and the grayscale values 128, 128, 130, 128, 130, 130, and 130) is 128, the tone value 120 is changed to 128 as shown in FIG. 8 (b). Similarly, when the second window W2 having the size of 3 * 3 is set around another tone value 110, the tone values 110 and the tone values in the second window W2 located in the vicinity of the tone value 110 130, 130, 130, 135, 100, 130, 130, 130 and 130 is 125, the gray level 120 is changed to 128 as shown in FIG. 8 (b). Here, the size of the first and second windows W1 and W2 is not limited to 3 * 3, and may be smaller or larger than the size.

Thus, the secondary controller CRU2 determines the gradation values (i.e., the gradation values in the depth map d-mp_L) corresponding to the intermediate gradation values G_M based on the adjacent gradation values spatially adjacent Quot ;, the secondary correction depth map d-mp_L_c2 is generated by this replacement operation.

On the other hand, errors and noise existing in the original depth map d-mp_L can be removed by the secondary correction alone. However, there is a possibility that the value of the minimum high frequency tone value G_min is set too small or the maximum high frequency tone value G_max is set too large depending on the image characteristics. As a result, the extremely low minimum high frequency tone value G_min, And an extremely high maximum high frequency tone value (G_max) are very likely to be noises. The present invention may further include a second histogram generator (HGG2) and a third generator (CRU3) capable of removing noise according to the image characteristics.

The second histogram generator ( HGG2 )

The second histogram generator HGG2 receives the secondary correction depth map d-mp_L_c2 from the secondary correcting unit CRU2 and outputs the second histogram information hgrm_2 to the secondary correction depth map d- . This will be described in more detail with reference to FIG.

Fig. 9 is a diagram showing the second histogram information hgrm_2.

According to the second histogram information hgrm_2 shown in FIG. 9, the frequency of a specific tone value (for example, a tone value of 100 and a tone value of 140) is smaller than the first histogram information hgrm_1 by a delta 1 (? 1) And delta 2 ([Delta] 2). This is because the gradation values in the depth map d-mp_L corresponding to some gradation values are replaced with the minimum high frequency gradation value G_min and the maximum high frequency gradation value G_max by the primary correction as described above This is because the frequency increases more. On the other hand, although not shown, the frequency of the tone values in the depth map d-mp_L corrected to the average value also appears in an increased state in the second histogram information hgrm_2 in Fig.

The Third Government ( CRU3 )

The tertiary information providing unit (CRU3) corrects the second histogram information hgrm_2 based on the preset lower limit blocking value and the upper limit blocking value. Further, the tertiary information providing unit (CRU3) sets the secondary correction depth map d-mp_L to 3 (3) based on the corrected second histogram information hgrm_2 and the threshold frequency value T_frq from the threshold frequency setting unit TFS Differential correction.

The lower limit cutoff value may be set to a tone value located in lower i% (i is a natural number smaller than 100) of the tone values included in the second histogram information hgrm_2, and the upper limit cutoff value may be set to the second histogram information hgrm_2 (J is a natural number smaller than 100) among the gray values included in the gray level value of the gray level value. For example, when the image data is 256 gradations, the lower limit cutoff value may be set to a gradation value of 40, and the upper limit cutoff value may be set to a gradation value of 215. [

The third order controller CRU3 classifies the tone values smaller than the lower limit cutoff value in the second histogram information hgrm_2 into the lower limit tone values and outputs the gradation values larger than the upper limit cutoff value in the second histogram information hgrm_2 Values are classified into upper limit excess tone values. The third transmission control unit CRU3 receives the corrected second histogram information hgrm_2 of the form in which the lower limit tone values and the upper limit tone values are removed and the threshold frequency value T_frq provided from the threshold frequency setting unit TFS (G_min) and the maximum high frequency tone value (G_max) on the basis of the maximum high frequency tone value (G_min) and the maximum high frequency tone value (G_max). Thereafter, the third transmission control unit (CRU3) transmits all of the tone values of the secondary correction depth map (d-mp_L_c2) corresponding to the lower limit tone values of the second histogram information hgrm_2 to the above- And the gray level values of the secondary correction depth map d-mp_L_c2 corresponding to the upper-limit excess gray level values of the second histogram information hgrm_2 are all reset to the maximum high frequency gray level value G_max, . By this replacement operation, a tertiary correction depth map d-mp_L_C is generated.

