KR101629459B1 - Image processing apparatus and method - Google Patents

Abstract

An image processing apparatus is provided. The low-resolution depth image is matched to the color image and upsampled to a high resolution. In this case, the weight calculation section of the image processing apparatus calculates, for the first pixel of the input depth image, at least one of the periphery of the first pixel based on at least one of a difference in color value, a difference in depth value, Of the pixel of interest.

Depth image, Image registration, Depth camera, Depth image, Depth camera, Image matching

Description

[0001] IMAGE PROCESSING APPARATUS AND METHOD [0002]

The present invention relates to image processing using a depth image for obtaining three-dimensional shape information on an object, and more particularly relates to image processing such as resolution adjustment and precision improvement of a depth image.

The 3D (3D) information of the object is composed of geometry information and color information.

The color information is generated using a camera based on a CCD (Charged Coupled Device) or a CMOS (Complementary Metal Oxide Semiconductor) sensor.

On the other hand, the shape information is generated by using a 3D scanner using a laser, a depth camera based on an infrared ray, etc. Especially, the latter is getting more attention because of its simple mobility and low cost.

An infrared (IR) -based depth camera measures the distance from the camera to the object by measuring the time of flight (TOF) by irradiating the object with infrared light and detecting the reflected light, , Depth camera, depth image) is calculated.

On the other hand, the resolution of such a depth camera is significantly lower than the resolution of a CCD camera or a CMOS sensor based color camera due to technical limitations for implementation, increase in manufacturing cost, and the like.

There is provided an image processing apparatus and method that improves noise characteristics and reduces errors in a process of matching a low-resolution depth image with a high-resolution color image and increasing a resolution to generate a high-resolution depth image.

According to an aspect of the present invention, there is provided a method for correcting an image of a first pixel of an input depth image matched to an input color image, based on at least one of a color value difference, And a weighting unit for calculating a corrected depth value of the first pixel using a weight for each of the at least one pixel, wherein the weighting unit calculates a weight for each of the at least one pixel, Device is provided.

The weight calculation unit may include a first calculation unit that calculates a first weight of each of the at least one pixel based on a difference between a color value of the first pixel and a color value of each of the at least one pixel, A second calculation unit for calculating a respective second weight of the at least one pixel based on a difference between a depth value of the first pixel and a depth value of each of the at least one pixel, A third calculation unit for comparing the respective distances of one pixel and calculating respective third weights of the at least one pixel, and a second calculation unit for calculating, using the first weight, the second weight, and the third weight, And a fourth calculation unit for calculating a final weight for each of the at least one pixel.

In this case, as the difference between the color value of the first pixel and the color value of each of the at least one pixel increases, the first calculation unit determines that the first weight of each of the at least one pixel is exponentially decreasing The first weight is calculated.

The second calculation unit may calculate the second weight so that the second weight of each of the at least one pixel is exponentially decreased as the difference between the depth value of the first pixel and the depth value of each of the at least one pixel increases. And calculates the second weight.

On the other hand, the third calculator calculates the third weight so that the third weight of each of the at least one pixel is exponentially decreased as the distance between the first pixel and the at least one pixel increases, do.

The fourth calculator then calculates a final weight for each of the at least one pixel so as to be proportional to the product of the first weight, the second weight, and the third weight, The final weight is calculated such that the sum of the final weights for each is equal to one.

In this case, the weight applying unit may perform mask filtering on the depth value of the first pixel of the input depth image according to the final weight of each of the at least one pixel, The depth value can be calculated.

The image processing apparatus may further include an image matching unit that matches the input depth image with the input color image when the input depth image is not matched with the input color image.

According to another aspect of the present invention, there is provided an image processing method including: a step of calculating, based on at least one of a color value difference, a depth value difference and an inter-pixel distance for a first pixel of an input depth image matched to an input color image, Calculating a weight for each of at least one pixel of the at least one pixel and calculating a corrected depth value of the first pixel using a weight for each of the at least one pixel, / RTI >

According to embodiments of the present invention, when a low-resolution depth image is matched to a high-resolution color image and the resolution is increased to generate a high-resolution depth image, noise characteristics are improved, errors are reduced, and image processing quality is improved .

Hereinafter, some embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the present invention is not limited to or limited by the embodiments. Like reference symbols in the drawings denote like elements.

FIG. 1 shows an image processing apparatus 100 according to an embodiment of the present invention.

