KR101019558B1 - Method for Compression in System for Editing Virtual Reality - Google Patents

Abstract

The present invention relates to a compression method in a virtual reality editing system that can increase the reality of the virtual reality content transmitted to a web server by reducing the file size by storing the virtual reality content data at a high compression ratio when the virtual reality editing program is running. Reading the 3D model data produced by the 3D application and generating the virtual reality content, losslessly compressing the internal information data of the virtual reality content, and lossy compressing the external information data of the virtual reality content. Include.

According to the present invention, the internal data of the 3D model data is compressed using a lossless compression algorithm, and the external data is compressed using the lossy compression algorithm, so that virtual reality content can be compressed at a rate of several tens of frames per second or more even in a generalized hardware specification. It is possible to provide a compression method in a virtual reality editing system that can render and store content at a high compression ratio, thereby reducing file size and increasing reality.

Description

Method for Compression in System for Editing Virtual Reality

The present invention relates to a compression method in a virtual reality editing system. Specifically, the virtual reality content data is stored at a high compression ratio when the virtual reality editing program is driven, thereby reducing the file size to improve the reality of the virtual reality content transmitted to the web server. A compression method in a virtual reality editing system that can be increased.

In the past, two-dimensional and static images and text based on Hyper Text Markup Language (HTML) have been the mainstay of Internet documents. However, in recent years, users can use Java, ActiveX, and VRML to communicate with users. Interactive three-dimensional images are expanding into the Internet.

This interactive 3D image is applied on the web through the Internet. In order to operate the 3D image directly on the web, the 3D image is first modeled by a 3D modeling program and converted into a form that can be driven on the web.

However, when the 3D image is expressed in detail in order to realistically express and even the material is coated in a similar manner to the reality, the size of the contents data file becomes very large. Due to the nature of the online medium of the Internet, the larger the file size, the lower the responsiveness the user feels due to the limitation of the transmission speed, and the corresponding page is turned off.

The solution is to reduce the quality, such as simplification of expression, to change the material in order to reduce the capacity, or to lower the rendering frame rate below 30 frames that can be perceived by the eyes. There is a problem that the reality of the three-dimensional image displayed on the Internet is falling.

An object of the present invention for solving the above problems is that the internal data of the virtual reality content data is compressed using a lossless compression algorithm, and the external data is compressed using the lossy compression algorithm, so that even in the generalized hardware specifications It is possible to render content at a rate of more than a few tens of frames per second, and to provide a compression method in a virtual reality editing system that can increase the reality by reducing the file size by storing the virtual reality content at a high compression ratio.

Compression method in the virtual reality editing system of the present invention for achieving the above object, (a) reading the three-dimensional model data produced by the three-dimensional application to create a virtual reality content, (b) the virtual reality content Lossless compression of the internal information data of and (c) Lossless compression of the external information data of the virtual reality content provides a compression method in a virtual reality editing system.

The step (b) may include (b1) compressing the internal information data of the virtual reality content using an LZ77 algorithm, and (b2) compressing the compressed data by Huffman coding.

In the step (c), (c1) dividing the external data of the virtual reality content into a matrix having a predetermined size to form a block, (c2) discrete cosine transforming matrix data of each block, (c3) Quantizing each discrete cosine transformed matrix data, (c4) zigzag scanning the quantized matrix data, and (c5) compressing the zigzag scanned data with Hoffman coding.

In the step (c1), (c11) reading the size of each block matrix preset by the user from a configuration file, (c12) dividing the external data of the virtual reality content by the size of each preset block matrix. Forming a block.

In the step (c), repeating the steps (c1) to (c5) with respect to at least one matrix size, comparing the sizes of the compressed data repeatedly repeated with the respective matrix sizes, and having the smallest compressed data. And selecting the size of the matrix, and recording and storing the size of the matrix used for compression in the header of the selected compressed data.