In this way, the tertiary information providing unit (CRU3) judges whether or not the secondary correction depth map (d-mp_L_c2) is finally corrected by using the lower limit cutoff value and the upper cutoff value, .

10 is a diagram for explaining a third-order correction method for the secondary correction depth map (d-mp_L_c2) that does not include the lower-limit gradation value and the upper-limit excess gradation values at all.

10 shows that the lower limit cutoff value CP_L and the upper limit cutoff value CP_H are applied to the second histogram information hgrm_2 of FIG. 9. As shown in the figure, the lower limit cutoff value CP_L and the upper limit cutoff value CP_H are applied to the second histogram information hgrm_2, (I.e., lower-limit over-gradation values) having no specific frequency among the gradation values that are higher than the upper-limit cut-off value CP_H ) Is not present at all. Therefore, the secondary correction depth map d-mp_L_c2 showing the same characteristics as the second histogram information hgrm_2 is outputted as it is through the third-order correction unit CRU3 without performing the third-order correction.

11 is a diagram showing a final correction depth map d-mp_L.

The final corrected depth map d-mp_L output through the third beam controller (CRU3) is as shown in Fig. That is, when the original depth map d-mp_L is corrected through the primary to the tertiary correction as described above, assuming that the drawing shown in FIG. 4B is the original depth map d-mp_L 11 shows the shape as shown in Fig.

In other words, the AA area AA of the original depth map d-mp_L shown in FIG. 4 (b) includes erroneous tone values generated by errors or noise, but as shown in FIG. 11 The erroneous tone values of the AA area AA are correctly corrected. This will be explained in more detail.

First, the A region A and the B region B in Fig. 4A correspond to the AA region AA and the BB region BB in Fig. 4 (b), respectively. As can be seen from Fig. 4 (a), the subject (part of the building) located in the area A is definitely behind the subject (car) located in the area B. However, as shown in FIG. 4 (b), the tone values of the AA area (AA (area corresponding to the A area) are higher than the gray values of the BB area (BB, area corresponding to the B area) Bright) can be known. This is because erroneous tone values have occurred in the area A of the original depth map d-mp_L in Fig. 4 (b). However, according to FIG. 11, it can be seen that the tone values of the AA area AA are significantly lowered (darkened) compared with the tone values of the AA area AA of FIG. 4 (b) have. That is, the tone values of the AA area AA were corrected correctly by the correction.

Next, the case where the correction is performed through the third transmission control unit (CRU3) will be described with reference to Figs. 12 to 14. Fig.

Figs. 12 to 14 are diagrams for explaining a third-order correction method for the secondary correction depth map d-mp_L including the lower-limit tone value.

12 (a) shows another original image, and Fig. 12 (b) shows a secondary correction depth map (d-mp_L_c2) for the image of Fig. 12 (a). The A region A and the B region B in Fig. 12A correspond to the AA region AA and the BB region BB in Fig. 12B, respectively.

As can be seen from Fig. 12 (a), the subject (a part of wallpaper) located in the area A is located farther behind the subject (part of the floor) located in the area B. However, as shown in FIG. 12 (b), it is found that the gray values of the AA area (area corresponding to the area A) and the gray values of the BB area (area corresponding to the BB area) . In other words, the gray level values of the BB area have a too dark value. This is because erroneous tone values (i.e., underexposure tone values) still exist in the BB area BB of the secondary correction depth map d-mp_L_c2 in FIG. 12 (b).

13 is a diagram showing second histogram information hgrm_2 for the secondary correction depth map d-mp_L_c2 in FIG. 12. As shown in the figure, the lower limit lower limit value CP_L G_nz) is present. The erroneous tone values in the BB area BB described above are generated by this lower limit tone value G_nz.

Therefore, as shown in FIG. 14, the minimum histogram value G_min and the minimum histogram G_min are calculated on the basis of the corrected histogram information and the above-described threshold frequency value T_frq, The maximum high frequency tone value G_max 'is reset. The method of setting the minimum high frequency tone value G_min 'and the maximum high frequency tone value G_max' is as described above.

Subsequently, all the tone values (i.e., the tone values in the secondary correction depth map d-mp_L) corresponding to the removed lower-limit tone value G_nz are replaced with the minimum-high tone tone values G_min ' do. Thus, a final corrected depth map d-mp_L_C is generated.

