KR20100000951A - Method and apparatus for converting stereoscopic image signals into monoscopic image signals - Google Patents

Abstract

PURPOSE: A method and an apparatus for converting stereoscopic image signals into monoscopic image signals are provided to realize an image in an existing 2D display without distortion of picture quality using the stereoscopic image signals, thereby activating stereoscopic contents. CONSTITUTION: Unmatched areas are specified from a reduced left image and a reduced right image(S10). Boundary information is extracted from the unmatched areas(S20). One or both between the reduced left image and the reduced right image are all corrected using the boundary information(S30). A 2D image is generated using the corrected image(S40).

Description

Method and apparatus for converting stereoscopic video signal to 2D video signal {Method and apparatus for converting stereoscopic image signals into monoscopic image signals}

The present invention relates to a procedure and apparatus for converting a video signal, and more particularly, to a procedure and apparatus for converting a stereoscopic video signal into a monoscopic video signal.

Although 3D image media is currently limitedly applied to special fields such as medical, game, and CAD, it is expected to grow rapidly as a next generation information communication media for display devices such as mobile devices and TVs in the future. Such 3D imaging technology is largely divided into stereoscopic and holographic methods. The stereoscopic method is a three-dimensional image expressing a sense of depth by using binocular parallax. The stereoscopic method provides a stereoscopic effect by directly wearing a secondary device such as glasses or by using a flat panel display device with special auxiliary means. In contrast, the holographic method implements a 3D stereoscopic image by optically projecting a 3D waveform of an object into the air. At present, the stereoscopic method is technically simple and inexpensive compared to the holographic method, and thus is more widely used as a 3D video media technology.

In the stereoscopic method, a pair of left and right images obtained by photographing the same subject with a left camera and a right camera spaced by a certain distance are used for a stereoscopic image. The left image and the right image are taken of the same subject, but because of different viewpoints, the image information may be different depending on the surface characteristics of the subject, the position of the light source, and the like.

1 is a diagram illustrating a process of obtaining a stereoscopic image. As shown in FIG. 1, the stereoscopic image captures the same subject 12 using the left camera 14a and the right camera 14b spaced apart by a predetermined distance. As a result of the shooting, the left image 16a is obtained through the left camera 14a, and the right image 16b is obtained through the right camera 14b.

2 is a view for explaining characteristics of a left image and a right image constituting a stereoscopic image. As described above, although the left image and the right image are taken of the same subject, differences in image information occur because of different viewpoints. As shown in FIG. 2, the image and the right image of the pentagonal body included in the left image are illustrated. It can be seen that the images of the pentagram contained in are not identical.

This stereoscopic image is used to implement a three-dimensional effect on the display. Using stereoscopic images, there are many ways to add stereoscopic effect to images reproduced on flat display devices such as liquid crystal displays (LCDs) and plasma display panels (PDPs). The method is to use a barrier type display device.

FIG. 3 is a diagram for describing a principle of reproducing a stereoscopic image using a barrier type display.

Referring to FIG. 3, (a) of FIG. 3 extracts only the vertical vertical lines of the left image 30a from the stereoscopic image and arranges them as odd vertical lines, and extracts only the even vertical lines of the right image 30b to make even vertical lines. Figure shows the integrated composite image 32 arranged in a line. 3B illustrates a process of viewing a stereoscopic image by mounting a barrier 34 formed of a polarizing film or polarizing glass on the flat panel display 36 on which the integrated composite image 32 is displayed. The barrier polarizer is provided with a line-type barrier pattern, which shows only the left image portion of the display image in the left eye and only the right image portion of the display image in the right eye. Therefore, since the viewer sees only the left image through the left eye and only the right image through the right eye, it is possible to view a three-dimensional image with the naked eye without the aid of an auxiliary device such as stereoscopic glasses.

As such, the content used in the stereoscopic method, that is, the stereoscopic video signal is composed of image information different from the conventional monoscopic video signal. However, at present, the diffusion of a barrier type display device for reproducing stereoscopic video signals is not enough, and most people use a general 2D display device. When the stereoscopic video signal is reproduced on a general 2D display device as it is, the image quality is not good, so the use range of the stereoscopic video content is quite limited, and as a result, the production shrinkage of the stereoscopic video content and the dissemination of the stereoscopic video display device accordingly. It can lead to a vicious cycle of degradation.