As described above, according to the present invention, the internal data of the three-dimensional model data is compressed using a lossless compression algorithm, and the external data is compressed using the lossy compression algorithm, so that virtual reality content can be stored in tens of seconds per the generalized hardware specification. It is possible to provide a compression method in a virtual reality editing system capable of rendering at a speed higher than a frame, and storing content at a high compression ratio to reduce file size, thereby increasing reality.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention.

1 is a block diagram showing a virtual reality editing system according to an embodiment of the present invention, Figure 2 is a block diagram showing a virtual reality content generating unit according to an embodiment of the present invention, Figure 3 is an embodiment of the present invention FIG. 3 is an exemplary diagram illustrating a state in which a virtual reality content generating unit is displayed on a user terminal.

In the virtual reality editing system according to an embodiment of the present invention, as shown in FIGS. 1 and 2, virtual reality content generation is performed by reading three-dimensional model data 160 produced by a three-dimensional application and generating virtual reality content. And a web server 200 for receiving and storing the virtual reality content generated through the user terminal 100 and the virtual reality content generation unit 150 provided with the unit 150.

The user terminal 100 is provided with a web browser 110 such as an internet explorer capable of accessing the internet 300, and has a temporary storage unit 130 for temporarily storing data such as content.

Here, the virtual reality content generating unit 150 is installed in the user terminal 100 in the form of software, and edits the 3D model data 160 produced by the 3D application to be driven on the web. (Editing)

As shown in FIG. 2, the virtual reality content generating unit 150 includes a three-dimensional engine 151, an object file generating unit 155, and an html file generating unit 156.

The 3D engine 151 is composed of a common library 152, an OpenGL wrapper, and an OpenGL implementation unit 153 defined for various objects related to all libraries and open links based on OpenGL.

The common library 152 is composed of mathematical (linear algebra) related functions and classes, data structure classes, and other utility functions and classes, and is statically linked throughout the system.

OpenGL implementation unit 153 includes the minimum unit class defined in the common library 153, and the classes constituting a single scene such as geometric entities, texture images, textures and the like optimally placed on the screen It consists of a pipeline, other utility classes, and so on.

Therefore, if a special conversion is required in addition to the functions supported by OpenGL, a function by following the above formula conversion is configured to make additional functions by calling.

This is not included in OpenGL's basic library, but it can be recognized as a library because it uses the same conversion formula. The 3D engine 151 is constructed by integrating the common libraries 152 configured to OpenGL.

The object file generator 155 corresponds to a user interface of the 3D engine 151, and stores the 3D model data 160 edited by the 3D engine 151, trigger actions, and events. Create an object file.

In addition, the object file executor (not shown) may be executed by receiving the posted object file stored in the web server. The object file executor is a module that reads an object file generated in the virtual reality editing system according to the present invention and executes the object file in an embedded form in a user's browser.

The html file generator 156 operates while being embedded in the web browser 110 and generates html (hypertext markup language) for embedding the object file. That is, an html script that calls the object file generator 155 is generated to receive and execute the object file posted on the web server.

In addition, the virtual reality content generating unit 150 according to an embodiment of the present invention is installed in the user terminal 100 in the form of software and displayed on the user terminal 100.

As shown in FIG. 3, the virtual reality content generating unit 150 includes a menu 400, a model viewer 500, a scene graph 600, and an action / event tree. Tree 700, Object Properties 800, and Event Map 900.

The menu 400 includes the main menu, action / event menu, playback menu, frame menu, default camera menu, rotate menu, zoom ( Zoom menu, View menu, etc. are configured.

The main menu sets up a new work frame and reads in the 3D model data 160 file. In addition, the current work window is terminated or the currently working file is stored, and the connected virtual browser is connected to a web browser 110 such as an internet explorer so that the stored virtual reality content can be viewed on the web.

The action / event menu defines the movement, visibility, camera, sound, etc. of the model, such as the 3D model data 160, and clicks the defined action with an appropriate event. It can be controlled through (Mouse Click), Keystroke, World Message, Internal Event, etc. It also binds actions and events to a trigger and executes the specified trigger.

The playback menu plays, pauses, or stops the 3D animation, and the 3D animation can be returned to the beginning or the end.

