KR20000041329A - Stereoscopic image converting method and apparatus - Google Patents

Abstract

PURPOSE: A stereoscopic image converting apparatus is provided to realize more precise stereoscopic images by providing three dimensional stereoscopic image materials with an original image synthesized to an object image viewed in the right eye. CONSTITUTION: An original image storage(10) inputs a two dimensional material of an original image to e converted into a stereoscopic image. The stored original image opens into an object extractor(20) to extract images of respective objects. The extracted images are separately stored by objects in an original cop object arranger(30). The extracted object images are removed from the object extractor(20). An object depth ranger(40) stores depths of images to determine how deep the images are seen. A synthesis object ranger(70) synthesizes the images of the original image storage(10) and images of an edit object ranger(50).

Description

Stereoscopic Image Conversion Method and Apparatus

The present invention relates to a three-dimensional image conversion method and apparatus for generating a three-dimensional stereoscopic image from a general two-dimensional image, and more particularly, after changing the image of each object on the general two-dimensional image data to three-dimensional, each object The present invention relates to a 3D stereoscopic image conversion method and apparatus for reducing a lot of effort and cost for producing 3D stereoscopic image data by synthesizing a right eye image with an original image.

A person feels three-dimensional by looking at an object from different angles in his left and right eyes. Based on this principle, stereoscopic images are artificially realized by showing two images taken separately at the visual position of the left eye and the right eye to the left eye and the right eye, respectively.

However, in order for the conventional technology to be effective, the image to be seen by the left eye and the image to be seen by the right eye must be present as described above. Therefore, in order to create a stereoscopic image, a new image must be produced using two cameras. There was a difficulty. This indicates that many existing two-dimensional images cannot be used in a stereoscopic image implementation system. Due to the above reasons, the lack of stereoscopic images that can be encountered by the general user is an obstacle to the popularization of stereoscopic imaging technology, and there is a problem in that application fields of stereoscopic imaging technology consume a lot of time and cost for producing image data.

Therefore, the present invention has been invented to solve the above problems, and its purpose is to change the image of each object on a general two-dimensional image data into a three-dimensional image, and then synthesize the right eye image of each object with the original image. The present invention provides a method and apparatus for converting a stereoscopic image image by reducing the effort and cost required for the production of the stereoscopic image data by making the 3D stereoscopic image data.

1 is a block diagram schematically showing the configuration and data flow of a stereoscopic image converting apparatus according to the present invention;

2 is a flow chart schematically showing a sequential algorithm of the present invention

3 is a flow chart showing in detail the steps involved in object image extraction and objectification of the portion of FIG.

4 is an explanatory diagram showing an embodiment of a three-dimensional image

Explanation of symbols on the main parts of the drawings

10: original storage 20: object extraction unit

30: original object array unit 40: object depth array unit

50: editing object array unit 60: composite stereoscopic image unit

70: composite object array unit 80,82,84: signal synthesis unit

In order to achieve the above object, the stereoscopic image image conversion method of the present invention,

A first step of reading an original two-dimensional image to be converted into a stereoscopic image, storing it, and initializing a data structure;

Extracting and objectifying each object image from the stored original image and storing the same;

A third step of adjusting and editing the extracted object image according to the depth of each pixel and the depth of the object;

And a fourth step of synthesizing the edited object image and the entire image and outputting them on the screen.

In this case, the method of extracting each object image in the second step includes extracting an outline of a desired object for each object image, and calculating a rectangle having the smallest size that includes the extracted outline. Allocating a space for storing the newly extracted object, initializing the allocated space, comparing which pixel of the pixel inside the rectangle is inside the outline line, and the comparing step Drawing a pixel of transparent color at the position corresponding to the original object array and the editing object array if a pixel is outside the outline, and editing the original object array if any pixel is inside the outline at the comparing step. Copying it to the object array and converting the pixel to transparent color at the corresponding position in the object extraction space. It characterized by that.

The stereoscopic image image conversion device of the present invention for achieving the above object,

Original storage means for receiving and storing the original two-dimensional image data to be converted to three-dimensional,

Object extraction means for extracting and objectifying each object image from the original image stored in the original storage means and storing the same;

Original object array means for storing the image of each object extracted from the object extraction means for each image;

Object depth arrangement means for adjusting and editing the extracted object image according to the depth of each pixel and the depth of the object;

An editing object arrangement means for changing the image of the original object arrangement means according to the depth value of the object depth arrangement means to create an image of the right side of the object and storing it;

Synthetic object array means for combining the image of the original object array means and the image of the edited object array means to store an image of each object actually seen in three dimensions;

And synthesizing three-dimensional image means for synthesizing both the image of the original storage means and the right eye image of each object stored in the editing object arrangement means, and finally storing the resultant image obtained by the user.