Therefore, for the development of stereoscopic video related technologies and widespread use of stereoscopic video display devices, a method of enabling stereoscopic video signals to be used in 2D video display devices is also required. As a method for reproducing stereoscopic images through existing two-dimensional image receivers (TVs, monitors), one of the reduced left image and the reduced right image constituting the stereoscopic image signal is enlarged and reproduced, or the reduced left image is doubled. A method of reproducing an integrated composite video (that is, a stereoscopic video signal) in which an image and a reduced right image are alternately arranged may be considered. However, in the former case, the image quality is not good because the resolution of the image is low, and in the latter case, the image quality is deteriorated due to the mismatch between the left and right images.

One problem to be solved by the present invention is a stereoscopic video signal conversion process and conversion apparatus that can reproduce the image without deterioration or distortion of the image quality even on a two-dimensional display using a stereoscopic video signal To provide.

One embodiment of the present invention for solving the above problems is a method for converting a stereoscopic video signal consisting of a reduced left image and a reduced right image to a monoscopic video signal, comparing the reduced left image and the reduced right image. Specifying an inconsistency region in each of the reduced left image and the reduced right image, extracting boundary information from each of the inconsistency regions of the reduced left image and the reduced right image, and using the boundary information of the two images. Correcting one or both images of the reduced left image and the reduced right image, and generating a 2D image using the corrected reduced left image and / or the corrected reduced right image.

According to an aspect of the embodiment, in the correcting step, the reduced left image and / or the reduced right image may be corrected such that an average value of boundary information of the two images becomes a boundary of the corrected image.

According to another aspect of the embodiment, the correcting step may be corrected by enlarging or reducing the mismatched region of the remaining images while referring to one of the reduced left image and the reduced right image.

According to another aspect of the embodiment, the correction step may use the camera parameter information included in the stereoscopic video signal.

Another embodiment of the present invention for solving the above problems is a method for converting a stereoscopic video signal into a monoscopic video signal, the stereoscopic video signal is used to obtain image information of the left and right images and the left and right images At least the camera parameter information may be used to generate the monoscopic video signal by including camera parameter information and correcting one or two images from the left and right images.

According to an aspect of the embodiment, a boundary line may be extracted from all or part of one or two images from the left and right images, and the extracted boundary line may be corrected using the camera parameter information.

According to another aspect of the embodiment, the monoscopic video signal may be an image including a subject image at a viewpoint at an intermediate position of each viewpoint of the left and right images.

Another embodiment of the present invention for solving the above problems is a method for converting a stereoscopic video signal consisting of a reduced left image and a reduced right image to a monoscopic video signal, each of the reduced left image and the reduced right image An image of an intermediate view of the reduced left image and the reduced right image is generated by using boundary information extracted from all or part of.

According to an aspect of the embodiment, the boundary information may be extracted only in an inconsistent region of the reduced left image and the reduced right image.

Another embodiment of the present invention for achieving the above object is a device for converting a stereoscopic video signal consisting of a reduced left image and a reduced right image to a monoscopic video signal, the reduced left image and the reduced right image A unit for specifying an inconsistency region in each of the reduced left image and the reduced right image, a unit for extracting boundary information from each of the inconsistency regions of the reduced left image and the reduced right image, and boundary information of the two images A unit for correcting one or both images of the reduced left image and the reduced right image by using; and a unit for generating a two-dimensional image using the corrected reduced left image and / or the corrected reduced right image. can do.

According to the above-described embodiment of the present invention, the stereoscopic video signal can be reproduced without distortion of the image quality in the existing 2D display. Therefore, the stereoscopic video signal can be compatible with the 2D display to expand the selection of stereoscopic broadcast providers, thereby contributing to the activation of stereoscopic content.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

The 'stereoscopic video signal' according to the embodiment of the present invention described below does not include both left and right image information acquired through the left camera and the right camera, respectively. In the present embodiment, the 'stereoscopic video signal' includes only data converted from the left image and right image information to be reproduced to a stereoscopic display apparatus such as a barrier type display apparatus. The converted data may be composed of 'image data' and 'meta data'.

