JP2011193486A - System and method for stereoscopic 3d video image digital decoding - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and system for digitally decoding an MPEG (Moving Picture Expert Group)2 compatible 3D stereoscopic video image having an original codification algorithm. <P>SOLUTION: By performing a hardware modification and a software modification in different portions of a decoding process in order to obtain a 3D image in a digital video stream, a structure and a video sequence of the video stream are modified for including a flag needed to identify a type of the TDVision(R) technology image at bit level. The modification is performed while decoding information via e.g., software and hardware in the decoding process, double output buffer are activated, a parallel decoding selector and a differential decoding selector are activated, a decompression process is executed and an image is displayed in the corresponding output buffer. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to stereoscopic video image display on a 3D Visor® device, and in particular, a video image with a digital data compression system that enables storage of 3D information by using standardized compression techniques. It relates to a decoding method.

  Currently, data compression techniques are used to reduce the bit consumption of a representation of an image or series of images. Standardization work was performed by a group of experts from the International Organization for Standardization. Currently, these methods are commonly referred to as JPEG (Joint Photographic Expert Group) and MPEG (Moving Pictures Expert Group).

  A common property of these techniques is that the image block is processed by applying an appropriate transform to the block, commonly referred to as the discrete cosine transform (DCT). The formed block is submitted to a quantization process and then encoded using a variable length code.

  The variable length code is a reversible process and allows accurate reconstruction of what was coded using the variable length code.

  The display of the digital video signal includes a number of image frames (30 fps to 96 fps) that are continuously displayed or presented at a frequency of 30 Hz to 75 Hz. Each image frame is still an image formed by the pixel array according to the display resolution of the particular system. For example, a VHS system has a display resolution of 320 columns and 480 rows, an NTSC system has a display resolution of 720 columns and 486 rows, and a high definition television system (HDTV) has a display resolution of 1360 columns and 1020 rows. Have. For a 320 column 480 row VHS format, a low resolution digitized format, a two hour movie can be equivalent to 100 gigabytes of digital video information. In comparison, the conventional compact optical disk has a capacity of about 0.6 gigabytes, the magnetic hard disk has a capacity of 1 to 2 gigabytes, and the current compact optical disk has a capacity of 8 gigabytes or more. .

  Every image we see on a movie or TV screen is based on the principle that it presents a complete image (still like a photo) at high speed. When they are presented in rapid and sequential form at a rate of 30 frames per second (30 fps), we perceive them as moving images due to the retention of the human eye.

  In order to classify images presented in sequential form and form a video signal, each image needs to be divided into rows, where each row is divided into pixels or pixels, Has associated values, luma and chroma. The luma represents the light intensity at each point, the luma represents the color as a function of a defined color space (RGB), and this color space can be represented by 3 bytes.

  Images are displayed on the screen periodically from top to bottom and left to right in a horizontal and vertical raster. The number of scan lines and the frequency of display can vary as a function of formats such as NTSC, PAL, or SECAM.

  This process can be digitized for storage in digital format after transmission, reception, and decoding for display on a normal television set or display device such as 3D Visor®, this process Is referred to as analog-digital video signal coding-decoding.

  By definition, MPEG has two different ways to interlace video and audio into the system stream.

  Transport streams are used in systems with a higher likelihood of error, such as satellite systems that are susceptible to interference. Each package is 188 bytes long and begins with an identification header that allows gap recognition and error correction. Various audio and video programs can be transmitted simultaneously over a single transport stream over a transport stream, and due to the header, these can be independently decoded and integrated into multiple programs can do.

  Program streams are used in systems that have a lower likelihood of error, such as during DVD playback. In this case, the package has a variable length and has a substantially larger size than the package used in the transport stream. As a key characteristic, the program stream allows only a single program content.

  Even when the transport stream and program stream handle different packages, the video and audio formats are decoded in the same way.

  Next, there are three compression types that apply to the above packages, such as temporal prediction, compression, and spatial compression.

  Decoding is associated with a lengthy mathematical process, the purpose of which is to reduce the amount of information. A full image of a full frame is divided into units called macroblocks, each macroblock being composed of a 16 pixel by 16 pixel matrix, ordered from top to bottom, left to right, and named. Even when using a matrix array on the screen, the information sent via the information stream follows a special sequential sequence, i.e. the macroblocks are ordered in ascending order, i.e. macroblock0, macroblock1, etc.