The frame menu converts viewpoints and coordinates so that the 3D model data 160 can be easily edited.

In addition, the 3D model data 160 may be moved to the screen center point, and the 3D model data 160 may be smoothly changed when the screen is changed.

The default camera menu converts the camera's coordinates to the default coordinates, converts the camera's viewpoint to the X-Z plane, the X-Y plane, or the Y-Z plane, and also the pi (π) plane.

The rotate menu rotates the model in the X, Y, and Z axes, and can rotate 15 degrees in the X-axis rectangle, 15 degrees in the Y-axis, or 15 degrees in the Z-axis.

The zoom menu can be used to zoom in / out the model or fit screen to fit the screen. In addition, the model can be scaled up or down by 5%.

The view menu enables or disables the windows constituting the model viewer 500, the scene graph 600, the action / event tree 700, the object properties 800, and the event map 900.

The model viewer 500 displays the virtual reality content downloaded from the web server 200 and displays the virtual reality content converted from the 3D model data 160.

The scene graph 600 may include an object tree, a camera, a light, a material list, a map property, and a texture image of the read 3D model data 160. Texture Images) and sound information.

The object tree information enumerates object information of the 3D model data 160. When the object is selected, the 3D model data 160 corresponding to the model viewer 500 is activated.

In addition, in the object tree, you can set the action / event, show / hide, and shade mode of the object, and provide a fade in / out action when selecting an object.

In the camera information, it is possible to set a camera that can observe the 3D model data 160. The basic camera consists of front, rear, left, right, top and bottom, and the camera properties are activated when the camera is selected. do.

Camera characteristics include the type, projection, position, field of view (Fov), distance, and movement restrict of the observation camera of the three-dimensional model data 160. Set.

The light information represents the light information of the 3D model data 160. In addition, Target Spot, Free Spot, Target Direct, Free Direct, and Omni Skylight provided by 3ds max are provided by default.

The substance list information represents the substance list information of the 3D model data 160 and can be edited by a user arbitrary substance.

In addition, it provides a color value conversion action of an object when selecting a material. For reference, the material has an important function of expressing the texture of the surface constituting the 3D model data 160.

The texture image information provides a texture image list and characteristics used for the 3D model data 160. The texture image feature performs information and editing of the selected image and converts the image into another image. For example, it performs functions such as image scaling, quality, size, zoom in / out, and the like.

The texture image list previews an image list used for the 3D model data 160.

In the action / event tree 700, the action tree consists of setting values of motion, visibility, camera, and sound.

Movememt associates Scaling, Rotation, Translation, and Combined settings of objects with events.

Visibility links the change color and fade in / out settings of an object with events.

The camera associates the camera's viewpoint with the event, and the sound associates the object's sound with the event.

In the action / event tree 700, the event tree associates an action with a mouse, keystroke, playback control, world message, and internal event. Provides an event.

At this time, one action should be linked with one or more events, and an action is performed when a specific linked event occurs.

The object property 800 provides information of a name, a transform, a translation, a rotation, a scale, and a material reference of the selected object.

One three-dimensional model data 160 is composed of a variety of objects, the object properties 800 is required to view the information of the selected object.

The event map 900 performs a function of connecting the event and the action of the 3D model data 160 made by the user. The time span system is introduced and the action / event is gradually progressed within the input time. The unit is 1/1000 second (millisec).

The generated event map can be executed by a Fire command.

4 is a flowchart illustrating a compression method in a virtual reality editing system according to an embodiment of the present invention, FIG. 5 is an exemplary view showing an LZ77 compression algorithm according to an embodiment of the present invention, and FIG. 6 is an embodiment of the present invention. An exemplary view showing a Hoffman coding algorithm according to an embodiment.

As described above, the virtual reality content generating unit 150 is installed in the user terminal 100, and the three-dimensional model is used for three-dimensional applications such as ASE, 3ds max, MAYA, etc., which can generate the three-dimensional model data 160. The data 160 is produced (step S110).