Accordingly, the 'stereoscopic video signal' includes 'zoom left image' extracted from the left or right in a horizontal or vertical direction as image data and 'zoom right image' extracted from a left or right in a horizontal or vertical direction. In addition, the 'stereoscopic video signal' includes metadata for decoding and playing back image information of an encoded stereoscopic video, for example, camera parameter information, which is information about left and right cameras used to acquire the left and right images. do. The information about the left and right cameras may include, for example, disparity information between the left and right images, information about a distance between the left and right cameras, information about a frame rate of the left and right images, and / or information about synchronization between the left and right images. And the like.

In the embodiment of the present invention, there is no limitation as to a method for generating a stereoscopic video signal from a left image and a right image. The stereoscopic video signal may not be generated from the left video and the right video, but may be a stereoscopic video signal obtained by converting the monoscopic video signal according to a predetermined conversion algorithm. Hereinafter, a process of acquiring a stereoscopic video signal will first be described before explaining a conversion procedure according to an embodiment of the present invention.

In the encoding apparatus, for example, the left and right images are each reduced to an image having a resolution of 1/2, and then the left and right reduced images are combined into a side-by-side format to form a separated composite image. By encoding the synthesized image, a stereoscopic video signal may be generated. Alternatively, a stereoscopic video signal may be generated by generating an integrated synthesis image by encoding the integrated synthesis image. In any case, there is no difference in that the stereoscopic video signal includes metadata along with the reduced left image and the reduced right image data.

Here, the "separated composite image" refers to a single image generated by arranging the reduced left image and the reduced right image side by side or up and down. On the other hand, the 'integrated composite image' refers to a single image in which the odd (or even) vertical lines extracted from the left image and the even (or odd) vertical lines extracted from the right image are alternately arranged. As a result, the integrated composite image corresponds to an image format finally reproduced in the stereoscopic image display apparatus. In order to reproduce the separated composite image on the stereoscopic display apparatus, a process of first converting the format into the integrated composite image is required. On the contrary, by converting the format of the integrated composite video, a separate composite video can be generated. Hereinafter, a process of generating a monoscopic video signal using a 'stereoscopic video signal' composed of a separated composite image will be described.

The stereoscopic video signal may be an image in which a reduced left image and a reduced right image are combined in a vertical direction (top-down format). In this case, the separated composite image is an image in which the set of odd-numbered horizontal lines extracted from the left image and the set of even-numbered horizontal lines extracted from the right image are arranged up and down, and the integrated composite image is the horizontal line of the reduced left image and the reduced right image. The images are arranged in alternating manner (hereinafter, referred to as "type 2 integrated synthesis video"). The stereoscopic video signal in the vertical direction is different from the above-described stereoscopic video signal in the horizontal direction only. Hereinafter, the stereoscopic video signal in the horizontal direction will be described.

4 is a diagram illustrating a separated composite image arranged in a side by side manner as an example of a stereoscopic image.

Referring to FIG. 4, a left image 40a having a predetermined size acquired through the left camera, for example, a 320x240 size, extracts only an odd number of vertical lines to form a reduced left image 42a having a size of 160x240, and is obtained through a right camera. The right image 40b having a predetermined size, for example, 320x240, extracts only the even-numbered vertical lines to form a reduced right image 42b which is also 160x240. Then, a separated composite image 44 having a 320x240 size is formed by combining the left and right reduced images 40a and 40b having the 160x240 size in the horizontal direction. In addition, the encoding apparatus encodes the separated composite image 44 according to a predetermined method, but the embodiment of the present invention described later has no relation to the encoding method of the separated composite image 44.

5 is a block diagram schematically illustrating an environment to which an embodiment of the present invention to be described below may be applied. In FIG. 5, the stereoscopic image acquisition unit generates a stereoscopic image signal having a format of a separated composite image, and uses the generated stereoscopic image signal through a predetermined transmission network such as the Internet, a wireless communication network, or a broadcasting network. The case of sending to is shown. The user may own a stereoscopic display device or a monoscopic display device.