  A set of consecutive macroblocks represents a slice. If a macroblock relates to a single row, any number of macroblocks can be provided in one slice. Similar to macroblocks, slices are numbered from left to right and from bottom to top. The slice must cover the entire image. This is because MPEG2 is a form of compressing video, and the coded image does not necessarily require samples per pixel. Some MPEG profiles require processing of a fixed slice structure, which must cover the entire image.

  U.S. Pat. No. 5,963,257, issued Oct. 5, 1999 to Katata et al., Separates coded data by location area and image format, lower layer code, predictive coding upper layer code, and thus coded data Protecting a flat video image decoding device having means for obtaining a hierarchical structure, the decoder comprises means for separating the data coded in the hierarchical structure in order to obtain a high quality image.

  U.S. Pat. No. 6,292,588 issued September 18, 2001 to Shen et al., In which the reconstructed flat image data is generated from the sum of the image data of a small area and the optimal prediction data of the image, Protect devices and methods for coding predictive flat images that are reconstructed from small areas and decoded. The predictive decoding device for the image data stream includes a variable length code of single dimensional DCT coefficients. US Pat. No. 6,370,276, granted to Boon on April 9, 2002, uses a decoding method similar to the above.

  U.S. Pat. No. 6,456,432, issued September 24, 2002 to Lazzaro et al., Protects a stereoscopic 3D image display system, which takes images from two perspective views and displays them on a CRT for viewing. These images are multiplexed in a field-sequential manner, without flicker for both eyes.

  U.S. Pat. No. 6,658,056, issued Dec. 2, 2003 to Durouz et al., Includes a digital display section corresponding to a "neighbor field" command to obtain a digital video field at a specified location in output memory. Protect the video decoder. The digital video display system includes an MPEG2 video decoder. The image is decoded as a memory buffer, which is optimized, maintains a compensation variable table, and accesses a fixed memory pointer table that is displayed as a data field.

  US Pat. No. 6,665,445 granted to Boon on December 16, 2003 protects data structures for image transmission, flat image coding methods, and flat image decoding methods. This decoding method consists of two parts, the first part classifies the image format information data stream, the second part is a decoding process relating to the pixel values of the image data stream, both parts being flat It can be switched according to the image signal coding.

  U.S. Pat. No. 6,678,331, issued January 13, 2004 to Moutin et al., Protects MPEG decoders that use shared memory. In practice, the circuit includes a microprocessor, an MPEG decoder that decodes a flat image sequence, a common memory for the microprocessor, and a decoder. This circuit also includes a circuit that evaluates the decoder delay and a control circuit that determines the memory priority for the microprocessor or decoder.

  US Pat. No. 6,678,424 granted to Ferguson on January 13, 2004 protects the behavior model of a real-time human vision system. In practice, this model processes two two-dimensional image signals in different channels, one derived from the other.

  It is an object of the present invention to provide a stereoscopic 3D video image digital decoding system and method consisting of software changes and hardware changes.

  An additional object of the present invention is that the normal video_sequence process is applied to the coded image data, ie, the variable_length_decoding (VLD), the inverse_scan, the inverse_quantization, the inverse_discrete_cosine_transform (encoding), and the encoding_smp .

  Another object of the present invention is to decode video format identification, 2D image MPEG2 backward compatibility, TDVision® type image discrimination, last image buffer storage, application of information decoding, error correction Applying and changing software information related to storing results in each channel buffer.

  Another object of the invention is to have the normal form of a video_sequence process, with the last complete image buffer being stored in the left or right channel buffer when a TDVision® type image is found. It is to provide a decoding method.

  Another object of the present invention is to provide a decoding process in which two mutually dependent (different) video signals can be sent in the same video_sequence, and that information decoding is applied and stored as a B-type frame. Is to provide.

  Another object of the present invention is to provide a decoding process in which error correction is applied to the last obtained image when motion vectors and color correction vectors are applied.

  Another object of the present invention is to program the decoder by software to receive and classify two independent program streams simultaneously.

  Another object of the present invention is to provide a decoding system that decodes 3D image information via hardware in which a dual output buffer is activated.

  Another object of the present invention is to provide a 3D image information decoding system that activates image decoding selectors in parallel.

  Another object of the present invention is to provide a 3D image information decoding system that performs a decompression process and displays a corresponding output buffer.