Here, the ASE format file is a file that outputs 3D data in ASCII format. Since the text format is text, the value can be easily seen and easily grasped. Therefore, it is free to simply remove or add specific data from the output ASE file.

On the other hand, ASE files are output as text, so to use them in a program, you need an additional parser to read the text. Therefore, the ASE file is loaded, the sentence of the ASE file is broken by word (TOKEN) unit, and the meaning of the word is analyzed.

Therefore, the word-cutting is called lexer, and the semantic analysis to determine whether the cut word is useful is called a parser.

Subsequently, the 3D model data 160 is read from the virtual reality content generation unit 150 through the input / output unit 154 and generated as graphic data using the object file generation unit 155.

In this way, when the user edits the 3D model data 160 input by the object file generator 155 through the user interface and generates the virtual reality content, the web server 200 increases the reality of the virtual reality content. Compress the data.

Internal data of virtual reality contents such as model vertex data, animation key data, and interaction response values cannot be used if any of the main data is lost. Therefore, the virtual reality content internal data is compressed using a Deflate algorithm, which is a lossless compression algorithm having high compression efficiency with little loss and no time spent in the compression operation (step S120).

The deplate algorithm is an algorithm published by Phil Ketz, who basically compresses the data inside the virtual reality content through the LZ77 algorithm, and once again uses Huffman coding for a pointer to the overlapping content, that is, the location and length of the data match. Compress.

The LZ77 algorithm was published in 1977 by Abraham Lempel and Jakob Ziv, and if the same pattern is found in the data, the content is recorded only by location and length. The length of the pattern is the algorithm that finds the longest possible length and improves its efficiency.

Hoffman coding is an algorithm that David Hoffmann published in 1958 that produces a prefix from the frequency of the characters (a code for which one character is not a prefix of another). The more characters that appear, the shorter the sign.

Huffman coding always produces an optimal prefix for a given frequency, and this prefix is equivalent to a simple binary block code if the frequency of each character is a power of two or all of the same.

For example, if there is data called 'AABBCCDAABBAABB', compression is performed in the order as shown in FIGS. 5 and 6 using a lossless compression algorithm.

First, as shown in FIG. 5A, since there is no data in front of the first 'A' of the string to be searched, (position, length) is all '0', and the first letter that does not match the preceding part is output. , (0, 0) A is output and the length becomes 0 + 1.

Subsequently, as shown in FIG. 5B, the search object has a length of 0 + 1, which is the second 'A', and 'A' is a value present in front, so that the position and length are replaced, and the first of the next non-matching characters is replaced. Print a letter. Therefore, (0,0) A (1,1) B will be output.

Subsequently, as shown in FIG. 5C, the search target is 1 + 1 + 1, that is, 1 + 2, which is the second 'B', and since 'B' is a value present in front, (0,0) A (1, 1) B (1,1) C is output

Subsequently, as shown in FIG. 5D, the search target becomes 1 + 2 + 1 + 1, that is, 1 + 2 + 2, and becomes the second 'C', and since 'C' is a value present before, 0,0) A (1.1) B (1,1) C (1,1) D is output.

Subsequently, as shown in Fig. 5E, the search target becomes 1 + 2 + 2 + 1 + 1, that is, 1 + 2 + 2 + 2, and becomes the third 'A'. Here, the output value is (0,0) A (1,1) B (1,1) C (1,1) D (7,4) because 'AABB' is present in front of 'A'. Becomes A.

Subsequently, as shown in Figs. 5F and 5G, the search target is 1 + 2 + 2 + 2 + 4 + 1, that is, 1 + 2 + 2 + 2 + 5, which is the last 'A'. Here, 'AB' starts with 'ABB', so the final output value is (0,0) A (1,1) B (1,1) C (1,1) D (7, 4) A (4,3).

The Encoding Data produced as a result of the LZ77 algorithm is again compressed by Hoffman Coding, and replaced with a small bit of data for a high frequency pattern. Do the work.

First, the data obtained from the LZ77 algorithm is the same as '00A11B11C11D74A43', and the frequency of each letter is measured and classified.