Referring to FIG. 5, the left image signal 50a and the right image signal 50b are input to the stereoscopic image acquisition unit 52 to extract only the odd vertical lines of the left image to reduce the reduced left image 54a. In addition, only the even-numbered vertical lines of the right image are extracted to form a reduced right image 54b. The left-right composite image 54 of a side-by-side method combining the left and right downscaled images in a horizontal direction may be encoded and then transmitted through a normal transmission channel 56. Then, the image received by the stereoscopic display 59 may be reconstructed into an integrated composite image in which the vertical pixel lines of the left and right images are alternately arranged again through a format converter to be reproduced as a stereoscopic image. Such stereoscopic images can be viewed and worn with polarized glasses on a TV equipped with a polarizing film, and can be viewed with a shutter glasses through a shutter method, and can be viewed in various other 3D display methods. In addition, the image received by the monoscopic display 58 may be converted into a 2D image signal and reproduced by the 3D / 2D conversion apparatus according to the embodiment of the present invention.

FIG. 6 is a block diagram schematically illustrating an environment to which an embodiment of the present invention to be described below may be applied, which is different from that of FIG. 5 in that a stereoscopic image is configured in a top-down manner. In FIG. 6, the stereoscopic image acquisition unit generates a stereoscopic image signal having a format of a separated composite image, and uses the generated stereoscopic image signal through a predetermined transmission network such as the Internet, a wireless communication network, or a broadcasting network. The case of sending to is shown. The user may own a stereoscopic display device or a monoscopic display device.

Referring to FIG. 6, the left image signal 60a and the right image signal 60b are input to the stereoscopic image acquisition unit 62 to extract only the odd horizontal lines of the left image to reduce the reduced left image 64a. And extracts only the even horizontal lines of the right image to form a reduced right image 64b. The left and right reduced images of the left and right images combined in the vertical direction are combined with each other, and then encoded, and then transmitted through a normal transmission channel 66. Then, the image received by the stereoscopic display 69 may be reconstructed into an integrated composite image in which horizontal pixel lines of the left and right images are alternately arranged again through a format converter to be reproduced as a stereoscopic image. Such stereoscopic images can be viewed and worn with polarized glasses on a TV equipped with a polarizing film, and can be viewed with a shutter glasses through a shutter method, and can be viewed in various other 3D display methods. In addition, the image received by the monoscopic display 68 may be converted into a two-dimensional image signal through a 3D / 2D conversion apparatus according to an embodiment of the present invention and reproduced.

7 is a flowchart illustrating a procedure of converting a stereoscopic video signal into a 2D video signal according to an embodiment of the present invention.

Referring to FIG. 7, first, a reduced left image and a reduced right image constituting a stereoscopic video signal are compared to extract a mismatched region of both images (S10). The process of extracting the inconsistency region may be performed in units of blocks of a predetermined size or in units of individual pixels, and the coincidence region of both images using data of the reduced left image and the reduced right image (image data and / or metadata). It can be regarded as an example of the process of distinguishing the region from the inconsistency. In the case of distinguishing the matched region from the mismatched region in block units, for example, as shown in FIG. 8 (e) and FIG. have. Here, the unit block may be a macroblock or a smaller block and may be a basic unit block of encoding in video encoding.

According to this embodiment, there is no limitation in the specific method of distinguishing the matching area and the inconsistent area. The 'unmatched region' refers to an area where the left and right images are not identical, whereas the 'matching region' refers to an area where the left and right images are the same. For example, a pixel or an area (a block of a predetermined size) in which there is a difference between the image information of the reduced left image and the image information of the reduced right image or the difference is greater than or equal to a predetermined threshold may be referred to as an inconsistency region. The image information may be represented by luminance and color difference information such as Y, Cb, Cr, or RGB information. In contrast, a pixel or an area in which there is no difference between the image information of the reduced left image and the image information of the reduced right image or the difference is less than or equal to a predetermined threshold may be referred to as a matching area.

Generally, there is little parallax between left and right images for a subject that is far from the camera. On the other hand, the object near the distance from the camera generates a parallax of the left and right images, which is inversely proportional to the distance from the camera. Therefore, the pixel or region extracted as the matching region in step S11 is likely to correspond to a background screen far from the camera, and the mismatched region is likely to correspond to an object or a person near the camera.

Using this principle, the process of distinguishing the matched region from the mismatched region may be obtained by using a depth map between left and right images. For example, a pixel or an area where disparity is large due to a depth map may be a mismatched area, and a pixel or an area where a disparity is small may be divided into a matching area.

(A) is a picture showing a left image of the stereoscopic image, (b) is a picture showing a right image, (c) is a picture consisting of a side-by-side separated composite image, (d) is (c) ), (E) shows a matching region partitioned into unit blocks (N x M), which is a coding unit of the image, in (d). In (D), the inconsistency region divided by the unit block (N x M) which is the coding unit of the image is shown.