According to the three-dimensional image image conversion method and apparatus of the present invention by the above configuration, after changing the image of each object on the general two-dimensional image data to a three-dimensional, and then synthesized the right eye image of each object with the original image three-dimensional three-dimensional By making it into video material, there is an effect that can save a lot of effort and cost in the production of stereoscopic video material.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the accompanying drawings. In addition, in all the drawings for demonstrating an embodiment, the thing with the same function uses the same code | symbol, and the repeated description is abbreviate | omitted.

First, the principle of the present invention will be described.

When there are two objects at different distances from the observer, the position of the focal point of vision is different when looking at a near object or when looking at a far object. Since the distance between both eyes of a person is constant, the angle between the image of both eyes is different when looking at a near object and when looking at a far object. Therefore, when the image of the left eye and the image of the right eye are projected on one screen, the distance difference between the two images is different depending on the distance between the object and the observer. Due to this effect, the human eye can feel the distance between the object and itself by the distance difference between the two images formed on both eyes.

This principle is applied to general video data taken with a single camera to create a video material that can feel three-dimensional. For example, the object A near the camera at the time of shooting and the object B farther than the object A will be described as an example. If the image was viewed from the left eye, in the image from the right eye, each object would have been formed at a position in which the object moved horizontally and horizontally from its original position. However, because the distance between the two objects and the observer is different, the displacement value to move in parallel as described above should be different. In other words, the near object has a large angle between the images of both eyes, and the far object has a small angle, so that the close object A is drawn in a parallel position with a large displacement, and the far object B is a small displacement. If you draw in the parallel position with, you can make an image of object A and object B seen with the right eye. Now, if the image of the left eye is shown only in the left eye and the newly created image of the right eye is shown only in the right eye, an observer who sees the two images with different eyes at the same time can feel the image of the object in three dimensions.

In the present invention, after converting the image of each object on the general two-dimensional image data to a stereo by this method, a new stereoscopic image is created by synthesizing the right eye image of each object with the original image. In other words, each object on the two-dimensional image is recognized as a separate object and then stereoscopically reconstructed and then stereoscopically combined with the original image to create a finally synthesized stereoscopic image. This is not a view of the original image itself as the left eye image, but a new right eye image, which is not synthesized. Instead, each object inside the original image is viewed as a single left eye image. It is characteristic to produce the whole image afterwards.

The structure and operation of the present invention constructed by the above principle will be described in more detail with reference to the accompanying drawings.

1 is a block diagram schematically showing the configuration and data flow of a stereoscopic image converting apparatus according to the present invention. FIG. 2 is a flowchart schematically showing a sequential algorithm of the present invention, and FIG. 3 is a flowchart showing the steps of object image extraction and objectization in detail in FIG. 2.

First, referring to FIG. 1, original 2D image data to be converted into stereoscopic images is received and stored in the original storage unit 10. The original image stored here is used as the image of the left eye to obtain a stereoscopic composite image. The original image stored here is copied to the object extractor 20 as it is, and the image of each object is extracted from the object extractor 20. The extracted image is stored separately for each object in the original object arranging unit 30, and the image of the extracted object is removed from the object extracting unit 20. If the image of each object is extracted from the original image as described above, the work in the future proceeds with the objects in consideration of the image of each object stored in the original object array unit 30 as one object.

Each object corresponds to three data spaces. The first is the original object array section 30 already described, and the second is to store a 'depth', which is a value that determines how much of each pixel of the object appears or pops out of the three-dimensionally edited image. The object depth arranging unit 40, and the third is the editing object arranging unit 50 'for storing the right eye image generated by editing the original object image.

If the depth of any pixel of the object currently being edited by the user's input is changed, the contents of the object depth arranging unit 40 are changed, and at the same time, the image of the original object arranging unit 30 is changed to the depth of the object depth arranging unit 40. Depending on the value, it works in the same manner as described above to produce the right image of the object in the editing object arranging unit 50.

As such, the object image is edited and a space for storing the synthesized object image and the entire image is needed to show the image and the whole image of the three-dimensionally synthesized object.

The composite object arranging unit 70 is a place where the image of the original object arranging unit 30 and the image of the editing object arranging unit 50 are synthesized to store an image of each object actually seen in three dimensions.

The composite stereoscopic image unit 60 synthesizes both the image of the original storage unit 10 and the right eye image of each object stored in the editing object arranging unit 50 to finally store a result image that the user will obtain. to be. The composition of the image is done in real time as the object is edited, thus allowing the user to work while visually seeing the results.

Based on such a structure, a stereoscopic image is produced by the following operation. FIG. 2 is a flowchart schematically showing the operation of the present invention. Hereinafter, each part of the operation will be described in detail with reference to FIG. 2.