A high-resolution color image and a low-resolution depth image are input to the image processing apparatus 100.

In this case, if the input depth image is not already matched with the color image, the image matching unit 110 matches the input depth image with the input color image. The image registration process will be described later in detail with reference to FIG.

Then, up-sampling is performed on the input depth image matched to the input color image. In this case, the weight calculation unit 120 calculates weights of peripheral pixels for mask filtering on the first pixel of the input depth image .

Details of this weight calculation will be described later in more detail with reference to FIG. 2, FIG. 6 to FIG. 8.

Then, the weight applying unit 130 performs mask filtering of the first pixel using the calculated weights, and corrects the depth value of the first pixel.

This process will be described in more detail below with reference to FIGS. 8 and 12. FIG.

2 shows a weight calculation unit 120 of the image processing apparatus 100 according to an embodiment of the present invention.

The first calculation unit 210 calculates a first weight (hereinafter, also referred to as a color range weight) of surrounding pixels for mask filtering of the first pixel of the upsampled depth image . Calculation of the color range weight is described in more detail below with reference to FIG.

The second calculation unit 220 calculates a second weight (hereinafter, also referred to as a depth range weight) of surrounding pixels for mask filtering of the first pixel of the up-sampled depth image. The calculation of the depth range weight will be described in more detail below with reference to FIG.

Then, the third calculator 230 calculates a third weight (hereinafter also referred to as a distance weight or position weight) of surrounding pixels for mask filtering of the first pixel of the up-sampled depth image . The calculation of the distance weight is described in more detail below with reference to FIG.

The fourth calculator 240 calculates a final weight to be applied to neighboring pixels of the first pixel using the first weight to the third weight. The calculation of the final weights will be described in more detail below with reference to Figures 8 and 12.

FIG. 3 illustrates an input color image 300 according to an embodiment of the present invention.

The input color image 300 input to the image processing apparatus 100 has a high resolution. And the pixels of the color image 300 correspond to any one of the plurality of objects.

For example, some pixels correspond to the human body portion 310, some other pixels correspond to the wall portion 320, and some other pixels correspond to the bottom portion 330.

According to the conventional image segmentation, each pixel of the color image 300 can be divided into a plurality of pixels based on the color value analysis of the color image 300.

Correction of the depth value can be performed using the segmentation result of the color image 300 with respect to the depth image matched to the color image 300.

However, even if the color image 300 is matched with the depth image, due to the hardware performance limitation of the depth camera, the edge of each area in the color image 300 and the depth image may not be exactly the same have.

Accordingly, in the case of correcting the depth image according to the color value analysis of the color image 300, in particular, an inaccurate result can be obtained at the edge portion of the region.

According to an embodiment of the present invention, the image processing apparatus 100 upsamples the depth image to a high resolution and corrects the depth image to obtain a high quality result. In addition, the image processing apparatus 100 may not only calculate the color value of the color image 300, Corrects the depth image based on the depth value of the upsampled depth image and the distance difference between the pixels to be corrected.

More details will be described later with reference to FIG. 4 and the following.

FIG. 4 illustrates a low-resolution depth image 400 input according to an embodiment of the present invention.

Due to the hardware limitations of the depth camera described above, the input depth image 400 has a lower resolution than the color image 300.

In the following description, it is assumed that the depth image 400 is lower in resolution than the color image 300, but the two images are matched. That is, it is assumed that the point of view of the depth image 400 is the same as that of the color image 300, and the distance between the depth camera and the object is the same as that of the color camera.

However, if the depth image input to the image processing apparatus 100 is not matched to the color image 300, the depth image input by the image matching unit 110 of the image processing apparatus 100 is converted into a color image (300) is required.

In this image matching process, feature points are extracted for each of the color image 300 and a depth image (not shown) that is not matched with the color image 300, and a conversion function between the coordinates of the feature points or a conversion matrix , And applying the obtained transform function (or transformation matrix) to all the pixels of the input depth image.

However, the present invention is not limited to the partial image matching method, and the image matching between the input depth image and the color image may be performed by various methods. Therefore, as described above, an embodiment assuming that the depth image 400 matched with the color image 300 is input to the image processing apparatus 100 will be described below.

The input depth image 400 has a lower resolution than the color image 300. However, since the two images are matched with each other, even if the resolution of the input depth image 400 is upsampled to the resolution of the color image 300, such matching is maintained.

However, if no image processing is performed in the upsampling process, the quality of the upsampled depth image is very low.