Then, as shown in Fig. 6B, this operation is repeated while grouping based on the value with the smallest frequency.

So, as shown in Figure 6c, 1 is '01', 0 is '101', A is '100', 4 is '001', B is '110', C is '0001', D is '0000' , 7 becomes '1111', 3 becomes '1110',

 The result of 00A11B11C11D74A43 is

A total of 7 bytes of '1011011000101110010100010101000011110011000011110' can be represented.

7 is an exemplary diagram illustrating a lossy compression algorithm according to an embodiment of the present invention, and FIG. 8 is an exemplary diagram illustrating virtual reality content displayed on a web server according to an embodiment of the present invention.

Subsequently, a loss loss algorithm is used for the external data such as the material of the virtual reality content whose size increases exponentially as the reality is increased, but the compression efficiency is excellent in terms of compression efficiency (step S130).

In the case of material data, color information is recorded for each coordinate of the virtual reality content. In the case of two-dimensional data, the value corresponding to the image has to be very large because the value corresponding to the image must be stored for each side in three dimensions. In general, however, the color of one point is very likely to be similar to the color of the surrounding points.

This is similar to photography, for example when photographing by the sea, the bottom is made of ocher sand and the sky is made up of blue and white clouds, and when you see a tiny point, the surroundings are most likely the same color.

This characteristic and the human eye are so insensitive that small dots are slightly different from the surrounding color, but if you change the color to the surrounding color, that is, at first glance, even if the data is lost, it looks more similar than the deflating algorithm. Excellent compression efficiency can be achieved.

Thus, in the case of material data, each coordinate system is divided into 8 * 8 matrices to form a block, and DCT (Discete Cosine Transform), quantization, and zigzag scanning are performed on the block. zag Scanning) and Hoffman coding algorithms are performed sequentially to increase the compression efficiency.

First, as shown in FIG. 7A, the virtual reality content image is divided into 8 * 8 matrix blocks. In this 8 * 8 block matrix, the amount of information is reduced by the DCT as shown in Fig. 7B.

DCT (Discete Cosine Transform) is an algorithm that transforms one-dimensional or two-dimensional data from the spatial domain to the frequency domain. When DCT operation is performed, large numbers are collected from the block to the upper left.

Thus, the large number at the top left is called the DC (low frequency) value, and the remaining 63 numbers are called AC (high frequency) values. In particular, the DC value and the numbers in the vicinity are very important information that controls the brightness of the entire block. Contains.

In the case of virtual reality content material data, since the change is small, the low frequency and zero frequency (DC) components have large values, and the high frequency components have relatively small values. Proper treatment can lead to high compression ratios.

Quantization is a process of dividing a matrix converted into DCT into a quantization matrix, which is an arbitrary integer matrix as shown in FIG. 7C. In this process, small values are changed to '0' as shown in FIG. 7D, so that the data becomes small, but some data information loss occurs due to the missing numbers.

The matrix, which has undergone the DCT process and the quantization process, is an integer such as [15, 0, -2, -1, -1, -1, -1, 0, 0, -1] by zigzag scanning as shown in FIG. 7E. Create a column and convert the integer column to binary.

Subsequently, using Hoffman coding using the neighbor value property and probability of binary numbers (how many 0s or 1s continue in a row, etc.), an 8 * 8 matrix is represented by [101 0110 111001 01 00 0 00 0 00 0 11011 0]. It is reduced to only a few combinations of zeros and ones.

Compressing this with Hoffman coding can result in compression efficiencies of as little as 10 to as much as 20 times.

Therefore, as the original 8 * 8 integer matrix goes through the process of DCT-quantization-zigzag scanning-Hoffman coding, it can be seen that the amount of data is greatly reduced.

However, very small AC values are lost in the quantization process, which results in a lossy compression method that slightly degrades the quality of the original image.

For reference, in the present invention, the virtual reality content image is divided into 8 * 8 matrix blocks, but the user can arbitrarily set the size of the matrix.