2 represents a technology map to which the subject purpose of the present invention relates. Figure 2 shows a flow diagram outlining the steps of the process. Fig. 4 shows an overview of the structure to be changed and the video_sequence of the data stream for identifying the TDVision ™ technology image type at the bit level. Fig. 6 shows a compiled software format (40) for a TDVision (R) decoding method. Decoding compilation format in the hardware section.

  The combination of hardware and software algorithms enables stereoscopic 3D image information compression, which sends 3 simultaneous programs with stereoscopic pair identifiers, thus accelerating the coding-decoding process. ) Received as video signals having two independent but identical time_codes corresponding to the left and right signals coming from the camera. Also, two interdependent video signals can be processed by obtaining the difference, which is stored as a “B” type frame with an image type identifier. Since the coding process is open to facilitate technology development, this decoding process, i.e. applying variable length decoding to the coded data (substantial reduction is obtained, but the lookup table Decoding), applying an inverse scan process, applying an inverse quantization process that applies a scalar to each data, applying an inverse cosine transform function, applying an error correction or motion compensation stage It is only necessary to follow the final obtaining of the decoded image.

  The novel features of the invention relating to structure and method of operation are better understood from the description of the accompanying drawings in conjunction with the accompanying specification, wherein like numerals refer to like parts and steps.

  FIG. 1 represents a technology map to which the subject purpose of the present invention pertains. FIG. 1 shows a stereoscopic 3D image coding and decoding system and corresponding method. The image comes from the stereo camera (32) and the information is compiled (31) and displayed by the appropriate system (30) or (33). The information is coded at (34) and can be transmitted to a system with a suitable previous decoding stage, such as (35), which includes a cable system (36), a satellite system (37 ), High-definition television system (38), or 3DVisions (registered trademark) (39) of TDVision (registered trademark).

  FIG. 2 shows a flow diagram outlining the steps of the process. The purpose is to obtain a 3D image from a digital video stream by making changes to the current MPEG2 decoder and to the software (3) and hardware (4) of the decoding process (2), and the decoder (1 ) Must be compatible with MPEG2-4.

  FIG. 3 shows an overview of the structure to be changed and the video_sequence of the data stream to identify the TDVision® technology image type at the bit level.

  Each of the stages of the decoding process is described in detail below (20).

  The coded data (10) is a byte with block information, macroblocks, fields, frames, and an MPEG2 format video image.

  Variable_length_decoding (11) (VLC, variable length decoder) is a compression algorithm in which the most frequent pattern is replaced by a shorter code and a pattern that does not occur frequently is replaced by a longer code. The compressed version of this information takes up less space and can be transmitted faster over the network. However, this is not an easily editable format and requires decompression using a lookup table.

For example, the word BEETLE has the following characters: ASCII code VLS
B 0100 0010 0000 0010 10
E 0110 0101 11
L 0110 1100 0001 01
T 0111 0100 0100

  Therefore, the ASCII code for this word is 0100 0010 0110 0101 0110 0101 0111 01000 0110 1100 1110 0110 0101 and 0000 0010 10 11 11 0100 0010 01 11 in VLC.

  Although a substantial decrease is observed, in order to return from the VLC to the word “Beetle”, a lookup in the lookup table is required to decode the bitstream, which is due to an accurate comparison of the read bits. Done.

  Reverse scan (12): Information must be grouped by block, and by coding the information using VLC, a linear stream is obtained. The block is an 8 × 8 data matrix, so the linear information needs to be converted to a square 8 × 8 matrix. This is done in the form of a zigzag descending from top to bottom and from left to right in both sequence types, depending on whether it is a progressive image or an interlaced image.

  Inverse quantization (13): This simply consists of multiplying each data value by a coefficient. When classified, most of the data in the block is quantized to remove information that cannot be perceived by the human eye, and this quantization allows to obtain a larger MPEG2 stream transform, and in the decoding process It is also necessary to perform an inverse process (inverse quantization).

  Inverse Cosine Transform (14) (IDCT, inverse_discrete_cosine_transform): The data processed in each block is related to the frequency domain, and this inverse cosine transform makes it possible to return to the spatial domain samples. Once the data is converted with IDCT, pixels, colors, and color corrections can be obtained.