First, the original image is read and initialized for each storage space (step 100).

The read original two-dimensional image is stored in the original storage unit (10 in FIG. 1) and the object extraction unit (20 in FIG. 1) is also initialized to the original image. Since the editing has not been done yet, the composite stereoscopic image part (60 in FIG. 1) is also initialized to the original image.

Thereafter, each object image is extracted from the original image (step 120). This process will be described in detail with reference to FIG. 3.

First, an outline of a desired object is extracted using an algorithm previously released on an image of the object extractor 20 (step 121). Next, the smallest rectangle containing the extracted outline is calculated (step 122).

Then, the space for storing the newly extracted object is stored in the original object array unit (30 in FIG. 1), the edit object array unit (50 in FIG. 1), the composite object array unit (70 in FIG. 1), and the object depth array unit. (40 in Fig. 1) '(step 123). Note that the size of the space to be allocated to the original object array unit (30 in FIG. 1), the edit object array unit (50 in FIG. 1), and the composite object array unit (70 in FIG. 1) is determined by the size of the rectangle. Instead of precisely assigning them, they should be allocated with the maximum depth left and right. This is because the original object image is drawn in a new position by moving from side to side according to the depth of each pixel. For example, if the size of the object image is "140 pixels x 60 pixels" and the maximum depth is 100, the size of the space to be allocated should be a size corresponding to "340 pixels (140 + 100 + 100) x 60 pixels". Since the object depth arranging unit (40 in FIG. 1) is used to store the depth of each pixel, as shown in the original object arranging unit (30 in FIG. 1) and the editing object arrangement (50 in FIG. Do not add

Next, the allocated space is initialized as described above (step 124).

The original object array unit 30 (FIG. 1) and the edit object array unit 50 (FIG. 1) are all initialized to transparent colors in order to prevent damage to the image. All depth values of the object depth arranging unit 40 in FIG. 1 are initialized to zero. Based on the computer's monitor, if the depth is zero, the pixel appears to be on the computer monitor; if the depth is positive, the pixel appears to be closer than the reference point. Conversely, if the depth is negative, the pixel appears to be farther from the reference point. In other words, positively-depth pixels appear to protrude from the computer monitor, while negative-depth pixels appear to be coming from the computer monitor.

In addition, the 'object depth' of the newly extracted image is also initialized to zero.

Here, 'object depth' refers to how each object image protrudes or is depressed compared to other objects, if the object image is referred to as an 'object'. By adjusting this value, you can locate each object in space. In fact, the depth of each pixel in the finally obtained composite stereoscopic image part (60 in FIG. 1) is calculated as "object depth + pixel depth".

Then, the image of the newly extracted object is copied to the center position of the allocated space (space except the left and right sides of the space) (steps 125 and 126).

The copying of the image is applicable not only to the original object arranging unit 30 (FIG. 1) but also to the editing object arranging unit 50 (FIG. 1). When the original object image is regarded as the image of the left eye in the editing object arrangement unit (50 in FIG. 1), the corresponding right eye image is stored, and since the object has not been edited yet, the original object arrangement unit (30 in FIG. 1). ) And the image of the editing object arrangement unit (50 in Figure 1) is the same. The copying process of the extracted image is performed for all the pixels in the object image stored in the original object array unit 30 of FIG. 1.

If a pixel is inside the outline line, the pixel at the corresponding position of the object extractor 20 of FIG. 1 is changed to a transparent color (step 126). By doing so, the image of the newly extracted object can be removed by the object extracting unit (20 in FIG. 1), and the image of another object to be extracted next will not be affected. If a pixel is outside the outline (step 127), transparent pixels are drawn at the corresponding positions of the original object array (30 in FIG. 1) and the editing object array (50 in FIG. 1). In addition, the left and right extra spaces are filled with transparent colors so that there is no damage to the image.

Meanwhile, the extracted image remaining is stored as a 'background object'. The background object is also initialized through the above process. However, since the background object is the portion of the original image corresponding to the background and should be the most depressed, the object depth of the background object is set to the largest negative value (−1 × maximum depth).

If all the images of each object are extracted and objectized through the above process, the object images are edited separately (step 140 of FIG. 2).

Editing an object means adjusting the depth of each pixel on the object and adjusting the object depth. The user changes the depth of each pixel to be positive or negative with respect to zero. Adjust the depth to a negative value for the part that needs to be seen and adjust the depth to a positive value for the part that needs to be popped out. Then adjust the absolute value of the depth according to the depth. Through this process, the depth of each pixel of the object depth array unit 40 of FIG. 1 is changed.