The final corrected depth map d-mp_L_C outputted through the third beam controller CRU3 is supplied to the stereoscopic image generating unit DVG.

Meanwhile, the threshold frequency value T_frq described above can be set in the following manner, which will be described in detail with reference to FIGS. 15 and 16. FIG.

FIG. 15 is a diagram showing accumulated histogram information for the first histogram information hgrm_1 in FIG. 5, and FIG. 16 is a diagram for explaining a method for setting a threshold frequency value T_frq from the first histogram information hgrm_1. to be.

The threshold frequency setting unit TFS generates the accumulated histogram information as shown in Fig. 15 based on the first histogram information hgrm_1 described above. The accumulated histogram information is generated by accumulating each tone value included in the first histogram information so as to include all of the frequency of all the tone values preceding (smaller) oneself.

The threshold frequency setting unit TFS sets the threshold frequency value T_frq based on the accumulated histogram information. Specifically, the threshold frequency setting unit TFS first differentiates the cumulative curve included in the cumulative histogram information from the lowest gradation value (0) to the highest gradation value (255) for each gradation value interval, and calculates a slope for each gradation interval do. Then, one of the grayscale values included in the grayscale section having the sharpest slope among the slopes is selected. For example, as shown in FIG. 15, if the interval between two adjacent gradation values G1 and G1 has a maximum slope, a lower gradation value G1 and an upper gradation value (&quot; G2 are selected.

16, the threshold frequency setting unit TFS finds a gradation value G1 equal to the gradation value G1 selected from the cumulative histogram from the gradation values in the first histogram information hgrm_1 . Then, the frequency value (the frequency value in the first histogram information hgrm_1) corresponding to the found tone value G1 is set as the threshold frequency value T_frq. For example, if the tone value selected from the accumulated histogram information is 115, the threshold frequency is determined as the frequency value corresponding to the 115 tone value in the first histogram information hgrm_1.

The cumulative curve of the above-described cumulative histogram information is generated by lining up cumulative tone values.

Meanwhile, the threshold frequency setting unit TFS may set the threshold frequency value T_frq in the following manner.

That is, the threshold frequency setting unit TFS sets a frequency value located at a lower k% (k is a natural number smaller than 100) among the frequency values included in the accumulated histogram information as a reference frequency value, and stores the frequency value in the accumulated histogram information The cumulative curve is subjected to first differentiation for each gradation value interval from the lowest gradation value (0) to the highest gradation value (255) to calculate a slope for each gradation period, and one of the gradation periods having the sharpest gradient One tone value is selected, and a frequency value corresponding to the selected tone value is compared with the reference frequency value. Thereafter, when the comparison result frequency value is larger than the reference frequency value, the same gradation value as the selected gradation value is searched from the gradation values in the first histogram information hgrm_1, and the first histogram information corresponding to the found gradation value (hgrm_1) is set as a threshold frequency value (T_frq).

On the other hand, when the comparison result indicates that the frequency value is not larger than the reference frequency value, the threshold frequency setting unit (TFS) sets the threshold value to be greater than the reference frequency value (For example, an upper gradation value) and a reference frequency value corresponding to another gradation value (for example, an upper gradation value) included in the gradation value or a gradation value of the next highest gradation value Or the lower limit gray level value) with the reference frequency value.

The above-described reference frequency value may be a frequency value located in the lower 1% of the frequency values included in the accumulated histogram information.

17 to 19 are diagrams for explaining the effect of the present invention.

The AA area in Fig. 17 means the AA area in Fig. 4 (B) described above. The AAA area in FIG. 17 is an enlarged view of an image (left eye image) created based on tone values (false tone values) of the AA area. As shown in FIG. 17, it can be seen that the image generated based on the original depth map d-mp_L before correction contains distorted information different from the original image.

The AA 'area in FIG. 18 means the AA area in FIG. 4 (B) described above. The AAA 'area in FIG. 18 is an enlarged view of the image (left eye image) created based on the tone values (false tone values) of the AA' area. As shown in FIG. 18, it can be seen that the image generated based on the corrected depth map d-mp_L_C contains the correct information for the original image.