Referring to (c) of FIG. 8, only the odd vertical lines are extracted with the left image 80a to form a reduced left image 82a having a half size in the horizontal axis. Then, only the even vertical lines are extracted with the right image 80b to form a reduced right image 82b having a half size of the horizontal axis. For example, if the size of the left and right images shown in (A) and (B) is 320x240, respectively, the reduced left image 82a is a 160x240 image created by extracting only odd vertical lines from the image of (A). The right image 82b is a 160x240 image made by extracting only even vertical lines from the image of (b). This method can extract only the even vertical lines with the left image and the odd vertical lines with the right image according to the convenience of the method. It can be configured differently depending on the convenience of configuring the system.

FIG. 8D illustrates a combined composite image 84 having a reduced left image 82a and a reduced right image 52b. As described above, the stereoscopic image can be viewed through the 3D display with the integrated synthetic image as shown in (d) of FIG. But when you see the image on a 2D display. In some areas, the normal two-dimensional image may not be visible, such as a double boundary of the subject.

A part corresponding to (e) of FIG. 8 is a part of the background screen and is located far from the camera. Such a part is likely to correspond to a matching area where there is no difference between the image information of the reduced left image and the reduced right image. As described above, the coincidence region may be an area in which there is little difference in image information or less than a predetermined threshold value between adjacent pixels or blocks in the integrated composite image. Even if the matched area is played back using the 2D display, the image quality is not bad and the boundary does not appear twice.

A part corresponding to (bar) of FIG. 8 is a part of the subject, not a background, and is located at a relatively close distance from the camera. Such a part is likely to correspond to an inconsistency region where there is a difference between the image information of the reduced left image and the reduced right image. As described above, the mismatched region may be a region in which there is a difference in image information or a predetermined threshold or more between adjacent pixels or blocks in the integrated composite image. When the discrepancy area is reproduced by using a 2D display, the boundary may be doubled, and the image information may vary depending on the line, and stripes may appear on the screen.

In step S10 described above, the matched region and the mismatched region may be distinguished using the integrated composite image. More specifically, by comparing the image information between pixels or blocks (horizontal or vertical directions, such as the composition of the integrated composite image) in the integrated composite image, that is, by using the difference in the image information of the adjacent pixels or blocks In each image, a matched area and a mismatched area may be distinguished.

As described above, in the step of extracting the inconsistency region from the reduced left image and the reduced right image (S10), a method of distinguishing the matched region and the mismatched region of the reduced left image and the reduced right image may be performed by one pixel unit. It may also be carried out in unit blocks (N × M). Preferably, it can be carried out in unit blocks (N × M). In the method of extracting an inconsistency region by one pixel unit, the 'inconsistency region' is an image information between neighboring pixels such as Y (luminance), Cb, Cr (color difference) or RGB (red, green, blue) of each pixel. ) May be the point where the value changes abruptly. In addition, in the method of classifying the inconsistency region by the unit block (N × M), the 'inconsistency region' is, for example, 16x8 block, 8x16 block, 8x8 block, 8x4 block, 4x8 block, or 4x4 block. It may be a block in which Y (luminance), Cb, Cr (color difference) or RGB (red, green, blue) values change rapidly.

In the step of extracting such a mismatched region (S10), when generating an integrated composite image from the reduced left image and the reduced right image, a predetermined threshold α for a change value of image information of adjacent pixels or blocks is set. Pixels or blocks below the threshold α may be defined as matching regions, and pixels or blocks above the threshold may be defined as inconsistent regions so as to extract only the mismatched regions. There is no particular limitation for setting the threshold value α, and the threshold value may be set by applying an average value according to the case where the ratio of the distant background screen in the image or frame is large or small.

Subsequently, referring to FIG. 7, edges are extracted from an inconsistency area between the reduced left image and the reduced right image extracted in step S10 (S20). The reason for extracting the boundary in this step is to correct an inconsistency region of both images using the extracted boundary. In general, when the boundary is changed, a large change in the image information occurs, but there is a high possibility that the image information located within the same boundary does not greatly differ. When the boundary is not clear in the image, it can be recognized that the image quality is lowered by that much. In the embodiment of the present invention, the image quality can be prevented from appearing clearly by displaying the boundary clearly.