  Motion compensation (15) makes it possible to correct some errors generated before the decoding stage of the MPEG format, motion compensation takes the previous frame as a reference and calculates a motion vector for the pixel. These can be used to create new images (up to four vectors can be calculated). This motion compensation is applied to P-type and B-type images, where the image position is located relative to the “t” time from the reference image. In addition to motion compensation, error correction is also applied. This is because it is insufficient to predict the position of a particular pixel, and its color change may exist. Thus, a decoded image is obtained (16).

  In order to decode a P-type image or a B-type image, a reference image is taken, motion vectors are added algebraically to calculate the next image, and finally error correction data is applied, thus decoding. The generated image is successfully generated. In fact, within the video_sequence there are two interdependent video signals “RL = Delta, where the delta difference is stored as a B-type stereo pair frame with a TDVision® identifier, At the moment of decoding, it is composed by the difference from the image, ie R-delta = L and L-delta = R, the left image is composed of the difference from the right image, the right image is the left image And the difference.

  The previous process takes the left or right signal, both are stored in a temporary buffer, then the difference between the left and right signals is calculated and then decoded later by the difference from the image, It was outlined in the form of being coded as a B-type image stored in video_sequence.

  In the decoding process, it can be deduced that the data input by the VLC stage is much smaller than the data output by the same stage.

MPEG Video Sequence Structure This is the largest structure used in the MPEG2 format and has the following format:
Video sequence (Video_Sequence)
Sequence header (Sequence_Header)
Sequence extension (Sequence_Extension)
User data (0) and extension (Extension_and_User_Data (0))
Image group header (Group_of_Picture_Header)
User data (1) and extension (Extension_and_User_Data (1))
Image header (Picture_Header)
Coded image extension (Picture_Coding_Extension)
User data (2) and extension (Extension_and_User_Data (2))
Image data (Picture_Data)
Slice
Macroblock
Motion vector (Motion_Vectors)
Coded block pattern (Coded_Block_Pattern)
Block
Final sequence code (Sequence_end_Code)

  These structures constitute a video sequence. The video sequence applies to the MPEG format, and to distinguish between versions, there must be a verification that a sequence extension exists immediately after the sequence header, and if the sequence extension does not follow the header, the stream Is the MPEG1 format.

  At the beginning of the video sequence, sequence_header and sequence_extension appear in video_sequence. The sequence_extension must be the same in the first trial, and the “s” iteration of the sequence_header should change little compared to the first occurrence, and only the part defining the quantization matrix must change. . Having sequence repetition allows random access to the video stream, i.e. this can be done if the decoder wants to start playing in the middle of the video stream. This is because in order to decode the next image, it is only necessary to find the previous sequence_header and sequence_extension. This also occurs for video streams that cannot start from the beginning, such as a satellite decoder tuned after the transmission time.

The full video signal coding-decoding process consists of the following steps.
Digitizing a video signal. This can be done in NTSC format, PAL format, or SECAM format.
Storing a video signal in digital format Sending a signal Recording a digital video stream on a physical medium (DVD, VCD, MiniDV) Receiving a signal Playing a video stream Decoding a signal Displaying a signal

  It is essential to have the possibility of doubling the memory processed by a suitable DSP and placing up to eight output buffers, thereby allowing devices such as TDVision® 3D Visor® Enables the previous and simultaneous representation of stereoscopic images.

In practice, for example, in the TMS320C62X DSP example case from Texas Instruments, two channels must be initialized when calling the DSP programming API.
MPEG2VDEC_create (const IMPEG2VDEC_fxns * fxns, const MPEG2VDEC_Params * params)

Here, IMPEG2VDEC_fxns y MPEG2VDEC_Params is a pointer structure that defines the operation parameters of each video channel.
3DLhandle = MPEG2VDEC_create (fxns3DLEFT, Params3DLEFT)
3DRhandle = MPEG2VDEC_create (fxns3DLIGHT, Params3DLIGHT
It is.

  This makes it possible to decode two video channels, one for each left and right stereo channel, resulting in two video handlers.

A dual display output buffer is required and the software defines which of the two buffers should display the output by calling the AP function.
That is, MPEG2VDEC_APPLY (3DRhandle, inputR1, inputR2, inputR3, 3doutright_pb, 3doutright_fb)
MPEG2VDEC_APPLY (3DLhandle, inputL1, inputL2, inputL3, 3outleft_pb, 3outleft_fb)

  This same procedure can be implemented for all DSPs, microprocessors, or electronic devices with similar functions.