Therefore, by dividing the size of the matrix including 8 * 8 into 16 * 16, 32 * 32, 64 * 64, 128 * 128, etc., the user arbitrarily constructs each block, and DCT, quantization, zigzag scanning, The Huffman coding algorithm is performed in sequence.

So, by cutting blocks into blocks of different sizes, you can choose the one with the best compression. At this time, the size of the matrix is stored in the compressed data header, and the dedicated player made in the ActiveX format can read the header and decompress it when executed.

Subsequently, as described above, after the virtual reality content compressed using the lossless compression algorithm and the lossy compression algorithm is stored or simultaneously with the storage, the virtual reality content may be transmitted to the web server 200 (step S140).

That is, the html file generated by the html file generating unit 156 allows the virtual reality content to be executed in the embedded state in the web browser 110 to be output from the user terminal 100 or the web server through the Internet 300. Send to (200). Subsequently, the received virtual reality content is driven in the web server 200.

By using this method, the final result of displaying the virtual reality content through the web server 200 by providing the user with a basic player capable of reading a compressed file of its own format in the form of an ActiveX control through the web server 200 is provided. To achieve.

For reference, since the virtual reality content driven in the web server 200 should have a reality that is indistinguishable from reality, the virtual reality content should be processed at a rendering of 30 frames or more per second.

Therefore, the virtual reality content generating unit 150 according to the present invention is the virtual reality content of tens of thousands of polygons applied to bump mapping (shadowing), shadowing (shadowing), etc., even in the generalized hardware specifications dozens of 30 to 99 frames per second You can render at rates higher than the frame. In addition, by storing the virtual reality content at a high compression ratio to reduce the file size, the reality can be driven in the web server 200.

In addition, the virtual reality content generation unit 150 according to the present invention enables the customer to quickly go to the desired space by enabling the expandability and applicability to the maximum access to the desired portion of the customer in the web server 200. It was.

In the above description, the present invention has been described with reference to preferred embodiments, but the present invention is not necessarily limited thereto, and a person having ordinary skill in the art to which the present invention pertains does not depart from the technical spirit of the present invention. It will be readily appreciated that various substitutions, modifications and variations can be made.

1 is a block diagram showing a virtual reality editing system according to an embodiment of the present invention;

2 is a block diagram showing a virtual reality content generating unit according to an embodiment of the present invention;

3 is an exemplary view illustrating a state in which a virtual reality content generating unit is displayed on a user terminal according to an embodiment of the present invention;

4 is a flowchart illustrating a compression method in a virtual reality editing system according to an embodiment of the present invention;

5 is an exemplary diagram illustrating an LZ77 compression algorithm according to an embodiment of the present invention;

6 is an exemplary diagram illustrating a Hoffman coding algorithm according to an embodiment of the present invention;

7 is an exemplary diagram illustrating a lossy compression algorithm according to an embodiment of the present invention;

8 is an exemplary view showing virtual reality content displayed on a web server according to an embodiment of the present invention.

Claims (5)

(a) reading the three-dimensional model data produced by the three-dimensional application to create a virtual reality content; (b) losslessly compressing the internal information data of the virtual reality content; And and (c) lossy compressing the external information data of the virtual reality content. In step (b), (b1) compressing internal information data of the virtual reality content with an LZ77 algorithm; And (b2) compressing the compressed data with Hoffman coding; In step (c), (c1) dividing the external data of the virtual reality content into a matrix having a predetermined size to form blocks respectively; (c2) discrete cosine transforming the matrix data of each block; (c3) quantizing each discrete cosine transformed matrix data; (c4) zigzag scanning each of the quantized matrix data; And (c5) compressing the zigzag-scanned data with Hoffman coding; Step (c1), (c11) reading the size of each block matrix preset by the user from the configuration file; And (c12) dividing the external data of the virtual reality content by the size of each preset block matrix to form blocks respectively; In step (c), Repeating steps (c1) through (c5) for at least one matrix size; Selecting the size of the compressed data and the matrix having the smallest size by comparing the size of the repeatedly compressed data with each matrix size; And And recording and storing the size of the matrix used in the compression in the header of the selected compressed data. delete delete delete delete

Images