In an embodiment of the present invention, in order to analyze image characteristics of the extracted mismatched region, a boundary between objects is preferably used. It is apparent to those skilled in the art that a boundary exists between all objects and objects, and a large discrepancy between left and right images occurs at this boundary.

Here, the boundary may be a point at which a pixel value between neighboring pixels or an average value of pixel values between neighboring blocks changes rapidly. In the embodiment of the present invention, there is no particular limitation on the method for obtaining the boundaries of the reduced left image and the reduced right image, for example, a boundary detection mask may be used. As a boundary detection mask, a Sobel Mask, a Prewitt Mask, etc. are used.

FIG. 9A illustrates a Sobel mask, FIG. 9B illustrates an image to which the Sobel mask of FIG. 9A is applied, and FIG. 9C illustrates a detection result of applying the Sobel mask of FIG. 9A to the image of FIG. 9B. When the boundary of the image of FIG. 9B is detected using the Sobel mask of FIG. 9A, the result as shown in FIG. 9C may be obtained. Here, the boundaries may be represented by a vertical position information value and a horizontal position information value at which the detected boundaries are located in the screen.

7, at least one of the reduced left image and the reduced right image is corrected using boundary information extracted from the disagreement region (S30). This process is to obtain a single monoscopic video signal from the stereoscopic video signal, the monoscopic video signal is obtained from the viewpoint of the intermediate position of the left and right cameras, not the same view as the left camera or the right camera In order to do so, it is preferable to use both boundary information of the left and right images.

10 is a conceptual diagram illustrating a boundary correction method according to an embodiment of the present invention. Referring to FIG. 10, the left and right composite images 120 of the side-by-side method are generated from the left image 110a and the right image 110b obtained by using both cameras, and then encoded. The coded signal can then be transmitted over a normal channel. The signal received by the 2D display is corrected by using one of the first boundary correction method 130a, the second boundary correction method 130b, and the third boundary correction method 130c to correct the boundaries of the inconsistency region. Create a 3D image.

As shown in FIG. 10, the first boundary correction 130a method includes position information values of boundaries of a mismatched region of a reduced left image 120a and a reduced right image 120b, for example, a horizontal position information value and a vertical position information value. The original image or the image is restored by correcting the value to a value through statistical processing such as a mean value, a median, and the like of the reduced left image 120a and the reduced right image 120b. The second boundary correction method 130b restores the original image by correcting the boundaries by enlarging or reducing the inconsistent region of the reduced right image 120b based on the viewing angle of the reduced left image 120a. The third boundary correction method 130c enlarges and reduces inconsistent regions of the reduced left image 120a based on the viewing angle of the reduced right image 120b. In addition to the above, it can be converted by various other similar methods that can be naturally connected and corrected in the inconsistency boundary of the image. It is apparent to those skilled in the art that the larger the size of the image, the more natural the image can be reproduced after the enlargement and reduction.

According to an aspect of an embodiment of the present invention, it is possible to correct using the median value of the reduced left image 120a and the reduced right image 120b of the boundary position information values of the pixel or unit block, which are the mismatched regions, One example is to use a median filter. The median filter may be applied to a predetermined number of pixels or blocks adjacent to each other. For example, if a plurality of input values (3, 6, 4, 8, 9) pass through the median filter, the intermediate value (6) is output to the output. In this way, it can be applied to the extracted boundary position information value. The filter that can be used to correct as described above is not limited to the median filter, but may be applied to filters that can be used to correct to an average value or other values through statistical processing.

10, an image filter may be used to enlarge or reduce the reduced left image and the reduced right image mismatched regions according to the second boundary correction 130b or the third boundary correction 130c. An interpolation filter may be used. When an image is enlarged or reduced in one of a reduced left image and a reduced right image, the effect of image restoration can be enhanced by using an interpolation filter by interpolating with reference to the remaining images.

Referring to FIG. 7 again, a two-dimensional image is generated by using the image information of the matched region divided and extracted in S10 and the image of the mismatched region corrected in step S30 (S40). Here, the 'two-dimensional image' is made using a reduced left image and / or a reduced right image including a matching area that has not been corrected in steps S20 and S30 and a mismatched area that has been corrected according to steps S20 and S30. Can lose.