Specifically, the edges of the body part 410, the wall part 420, and the bottom part 430 are blurred, and more distortion due to noise is observed.

FIG. 5 illustrates a depth image 500 generated by upsampling the depth image 400 of FIG. 4 and having the same resolution as the color image 300 of FIG. 3, according to an embodiment of the present invention.

The edges of the human body portion 510, the wall portion 520 and the bottom portion 530 of the depth image 500 are observed to be crushed and a portion 541 which is not distorted smoothly due to noise is also observed.

According to an embodiment of the present invention, this quality deterioration in the up-sampling process is improved through mask filtering.

The weight calculation unit 120 of the image processing apparatus 100 calculates the values of the elements of the mask for mask filtering, that is, the weights.

The mask for mask filtering is determined, for example, as a matrix of 3x3 or 5x5 size, and each element constituting this matrix is a weight for mask filtering.

For example, a mask of size 3x3 is used [0.1, 0.1, 0.1; 0.1, 0.2, 0.1; 0.1, 0.1, 0.1], the result of applying this mask is that a pixel value of eight pixels surrounding a specific pixel is given a weight of 0.1, a specific pixel itself is given a weight of 0.2, As the weighted sum of the nine pixel values multiplied by the weights.

The basic contents of the mask filtering are obvious to those skilled in the art of image processing, so that detailed description is omitted.

According to an embodiment of the present invention, in order to calculate weights, which are elements of a mask for correcting the depth value of each pixel of the depth image 500, the pixel value of the color image 300 corresponding to the depth image 500 The depth value of the depth image 500, and the inter-pixel distance are taken into account.

6 is a graph 600 illustrating the relationship between the difference in pixel color values and the first weight, in accordance with an embodiment of the present invention.

The first calculation unit 210 of the weight calculation unit 120 calculates the color value of the color image 300 having the same index as the first pixel and the color value of the surrounding pixels The color range weight is calculated based on the difference of the color range.

If the size of the mask is 5x5, the color range weights to be applied to the 24 pixels around the first pixel are calculated based on the difference from the color value of the first pixel itself.

According to an embodiment of the present invention, a color value of a first pixel having coordinates (i, j) is referred to as C (i, j) (M, n), the first weight applied to the neighboring pixels decreases exponentially as the difference between C (i, j) and C (m, n) increases.

For example, the following equation holds between the difference in color value and the first weight Wc (m, n).

[Equation 1]

는 칼라 값의 차이가 증가할수록 제1 가중치 Wc(m, n)가 지수적 으로 감소하는 정도를 조절하기 위한 상수이며, 실시예에 따라 다르게 설정될 수 있다.here

Is a constant for adjusting the degree to which the first weight Wc (m, n) decreases exponentially as the difference in color values increases, and may be set differently according to the embodiment.

In the above embodiment, the color values are represented by a single value and can be compared with each other by a method of using a distance difference on a color space of color values.

However, according to another embodiment of the present invention, the color value may be compared for each channel of R (Red), G (Green), and B (Blue). In this case, The final color range weight can be calculated by the arithmetic mean of each channel and then calculated.

This case can be understood by the following equation.

&Quot; (2) "

Figure 7 is a graph 700 illustrating the relationship between the difference in pixel depth values and the second weight, in accordance with an embodiment of the present invention.

The second calculation unit 220 of the weight calculation unit 120 calculates a depth range based on the difference between the depth values of the first pixel and the surrounding pixels for the first pixel in the depth image 500, Calculate the weights.

As described above, if the mask size is 5x5, the depth range weights to be applied to the 24 pixels around the first pixel are calculated based on the difference from the depth value of the first pixel itself.

According to an embodiment of the present invention, a depth value of a first pixel having coordinates (i, j) is referred to as D (i, j) (M, n), the second weight applied to the neighboring pixels decreases exponentially as the difference between D (i, j) and D (m, n) increases.

For example, the following equation holds between the difference in depth value and the second weight Wd (m, n).

&Quot; (3) "

는 깊이 값의 차이가 증가할수록 제2 가중치 Wd(m, n)가 지수적으로 감소하는 정도를 조절하기 위한 상수이며, 실시예에 따라 다르게 설정될 수 있다.here

Is a constant for adjusting the degree to which the second weight Wd (m, n) decreases exponentially as the difference in depth value increases, and may be set differently according to the embodiment.