FIG. 19 shows an original depth map d-mp_L and a corrected depth map d-mp_L_C for each of the two original images. As shown in Fig. 19, the corrected depth map d-mp_L in the present invention can substantially correct errors and noise of the original depth map d-mp_L.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the general inventive concept as defined by the appended claims and their equivalents. Will be clear to those who have knowledge of.

HGG1: first histogram generation section TFT: threshold frequency setting section
GLC: gradation value classifying unit CRU1: primary correcting unit
CRU2; Second secondary control unit CRU3: Third secondary control unit
HGG2: second histogram generator DMC_L: depth map corrector
d-mp_L: depth map hgrm_1: first histogram information
hgrm_2: second histogram information T_frq: threshold frequency value
d-mp_L_c1: primary correction depth map d-mp_L_c2: secondary correction depth map
d-mp_L_C: Final calibration depth map

Claims (26)

The method comprising: generating first histogram information for a depth map, setting a threshold frequency value based on the first histogram information, and resetting the tone values appearing at a frequency lower than the threshold frequency value to different values, And a depth map correction unit for correcting the depth map,
The depth-
A first histogram generator for generating first histogram information for the depth map;
A threshold frequency setting unit for setting a threshold frequency value based on first histogram information from the first histogram generation unit;
The histogram information included in the first histogram information based on the first histogram information from the first histogram generation unit and the threshold frequency value from the threshold frequency setting unit to the high frequency tone values, the low frequency tone values, A high frequency tone value, a maximum high frequency tone value, a low region tone value, an intermediate region tone value, and a high region tone value;
The gradation value of the depth map corresponding to the low region gradation values of the first histogram information is replaced with the minimum high frequency gradation value by referring to the classification result from the gradation value classifying portion, A first correcting unit for first correcting the depth map by replacing all the gray level values of the depth map corresponding to the high gray level values of the information with the maximum high frequency gray value; And
Based on the gradation values of the primary correction depth map corresponding to the intermediate region gradation values of the first histogram information, based on the gradation values spatially adjacent to each of the secondary gradation values of the first histogram information, And a second correction unit for correcting the depth map of the three-dimensional image. delete delete The method according to claim 1,
Wherein the tone value classifying unit classifies the tone values of the plurality of tone values based on the first histogram and the threshold frequency value,
Classifying the gradation values appearing at a frequency lower than the critical frequency value into low frequency gradation values and classifying the gradation values appearing at a frequency higher than or equal to the critical frequency value into high frequency gradation values, Frequency tone gradation value, and classifies the highest gradation value among the high frequency gradation values into a maximum high frequency gradation value, and classifies the lowest frequency gradation value smaller than the minimum high frequency gradation value Frequency gradation values, and classifying the low frequency gradation values that are larger than the maximum high frequency gradation value into high region gradation values, and dividing the low frequency gradation values into a high region gradation value, Wherein the frequency gradation values are classified into intermediate gradation values. The method according to claim 1,
When the second gradation correction unit corrects one gradation value in the primary correction depth map corresponding to any one of the intermediate area gradation values,
Calculating an average value of the total sum of all the gray level values in the first correction depth map located spatially close to the one gray level value and replacing the one gray level value with the calculated average value; An apparatus for correcting a depth map of a three - dimensional image. The method according to claim 1,
The threshold frequency setting unit may set,
And generating the accumulated histogram information based on the first histogram information and setting the threshold frequency value based on the accumulated histogram information. The method according to claim 6,
The threshold frequency setting unit may set,
Calculating a slope for each gradation section by first differentiating an accumulation curve included in the cumulative histogram information by a gradation value interval from a lowest gradation value to a highest gradation value;
Selects one of the gradation values included in the gradation period having the sharpest slope among the gradations;
From the grayscale values in the first histogram information, a grayscale value equal to the selected grayscale value; And,
And sets the frequency value in the first histogram information corresponding to the found gray level value as the threshold frequency value. The method according to claim 6,
The threshold frequency setting unit may set,
Setting a frequency value located at a lower k% (k is a natural number smaller than 100) among the frequency values included in the accumulated histogram information as a reference frequency value;
Calculating a slope for each gradation section by first differentiating an accumulation curve included in the cumulative histogram information by a gradation value interval from a lowest gradation value to a highest gradation value;