  Here, 3DLhandle is a pointer to the handle returned by the DSP creation function, input1 parameter is the address of FUNC_DECODE_FRAME or FUNC_START_PARA, input2 is a pointer to the external input buffer address, and input3 is the external input The size of the buffer.

  3dleft_pb is the address of the parameter buffer, and 3dleft_fb is the head of the output buffer where the decoded image is stored.

  Timecode and timestamp are used for output to the final device in a sequential synchronous fashion.

  It is essential to double the memory processed by the DSP and have the possibility of placing up to eight output buffers, which allows devices such as 3D Visor (registered trademark) from TDVision (registered trademark) Corporation Previous display and simultaneous display of stereoscopic images are possible.

  The integration of software and hardware processes is performed by a device called a DSP, which performs most of the hardware processes. These DSPs are programmed by a hybrid of C and assembly language supplied by the manufacturer. Each DSP has its own API that consists of a function list or procedure call that is placed in the DSP and called by software.

  This reference information is used to create an application relating to MPEG2 format compatible 3D image decoding.

  In practice, a sequence header (sequence_header) and a sequence extension always appear at the beginning of a video sequence. The sequence extension iteration must be the same as the first. In contrast, sequence header iterations change little compared to the first occurrence, and only the part defining the quantization matrix must change.

  FIG. 4 shows the compiled software format (40) of the TDVision® decoding method, where the video_sequence (41) of the digital stereoscopic image video stream is identified in the sequence_header (42), which depends on Or it can be independent (parallel image). If the image is TDVision®, the double buffer is activated and the change in aspect_ratio_information is identified. Information corresponding to the image, which can be found here, is read into user_data (43). The sequence_scalable_extension (44) defines the information contained in it, the base layer and the enhancement layer, the video_sequence can be located here, the scalable_mode and the layer identifier are defined, and the extra_bit_picture (45) defines the picture_eture , Picture_coding_extension (46) reads the “B” type image and decodes the second buffer if it is a TDVision® type image. picture_temporal_scalable_extension () (47) is used to decode a B-type image if it has temporal scalability.

That is, the sequence header (sequence_header) provides a higher level of information for the video stream, and for the sake of clarity, the corresponding number of bits is also shown, and the higher order bits are placed in the sequence extension (Sequence_Extension) structure, Formed by the following structure.

Picture_temporal_scalable_extension ()
In the case of temporal scalability, there are two spatial resolution streams, the lower layer provides a smaller index version of the video frame, and the upper layer derives a larger index version of the same video frame. Can be used for Temporal scalability can be used by low quality, low cost, or free decoders, with higher frames per second being used for a fee.

In the case of image space scalability, data is included in the enhancement layer, which allows for a better resolution of the base layer and can thus reconstruct a better resolution of the base layer. When the enhancement layer is used as a function of the base layer as a reference for motion compensation, the lower layer must be escalated and offset to obtain a higher resolution of the enhancement layer.

EXTENSION_AND_USER_DATA (2)
The image is
DVD (digital versatile disc)
DTV (digital television)
HDTV (High Definition Television)
Cable (DVB, digital video broadcasting)
Satellite (DDS, digital satellite system)
This is the integration of software and hardware processes.

  The decoding compilation format (50) in the hardware section of FIG. 5 is duplicated in the DSP input memory and at the same time two independent stereo signals corresponding to the left and right stereo existing signals taken by the stereo TDVision® camera. Simultaneous or dependent video signals are allowed. In this procedure, video_sequence (51) is detected to alternate left and right frames or send them in parallel, sequence_header (52) identification, image type (53) is identified and passed to normal video stream (54). And then submitted to the error correction process (55), the video image information is sent to the output buffer (56), which shares this information and the video stream information is in 3D or 2D. Distribute to the left channel (57) and right channel (58) within the displayed channel.

  It is to store both L (left) and R (right) video streams in a simultaneous format as video streams synchronized to two independent but identical time_codes, so they are later decoded and more Can be played back in parallel in a large output buffer. These can be made dependent and declassified by difference.

  With regard to hardware, most of the process is performed by a device called a DSP (Digital Signal Processor). By way of example, for example, the Motorola model and the Texas Instruments model (TMS320C62X) can be used.