According to the contents of the changed object depth arrangement unit 40 (FIG. 1), the image of the right side of the object is changed from the image stored in the original object arrangement unit 30 (FIG. 1) to the editing object arrangement unit 50 (FIG. 1). Make it up As shown in FIG. 4, when the depth is positive, the image of the right eye moves parallel to the left from the position of the image of the left eye. Conversely, if the depth is negative, it will translate parallel to the right from the position of the image of the left eye. And the greater the absolute value of the depth, the greater the amount of parallel translation. In this manner, the pixels of the original object array unit 30 (of FIG. 1) having the changed depth are moved to the left or right in parallel according to the depth, and copied to the corresponding positions of the editing object array unit (50 of FIG. 1).

When the image of the editing object arrangement unit 50 of FIG. 1 is changed through the above process, the image of the original object arrangement unit 30 of FIG. 1 and the image of the editing object arrangement unit 50 of FIG. 1 are synthesized. An image of the object array unit 70 of FIG. 1 is generated to allow a user to check an image of a three-dimensionally synthesized object in real time (step 160 of FIG. 2).

In synthesizing the images, the image of the left eye is an image of the original object array unit (30 in FIG. 1), and the image of the right eye is using the image of the editing object array unit (50 in FIG. 1). use. Read a line of image data from the left eye image and copy it to the top line of the buffer to store the synthesized image. Next, we read one line of image data from the right eye image and copy it to the next line of the buffer. In this way, the left eye image and the right eye image are read one by one, and the contents are written one by one in the buffer one after the other. The synthesized stereoscopic image can be used in a stereo viewer that supports line blanking.

The synthesized object image is stored in the synthesized object array unit 70 of FIG. 1 and output to the screen as it is so that the user can check it.

In the case of the entire image, the object having the smallest object depth among the objects in the editing object arrangement unit 50 of FIG. 1 is synthesized using the original image stored in the original storage unit 10 of FIG. 1 as the left eye image. In other words, it combines the background object with the original image and then combines it with the image that has been synthesized here and the object with the next smaller object depth value. When compositing sequentially from the smallest object depth like this, objects that should appear to be far away (objects with small object depth) are overwritten by objects that need to be seen as near (objects with large object depth), It fits well with real situations that are not covered by nearby objects.

As described above, according to the stereoscopic image image conversion method and apparatus according to the present invention, after changing the image of each object on the general two-dimensional image data to a stereoscopic image, the right eye image of each object is synthesized with the original image Since it is configured to make three-dimensional stereoscopic image data, the image of the object itself can be edited, so more detailed three-dimensional image can be realized. In addition, it is possible to obtain a more solid three-dimensional effect by composing the entire image by integrating independently edited objects instead of editing the entire image, and utilizing a large number of existing two-dimensional images. It's a great way to save time and money on creating your own data.

In addition, preferred embodiments of the present invention are disclosed for the purpose of illustration, those skilled in the art will be able to various modifications, changes, additions, etc. within the spirit and scope of the present invention, these modifications and changes should be seen as belonging to the following claims. something to do.

Claims (3)

A first step of reading an original two-dimensional image to be converted into a stereoscopic image, storing it, and initializing a data structure; Extracting and objectifying each object image from the stored original image and storing the same; A third step of adjusting and editing the extracted object image according to the depth of each pixel and the depth of the object; And a fourth step of synthesizing the edited object image and the entire image and outputting the synthesized image on the screen. The method of claim 1, The method of extracting each object image in the second step, Extracting the outline of the desired object for each object image, Calculating a rectangle of the smallest size containing the extracted outer line; Allocating a space to store the newly extracted object; Initializing the allocated space; Comparing which of the pixels inside the rectangle are inside the outline; Drawing a pixel of transparent color at a position corresponding to the original object array and the edited object array if a pixel is outside the outline in the comparing step; And if the pixel is in the outline in the comparing step, copying it to the original object array and the editing object array, and converting the pixel into a transparent color at a corresponding position in the object extraction space. In the stereoscopic image conversion device, Original storage means for receiving and storing the original two-dimensional image data to be converted to three-dimensional, Object extraction means for extracting and objectifying each object image from the original image stored in the original storage means and storing the same; Original object array means for storing the image of each object extracted from the object extraction means for each image; Object depth arrangement means for adjusting and editing the extracted object image according to the depth of each pixel and the depth of the object; An editing object arrangement means for changing the image of the original object arrangement means according to the depth value of the object depth arrangement means to create an image of the right side of the object and storing it; Synthetic object array means for combining the image of the original object array means and the image of the edited object array means to store an image of each object actually seen in three dimensions; 3D image comprising a composite stereoscopic image means for synthesizing both the image of the original image storage means and the right eye image of each object stored in the editing object array means and finally stores the resultant image to be obtained by the user Image conversion device.