Selects one of the gradation values included in the gradation period having the sharpest slope among the gradations;
Comparing the frequency value corresponding to the selected gradation with the reference frequency value;
And if the frequency value is greater than the reference frequency value, searching for the same tone value as the selected tone value from the tone values in the first histogram information; And,
And sets the frequency value in the first histogram information corresponding to the found gray level value as the threshold frequency value. 9. The method of claim 8,
The threshold frequency setting unit may set,
And when it is determined that the frequency value is not larger than the reference frequency value, searching for another frequency value having a value larger than the reference frequency value, the frequency value corresponding to another one of the gradation values included in the gradation section having the sharpest slope Wherein the comparison unit compares the frequency value with the reference frequency value or compares the reference frequency value with a frequency value corresponding to any one of the tone values included in the gradation section having a sharp slope, Of the depth map. 9. The method of claim 8,
Wherein the reference frequency value is a frequency value located in the lower 1% of the frequency values included in the accumulated histogram information. The method according to claim 1,
The depth-
A second histogram generator for generating a second histogram based on a secondary correction depth map provided from the secondary controller; And
Correcting the second histogram information based on a preset lower limit cutoff value and an upper limit cutoff value and correcting the secondary correction depth map based on the corrected second histogram information and the threshold frequency value from the threshold frequency setting unit And a third correcting unit for correcting the depth map of the three-dimensional image. 12. The method of claim 11,
The lower limit cutoff value is a gradation value located at a lower i% (i is a natural number smaller than 100) among the gradation values included in the second histogram information, and the upper limit cutoff value is a gradation value contained in the second histogram information (J is a natural number smaller than 100) among the plurality of pixels of the three-dimensional image. 13. The method of claim 12,
The tertiary transmission /
Classifying the gradation values smaller than the lower limit cutoff value in the second histogram into lower limit gradation values;
Classify the gradation values larger than the upper limit cutoff value in the second histogram into the upper limit excess gradation values;
The minimum high frequency tone value and the maximum high frequency tone value are reset based on the corrected second histogram information in which the lower limit tone values and the upper limit tone values are removed and the threshold frequency value from the threshold frequency setting unit ;
Replacing all the tone values of the secondary correction depth map corresponding to the lower-limit tone values of the second histogram information with the reset minimum high-frequency tone value; And,
Dimensional depth map by replacing all of the tone values of the secondary correction depth map corresponding to the upper-limit excess tone values of the second histogram information with the reset maximum high frequency tone value, An apparatus for correcting depth map of an image. The method comprising: generating first histogram information for a depth map, setting a threshold frequency value based on the first histogram information, and resetting the tone values appearing at a frequency lower than the threshold frequency value to different values, A first step of correcting the depth map,
In the first step,
1-A step of generating first histogram information for the depth map;
A 1-B step of setting a threshold frequency value based on the first histogram information generated from the step 1-A;
Wherein the gray level values included in the first histogram information are calculated based on the first histogram information generated from the 1-A step and the threshold frequency value generated from the 1-B step to a minimum high frequency tone value, a maximum high frequency tone value, A 1-C step of classifying the image data into low region tone values, intermediate region tone values, and high region tone values;
The gradation values of the depth map corresponding to the low region gradation values of the first histogram information are all replaced with the minimum high frequency gradation value by referring to the classification result from the 1-C step, and the first histogram A 1-D step of first adjusting the depth map by replacing all the gray level values of the depth map corresponding to the high gray level values of the information with the maximum high frequency gray level value; And
Referring to the classification result from the 1-C step, the gradation values of the primary correction depth map corresponding to the intermediate region gradation values of the first histogram information may be referred to as secondary images based on the gradation values spatially adjacent to each other, Wherein the depth map comprises a 1-E step to determine a depth map of the three-dimensional image. delete delete 15. The method of claim 14,
The method of claim 1, wherein, in step 1-C, the first histogram and the threshold-