8 is a graph 800 illustrating the relationship between the inter-pixel distance and the third weight, in accordance with an embodiment of the present invention.

The third calculation unit 230 of the weight calculation unit 120 calculates a distance weight for a first pixel in the depth image 500 based on the distance between the first pixel and the surrounding pixels.

As described above, if the size of the mask is 5x5, the distance weights to be applied to the 24 pixels around the first pixel are calculated based on the distance to the position of the first pixel itself.

According to an embodiment of the present invention, the coordinate value (i, j) and the coordinate value (m, n) of the coordinate (m, n) As the distance increases, the third weight applied to the surrounding pixels exponentially decreases.

For example, the following equation holds between the pixel-to-pixel distance and the third weight Wp (m, n).

&Quot; (4) "

는 픽셀 간의 거리가 증가할수록 제3 가중치 Wp(m, n)가 지수적으로 감소하는 정도를 조절하기 위한 상수이며, 실시예에 따라 다르게 설정될 수 있다.here

Is a constant for adjusting the degree to which the third weight Wp (m, n) decreases exponentially as the distance between the pixels increases, and may be set differently according to the embodiment.

When the first weight Wc (m, n), the second weight Wd (m, n) and the third weight Wp (m, n) are calculated as described above, the fourth calculator 240 of the weight calculator 120 calculates A final weight to be applied to each of the peripheral pixels of the first pixel in the depth image 500 is calculated.

According to one embodiment of the present invention, the final weight is determined to be proportional to the product of the first weight Wc (m, n), the second weight Wd (m, n) and the third weight Wp (m, n) In this case, the sum of the total final weights entering the mask filtering is normalized to be one.

The calculation of the final weight W _CDP (m, n) of the fourth calculator 240 is, for example, according to the following equation.

&Quot; (5) "

If the final weight W _CDP (m, n) is expressed simply as w (m, n), the weight applying unit 130 of the image processing apparatus 100 calculates the coordinates (i, j) The depth value of the position first pixel is corrected as follows.

&Quot; (6) "

으로 규준화되어 있음은 상기한 바와 같다.here,

As described above.

In order to facilitate the understanding of the above process, the process of applying the above process to the depth pixels indicated by the one-dimensional index and the matched color pixels separately from the depth image 500 is described with reference to FIGS. 11.

9 is a graph 900 illustrating a change in color value according to a pixel index, in accordance with an embodiment of the present invention.

The color value is changed from about 100 to about 0 according to the pixel index. In this case, the reduction of the color value is prominent in the portion where the pixel index is about 11 to about 15, and it can be understood that the boundary between the different portions (for example, the human portion and the wall portion in Fig. 3) have.

FIG. 10 is a graph 1000 illustrating a change in depth value according to a pixel index, in accordance with an embodiment of the present invention.

A change in the depth value from about 3 to about 5 is observed according to the pixel index. In this case, the increase in the depth value is prominent in the portion where the pixel index is from about 20 to about 25, and it can be understood that the boundary of different portions is located therebetween.

The pixel indices of FIG. 9 and FIG. 10 are matched, but it can be understood that the boundaries do not coincide with each other due to errors in the depth value acquisition and the upsampling process.

Further, in addition to the increasing portion 1020 of the depth value, it can be understood that the portions 1010 and 1020 where the depth value continuously changes are changed because the depth value is not constant due to noise.

11 is a graph 1100 showing a result of correcting the depth value of FIG. 10 by the image processing apparatus 100 according to an embodiment of the present invention.

The weight calculation unit 120 of the image processing apparatus 100 calculates a final weight according to the procedure described with reference to Equations 1 to 5 in consideration of a difference in color values, a difference in depth values, and a pixel distance .

The weight applying unit 130 of the image processing apparatus 100 performs mask filtering according to the final weight to correct the depth value graph 1000 in the form of a depth value graph 1100.

In the case of the partial boundary, no significant change has occurred, but it can be observed that the depth value noise of the pixel is greatly reduced as a whole. When applied to an actual depth image, the position of the boundary indicating different objects in the depth image is also modified, and the overall quality improvement is expected.

FIG. 12 shows a depth image 1200 generated by correcting the depth image 500 of FIG. 5 by the image processing apparatus 100 according to an embodiment of the present invention.