  These DSPs are programmed with a hybrid of C and assembly language supplied by the manufacturer. Each DSP has its own API that consists of a function list or procedure call that is placed in the DSP and called by software. From this reference information, a 3D image is coded, which is compatible with the MPEG2 format and its own coding algorithm. When the information is coded, the DSP is responsible for performing a prediction process, a comparison process, a quantization process, and a DCT function application process to form an MPEG2 compressed video stream.

  In order to obtain a 3D image from a digital video stream, certain changes to the current MPEG2 decoder have been made by software and hardware changes in different parts of the decoding process. The structure and video_sequence of the video data stream must be changed to include the flags necessary to identify the TDVision ™ technology image type at the bit level.

  This change is made in the next decoding step.

Software-Video format identification.
-Application of logic "and" for MPEG2 backward compatibility when not TDVision (R) video.
-Image decoding in the usual way (previous technique), scanning video_sequence.
-For TDVision (R) type images,
Distinguishing whether they are subordinate or independent video signals.
Store the last complete image buffer in the left or right channel buffer.
Apply B-type frame information decoding.
Apply error correction to the last obtained image by applying motion vectors and color correction vectors.
Store the result in each channel buffer.
_ Continue video sequence reading.

When hardware information is decoded through the hardware,
Distinguish whether the image is 2D or 3D.
_Activate the dual output buffer (increase memory).
_Activate the difference decoding selector.
Activate the parallel decoding selector.
Perform the decompression process.
Display the image in the corresponding output buffer.

The following structures, substructures, and sequenced are used in a specific way and belong to the video_sequence structure of the hardware implementation of MPEG2 backward compatible TDVision® technology.
actually,
Sequence_header
Aspect_ratio_information
1001 Not applicable with TDVision (registered trademark) 1010 TDVision (registered trademark) 4: 3
1011 TDVision® 16: 9
1100 TDVision (R) 2.21: 1

An indication that the logical “and” with 0111 is performed to obtain backward compatibility with 2D systems, but when this is done, the solid pair (left or right) buffer must be equal to the source Are sent to the DSP, so all images to be decoded are sent to both output buffers to allow image display on all devices.
Frame_rate_code
1001 TDVision® format 24000/101 (23.976)
1010 24 in TDVision® format
1011 25 in TDVision® format
1100 30000/1001 (29.97) in TDVision® format
1101 30 in TDVision (registered trademark) format
1110 TDVision (registered trademark) format 50
1111 TDVision® format 60,000 / 1001 (59.94)

In order to obtain backward compatibility with 2D systems, a logical “and” with 0111 is performed.
User_data ()
Sequence_scalable_extension
Picture_header
Extra_bit_picture
0 = TDVision (registered trademark)
1 = Normal Picture_coding_extension
Picture structure 00 = Image in TDVision (registered trademark) format Picture_temporal_scalable_extension ()

  At the moment of coding information, DPS is used, which is responsible for performing the prediction, comparison, and quantization processes, applying DCT to form an MPEG2 compressed video stream, 2D image And 3D images.

  Two video signals are coded in a format that has independent but identical time_codes, these signals correspond to the left and right signals coming from the 3DVision® camera, both programs being TDVision® stereoscopic Sent together with a pair identifier. This type of decoding is referred to as “by parallel images” and stores both left and right (L and R) video streams simultaneously as two independent but time_code synchronized video streams. It is in. These are later decoded and played back in parallel. Only the decoding software needs to be decoded and the transport stream coding and compression algorithm is the same as the current one.

  Software changes at the decoder

  At the decoder, in this example case between the two program streams or both stored as a B-type frame with two interdependent video signals or identifiers according to the programming API with the use of the Texas Instruments DSP of the TMS320C62X family The video signal composed of the differences must be programmed simultaneously.

DSP Programming Algorithms and Methods-Creates two process channels when starting the DSP (left and right when calling the main buffer and secondary buffer or API).
Get the RAM memory pointer (RAM address in the memory map) for each channel.
-When obtaining a TDVision (R) type video sequence, interpret it as type B, decode the image in real time, apply changes or differences to the complementary buffer, and store the result in the secondary buffer.

The following two options are implemented for software in the video_sequence data stream:
1. One changes only the software and uses the user_data () section to store error corrections that allow the stereo signal to be regenerated.
2. The other enables the PICTURE_DATA3D () function by hardware, which is transparent to an MPEG2 compatible reader and can be decoded by a TDVision ™ compatible DSP.