Classifying the gradation values appearing at a frequency lower than the critical frequency value into low frequency gradation values and classifying the gradation values appearing at a frequency higher than or equal to the critical frequency value into high frequency gradation values, Frequency tone gradation value, and classifies the highest gradation value among the high frequency gradation values into a maximum high frequency gradation value, and classifies the lowest frequency gradation value smaller than the minimum high frequency gradation value Frequency gradation values, and classifying the low frequency gradation values that are larger than the maximum high frequency gradation value into high region gradation values, and dividing the low frequency gradation values into a high region gradation value, Wherein the frequency gradation values are classified into intermediate gradation values. 15. The method of claim 14,
In the 1-E step, when any one of the intermediate area tone values is corrected,
Wherein the average value of the sum of all the gray-scale values located spatially close to any one of the intermediate gray-scale values and the periphery thereof is calculated and the one intermediate gray-scale value is replaced with the calculated average value. Correction method of depth map of 3D stereoscopic image. 15. The method of claim 14,
In the 1-B step,
And generating the accumulated histogram information based on the first histogram information and setting the threshold frequency value based on the accumulated histogram information. 20. The method of claim 19,
In the step 1-B,
A 1-B1 step of firstly differentiating an accumulated curve included in the cumulative histogram information by a gradation value section from a lowest gradation value to a highest gradation value to calculate a slope for each gradation section;
A 1-B2 step of selecting one of the grayscale values included in the grayscale section having the sharpest slope among the slopes;
1-B3 step of finding the same tone value as the selected tone value from the tone values in the first histogram information; And,
And a 1-B4 step of setting a frequency value in the first histogram information corresponding to the found gray value as the threshold frequency value. 20. The method of claim 19,
In the step 1-B,
A 1-B1 step of setting a frequency value located at a lower k% (k is a natural number smaller than 100) among the frequency values included in the accumulated histogram information as a reference frequency value;
A 1-B2 step of firstly differentiating an accumulated curve included in the cumulative histogram information from a lowest gradation value to a highest gradation value according to a gradation value interval to calculate a slope for each gradation interval;
A 1-B3 step of selecting one of the grayscale values included in the grayscale having the sharpest slope among the slopes;
A 1-B4 step of comparing the frequency value corresponding to the selected gradation with the reference frequency value;
A 1-B5 step of finding, from the tone values in the first histogram information, the same tone value as the selected tone value when the frequency value is greater than the reference frequency value; And,
And a 1-B6 step of setting a frequency value in the first histogram information corresponding to the found gray value as the threshold frequency value. 22. The method of claim 21,
In the step 1-B,
When it is determined that the frequency value is not greater than the reference frequency value in step 1-B4, in order to find another frequency value having a value larger than the reference frequency value, the other one included in the gradation section having the sharpest slope Which compares the frequency value corresponding to the gray level value of the first gray level value with the reference frequency value or compares the reference frequency value with a frequency value corresponding to any one gray level value included in the gray level section having a sharp gradient, The method of claim 1, further comprising a step B7. 22. The method of claim 21,
Wherein the reference frequency value is a frequency value located in the lower 1% of the frequency values included in the accumulated histogram information. 15. The method of claim 14,
In the first step,
A 1-F step of generating a second histogram based on the secondary correction depth map generated from the 1-E step; And
Correcting the second histogram information based on a preset upper limit cutoff value and a lower limit cutoff value and correcting the secondary correction depth map based on the corrected second histogram information and the threshold frequency value from the 1-B step And a 1-G step of correcting the depth map of the three-dimensional image. 25. The method of claim 24,
The lower limit cutoff value is a gradation value located at a lower i% (i is a natural number smaller than 100) among the gradation values included in the second histogram information, and the upper limit cutoff value is a gradation value contained in the second histogram information (J is a natural number smaller than 100) among the plurality of pixels of the three-dimensional image. 26. The method of claim 25,
In the step 1-G,
1-G1 step of classifying gradation values smaller than the lower limit cut-off value in the second histogram into lower-limit gradation values;
1-G2 step of classifying the tone values larger than the upper limit cut-off value in the second histogram into the upper limit tone values;
The minimum high frequency tone value and the maximum high frequency tone value are reset based on the corrected second histogram information in which the lower-limit tone values and the upper-limit excess tone values are removed and the threshold frequency value from the 1-B step 1-G3 step;
1-G4 step of replacing all the tone values of the secondary correction depth map corresponding to the lower-limit tone values of the second histogram information with the reset minimum-high-frequency tone values; And,
And a 1-G5 step of generating a third correction depth map by replacing all of the tone values of the secondary correction depth map corresponding to the upper-limit excess tone values of the second histogram information with the reset maximum high frequency tone value Dimensional depth map of the three-dimensional image.

Images