As described above with reference to Equations 1 to 5, the weight calculation unit 120 calculates weights to be applied to the respective elements of the mask filter, i.e., neighboring pixels, with respect to the first pixel of the depth image 500, The weighting unit 130 applies the mask filter to the first pixel of the depth image 500. [

When the calculation of the mask filter and the application of the mask filter are applied to all the pixels of the depth image 500, the corrected depth image 1200 is generated.

The improved portion 1241 of the portion 541 where an error of the edge is found in the depth image 500 is observed.

In addition to the improvement of the border error, it is possible to improve the blurring of the border line and to reduce the noise. Thus, each object portion 1210 to 1230 is expressed clearly and accurately.

13 illustrates an image processing method according to an embodiment of the present invention.

A high-resolution color image 300 and a low-resolution depth image 400 are input to the image processing apparatus 100.

In this case, as in the case of the input depth image 400 of FIG. 4, if the already matched color image 300 is input, step 1310 may be omitted.

However, if an unmatched depth image (not shown) is input, the depth image is matched to the color image 300 in step 1310, and the matched depth image 400 is input to the weight calculation unit 120 / RTI >

Then, up-sampling of the depth image 400 is performed to generate a depth image 500, and the weight calculation unit 120 performs mask filtering on the first pixel of the depth image 500 in step S1320 ≪ / RTI >

Details of this weight calculation are as described above with reference to Figs. 6 to 8, and Equations 1 to 5.

Then, in step S1330, the weight applying unit 130 performs mask filtering of the first pixel using the calculated weights, thereby correcting the depth value of the first pixel.

This process is as described above with reference to Equation (6).

FIG. 14 shows a weight calculation step (S1320) of an image processing method according to an embodiment of the present invention.

The color range weight of peripheral pixels for mask filtering of the first pixel of the depth image 500 is calculated by the first calculation unit 210 of the image processing apparatus 100 in step S1410. The calculation of the color range weight is as described above with reference to equations (1) to (2).

In step S1420, a depth range weight of peripheral pixels for mask filtering of the first pixel of the depth image 500 is calculated by the second calculation unit 220 of the image processing apparatus 100. [ The calculation of the depth range weight is as described above with reference to Equation (3).

In step S1430, a distance weight of neighboring pixels for mask filtering of the first pixel of the depth image 500 is calculated by the third calculation unit 230 of the image processing apparatus 100. [ The calculation of the distance weight is as described above with reference to Equation (4).

On the other hand, there is no precedence of the execution sequence of the steps S1410 to S1430, and can be performed simultaneously or sequentially.

When the color range weight, the depth range weight, and the distance weight to be applied to the neighboring pixels of the first pixel are calculated, the final weight to be applied to the neighboring pixels of the first pixel is calculated in step S1440, And is calculated by the fourth calculation unit 240. The calculation of the final weight is as described above with reference to Equation (5).

The process of correcting the depth value of the first pixel using the calculated final weight is as described above with reference to Equations (6) and (12).

The method according to an embodiment of the present invention can be implemented in the form of a program command which can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions recorded on the medium may be those specially designed and constructed for the present invention or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

While the invention has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by the equivalents of the claims, as well as the claims.

FIG. 1 illustrates an image processing apparatus according to an embodiment of the present invention.

2 shows a weight calculation unit of the image processing apparatus according to an embodiment of the present invention.

FIG. 3 illustrates an input color image according to an embodiment of the present invention.

FIG. 4 illustrates a low-resolution depth image input according to an embodiment of the present invention.

FIG. 5 illustrates a depth image generated by upsampling the depth image of FIG. 4 and having the same resolution as the color image of FIG. 3, according to an embodiment of the present invention.

6 is a graph showing a relationship between a difference between pixel values of a pixel and a first weight according to an embodiment of the present invention.

7 is a graph showing a relationship between a difference between depths of pixels and a second weight according to an embodiment of the present invention.

8 is a graph showing the relationship between the inter-pixel distance and the third weight according to an embodiment of the present invention.

9 is a graph showing a change in color value according to a pixel index according to an embodiment of the present invention.

10 is a graph showing a change in depth value according to a pixel index according to an embodiment of the present invention.

11 is a graph showing a result of correcting the depth value of FIG. 10 by the image processing apparatus according to an embodiment of the present invention.

FIG. 12 shows a depth image generated by correcting the depth image of FIG. 5 by the image processing apparatus according to an embodiment of the present invention.

13 illustrates an image processing method according to an embodiment of the present invention.

FIG. 14 shows weight calculation steps of an image processing method according to an embodiment of the present invention.