For example, a monoscopic image may be generated by doubling the image of either the corrected reduced left image or reduced right image. In this case, the correcting step S30 may be performed only for one image to be enlarged. Alternatively, the monoscopic image may be generated by alternately arranging vertical lines extracted from the corrected reduced left image and the corrected reduced right image.

FIG. 11 is a diagram illustrating reproducing an image of a pentagonal body in a 2D display by generating a 2D image according to an exemplary embodiment of the present invention.

 Referring to FIG. 11, it can be seen that the pentagonal shapes shown in the reduced left image 102a and the reduced right image 102b are different from each other. As shown in FIG. 11, the left-right composite image 102 of the side-by-side method is generated from the reduced left image 102a and the reduced right image 102b and then encoded. Then, when the integrated composite image is generated from the reduced left image 102a and the reduced right image 102b, the change values of the reduced left image 102a image information and the reduced right image 102b image information of adjacent pixels or blocks are changed. An inconsistency region having a threshold value α or higher is extracted. In FIG. 11, the mismatched area may be an area occupied by the pentagonal body in the image. This part appears to be enlarged or reduced in size according to the angle of the left and right cameras, and appears as a mismatched area when the reduced left image 102a and the reduced right image 102b are combined. Therefore, the reduced right image 102b is converted into the inconsistency region based on the reduced left image 102a using an interpolation filter. Then, when the vertical lines of the converted reduced left image 104a and the reduced right image 104b are alternately arranged and reconstructed into the first type integrated synthesis image 106, the 2D display may be naturally reproduced.

Next, an apparatus for converting a stereoscopic video signal into a 2D video signal according to an embodiment of the present invention will be described. The conversion apparatus according to the embodiment of the present invention may be included in the 2D display apparatus, or may be included in a means such as a receiver or a server that receives a stereoscopic video signal and outputs it to the 2D display apparatus. Alternatively, the conversion apparatus may be included in a transmission server for supplying a stereoscopic video signal. In this case, for example, when the user requests a user, the conversion apparatus may convert the stored stereoscopic video signal and transmit the converted stereoscopic video signal to the user.

12 is a block diagram illustrating a configuration of an apparatus for converting a stereoscopic video signal into a 2D video signal according to an embodiment of the present invention. The block diagram of FIG. 12 is for implementing the conversion procedure shown in FIG. 7, and each step in the conversion procedure shown in FIG. 7 may be performed in one unit of FIG. 12. However, this is exemplary and may be configured such that one step in FIG. 7 is performed in two or more units or two or more steps in FIG. 7 are performed in one unit.

Referring to FIG. 12, the apparatus 200 for converting a stereoscopic video signal into a 2D video signal includes an inconsistency region extractor 210, an inconsistency region boundary extractor 220, an inconsistency region boundary corrector 230, and It includes a 2D image generating unit 240. The mismatched region extractor 210 distinguishes a matched region and a mismatched region of each image by using a stereoscopic video signal including a reduced left image and a reduced right image. For example, in the case where the input signal has the form of an integrated composite image, an inconsistency between the reduced left image and the reduced right image in a region where a difference between the adjacent reduced left image image information and the reduced right image image information is greater than or equal to a threshold α Can be considered an area. On the other hand, the matched area may be an area far from the camera, for example, a background screen, and the mismatched area may correspond to an object or a person close to the camera.

The disparity region boundary extractor 220 extracts boundaries in the extracted disparity region of the reduced left image and the reduced right image, respectively. The mismatched region boundary corrector 230 corrects at least one of the reduced left image and the reduced right image by using boundary information extracted in the mismatched region. Preferably, the image is corrected by statistical processing such as mean, median, etc. of boundary position information values of the reduced left image and the reduced right image, or is reduced based on the viewing angle of the reduced left image. Zoom in and out the inconsistent area of the zoomed-out image, or zoom in and out based on the viewing angle of the zoomed-in right image.

The 2D image generator 240 generates a 2D image by alternately arranging vertical or horizontal components of the reduced left image and the reduced right image, which are composed of the matching region and the disparity region corrected by the boundary corrector 230. If the input image is a side-by-side split composite image signal, the integrated composite image may be generated by the 2D image generating unit 240 alternately arranged in a vertical component. When the integrated synthetic video signal of the method is input, an integrated composite video that is alternately arranged and integrated with horizontal components may be generated.