Claims (17)

For each first pixel of an input depth image matched to an input color image, based on at least one of a color value difference, a depth value difference, and an inter-pixel distance, at least one pixel in the vicinity of the first pixel A weight calculation unit for calculating a weight for the object; And A weight applying unit for calculating a corrected depth value of the first pixel using a weight for each of the at least one pixel, Lt; / RTI > The weight calculation unit may calculate, A first calculation unit for calculating a first weight of each of the at least one pixel based on a difference between a color value of the first pixel and a respective color value of the at least one pixel; A second calculation unit for calculating a second weight of each of the at least one pixel based on a difference between a depth value of the first pixel and a depth value of each of the at least one pixel; A third calculator for comparing the distance of each of the first pixel and the at least one pixel to calculate a third weight of each of the at least one pixel; And Calculating a final weight for each of the at least one pixel using the first weight, the second weight, and the third weight, And the image processing apparatus. delete The method according to claim 1, The first calculation unit may calculate, Calculating the first weight so that each first weight of the at least one pixel is exponentially reduced as the difference between the color value of the first pixel and the respective color value of the at least one pixel increases, Image processing apparatus. The method according to claim 1, The second calculation unit calculates, Calculating the second weight so that each second weight of the at least one pixel is exponentially reduced as the difference between the depth value of the first pixel and the depth value of each of the at least one pixel increases, Image processing apparatus. The method according to claim 1, The third calculation unit calculates, Wherein each third weight of the at least one pixel is exponentially decremented as the distance of each of the first pixel and the at least one pixel increases. The method according to claim 1, The fourth calculation unit calculates, Calculating a final weight for each of the at least one pixel such that the final weight is proportional to a product of the first weight, the second weight, and the third weight, and wherein the sum of the final weights for each of the at least one pixel is 1, < / RTI > The method according to claim 1, Wherein the weight applying unit comprises: And performing a mask filtering on the depth value of the first pixel of the input depth image according to a final weight for each of the at least one pixel to calculate a corrected depth value of the first pixel, Processing device. The method according to claim 1, And an image matching unit for matching the input depth image with the input color image when the input depth image is not matched with the input color image. For each first pixel of an input depth image matched to an input color image, based on at least one of a color value difference, a depth value difference, and an inter-pixel distance, at least one pixel in the vicinity of the first pixel Calculating a weight for the object; And Calculating a corrected depth value of the first pixel using a weight for each of the at least one pixel Lt; / RTI > The step of calculating the weight includes: Calculating each first weight of the at least one pixel based on a difference between a color value of the first pixel and a respective color value of the at least one pixel; Calculating a respective second weight of the at least one pixel based on a difference between a depth value of the first pixel and a depth value of each of the at least one pixel; Comparing respective distances of the first pixel and the at least one pixel to calculate a respective third weight of the at least one pixel; And Calculating a final weight for each of the at least one pixel using the first weight, the second weight, and the third weight, And an image processing method. delete 10. The method of claim 9, Wherein the step of calculating the first weight comprises: Calculating the first weight so that each first weight of the at least one pixel is exponentially reduced as the difference between the color value of the first pixel and the respective color value of the at least one pixel increases, Image processing method. 10. The method of claim 9, Wherein the calculating the second weight comprises: Calculating the second weight so that each second weight of the at least one pixel is exponentially reduced as the difference between the depth value of the first pixel and the depth value of each of the at least one pixel increases, Image processing method. 10. The method of claim 9, Wherein the step of calculating the third weight comprises: Wherein the third weight is computed such that each third weight of the at least one pixel is exponentially reduced as the distance of each of the first pixel and the at least one pixel increases. 10. The method of claim 9, Wherein calculating the final weight comprises: Calculating a final weight for each of the at least one pixel such that the final weight is proportional to a product of the first weight, the second weight, and the third weight, and wherein the sum of the final weights for each of the at least one pixel is 1, < / RTI > 10. The method of claim 9, Wherein calculating the corrected depth value of the first pixel comprises: Performing mask filtering on a depth value of a first pixel of the input depth image according to a final weight for each of the at least one pixel to calculate a corrected depth value of the first pixel. 10. The method of claim 9, Further comprising the step of matching the input depth image to the input color image prior to the weighting step if the input depth image is not matched to the input color image. A computer-readable recording medium storing a program for performing the image processing method of any one of claims 9 and 11 to 16.

Images