The present invention is not limited to the embodiments described above, and various modifications and changes can be made by those skilled in the art, which are included in the spirit and scope of the present invention as defined in the appended claims.

1 and 2 illustrate an example of a stereoscopic image.

FIG. 3 is a diagram for describing a principle of reproducing a stereoscopic image using a barrier type display.

FIG. 4 is a diagram illustrating a side by side scheme of left and right composite images generated from stereoscopic images.

5 is a conceptual diagram illustrating an embodiment of reproducing a side-by-side type left and right composite image on a 2D display according to the present invention.

6 is a conceptual diagram illustrating an embodiment of reproducing a top-and-down left and right composite image on a 2D display according to the present invention.

7 is a flowchart illustrating a procedure of converting a stereoscopic video signal into a 2D video signal according to an embodiment of the present invention.

(A) is a picture showing a left image of the stereoscopic image, (b) a picture showing a right image, (c) a picture constituting a left-right composite image of the side-by-side method, (d) a first type Fig. (E) shows the coincidence region divided by unit block (N x M) which is the coding unit of the image in (D), and (F) shows the image of the image in (D). This figure shows the inconsistency region partitioned into unit blocks (N x M) which are coding units.

FIG. 9A illustrates a Sobel mask, FIG. 9B illustrates an image to which the Sobel mask of FIG. 9A is applied, and FIG. 9C illustrates a detection result of applying the Sobel mask of FIG. 9A to the image of FIG. 9B.

10 is a conceptual diagram illustrating a boundary correction method according to an embodiment of the present invention.

FIG. 11 is a diagram illustrating reproducing an image of a pentagonal body in a 2D display by generating a 2D image according to an exemplary embodiment of the present invention.

12 is a block diagram showing the configuration of an apparatus for converting a stereoscopic video signal into a 2D video signal according to an embodiment of the present invention.

Claims (10)

A method for converting a stereoscopic video signal consisting of a reduced left image and a reduced right image into a monoscopic video signal, Comparing the reduced left image with the reduced right image and specifying an inconsistency region in each of the reduced left image and the reduced right image; Extracting boundary information from an inconsistency area between each of the reduced left image and the reduced right image; Correcting one or both images of the reduced left image and the reduced right image by using boundary information of the two images; And And generating a two-dimensional image using the corrected reduced left image and / or the corrected reduced right image. The conversion method according to claim 1, wherein in the correcting step, the reduced left image and / or the reduced right image are corrected so that an average value of boundary information of the two images becomes a boundary of the corrected image. The conversion method of claim 1, wherein in the correcting, the inconsistency region of the remaining image is enlarged or reduced while correcting by referring to one of the reduced left image and the reduced right image. The conversion method according to claim 1, wherein the correction step uses camera parameter information included in the stereoscopic video signal. A method for converting a stereoscopic video signal into a monoscopic video signal, The stereoscopic video signal includes image information of left and right images and camera parameter information used to obtain the left and right images. And the camera parameter information is used to generate the monoscopic video signal by correcting one or two images from the left and right images. The conversion method according to claim 5, wherein all or part of a boundary line is extracted from one or two images of the left and right images, and the extracted boundary line is corrected using the camera parameter information. The conversion method according to claim 5, wherein the monoscopic video signal is an image including a subject image at a viewpoint at an intermediate position of each viewpoint of the left and right images.  A method for converting a stereoscopic video signal consisting of a reduced left image and a reduced right image into a monoscopic video signal, And converting the reduced left image and the reduced right image by using boundary information extracted from all or a part of the reduced left image to generate an image including a subject image at an intermediate point in time between the reduced left image and the reduced right image. . The method of claim 8, wherein the boundary information is extracted only in an inconsistency region between the reduced left image and the reduced right image. An apparatus for converting a stereoscopic video signal composed of a reduced left image and a reduced right image into a monoscopic video signal, A unit for comparing the reduced left image and the reduced right image to specify an inconsistency region in each of the reduced left image and the reduced right image; A unit for extracting boundary information from an inconsistency area of each of the reduced left image and the reduced right image; A unit for correcting one or both of the reduced left image and the reduced right image by using boundary information of the two images; And And a unit for generating a two-dimensional image using the corrected reduced left image and / or the corrected reduced right image.

Images