  The MPEG2 decoder searches for the 3DVISION_START_IDENTIFIER = 0X000ABCD 32-bit identifier at the moment it detects the user_data () code, which is extremely expensive to recreate the code, which is difficult or does not represent data. Next, the 3D block length to be read is taken into account, which is 32-bit “n” data. When this information is detected in USER_DATA (), a call to a special decoding function is made, which is then compared to the output buffer and as a normal correction for B-type frames, the current reading offset of video_sequence That is, it is applied from n bytes. The output of this correction is sent to another output address, which is directly related to the video output added to what is present on the electronic display device.

  If the PICTURE_DATA3D () structure is recognized, the process proceeds directly to reading the information by the decoder, but the information is written to the second output buffer, which is added to what is present in the electronic display device. Also connected to the output.

  In the case of a program stream, the two signals (left and right) are synchronized by time_code, and these signals are either decoded in parallel by an MPEG decoder with sufficient simultaneous multiple video channel decoding capability, or the same video_sequence Two interdependent video signals, e.g., "RL = delta" can be sent, where delta is the difference stored as a "B" type frame with a stereo pair TDVision (R) identifier Yes, the two signals can be reconstructed at the instant of decoding by the difference from the image, as in the case of the Texas Instruments DSP mentioned above, ie "R-delta = L" or "L- Delta = R ”, but this is exemplary Not considered to be a limiting example Te.

  Video including a single video sequence is also implemented, but the left and right frames alternate at 60 frames per second (30 frames each) and when decoded, the video buffer image is placed in the corresponding left or right channel. .

  There is also the ability to detect whether the signal is of TDVision® type via hardware, in which case the signal is transport stream, program stream, or left and right multiplexed at 60 frames per second Is identified.

  In the case of transport streams, backward compatible systems can be used in current decoders that have the ability to display the same video only in 2D without 3D properties, in which case the DSP can use TDVision (registered) Trademarked) or disabled for displaying images on devices of previous techniques.

  In the case of program streams, unaltered coders such as those currently used in satellite transmission systems are used, but the receptors and decoders have a TDVision® flag identification system, so Enable the second video buffer to form.

  Finally, in the case of multiplexed video, an MPEG decoder with two video buffers (left and right) is enabled to identify the appropriate frames and separate each signal at 30 frames per second, thus flickerless images As a result, the multiplexing effect is not perceived because the video stream is constant and due to the characteristic retention wave of the human eye.

  While particular embodiments of the present invention have been illustrated and described, it would be obvious to those skilled in the art that multiple modifications or changes can be made without departing from the scope of the invention. All such modifications and changes are intended to be included within the scope of the appended claims, and as a result, all changes and modifications are intended to be included within the scope of the present invention.

Claims (32)

A system for displaying stereoscopic digital video to a user,
A plurality of headers each having data indicating whether or not a delta image derived from a difference between the first field image and the second field image exists, and a plurality of the first fields each having the corresponding header. Receiving means for receiving one or more video streams having one field-of-view image;
Reading the header with the data to determine whether the one or more video streams also have the delta image derived from a difference between the first field image and the second field image; A decoder that constructs the second field image from the delta image and the first field image if the data of the header indicates that a delta image is present;
Video output means for outputting stereoscopic digital video by the first field image and the second field image;
A system comprising: The system is part of at least one of a cable system, a high-definition television system, a satellite television system, an optical disc system, a DVD system, and a digital television system;
The system of claim 1. The one or more video streams have a time code;
The system of claim 1. The frame of the first field image and the frame of the second field image are output to a high definition display or a stereoscopic vision system;
The system of claim 1. The decoder constructs the second field image by adding the delta image to the first field image;
The system of claim 1. The decoder reads a predetermined set of bits of the header to determine whether the one or more video streams have a delta image;
The system of claim 1. The delta image is stored in video stream data specific to the delta image;
The system of claim 1. The delta image has a frame of a P-type image or a B-type image;
The system of claim 1. The one or more video streams comprise a plurality of video streams;
The system of claim 1. The decoder reads the header and the first field-of-view image from a first video stream, and reads the delta image from a predetermined data structure;
The system of claim 1. The delta image includes a plurality of motion vectors calculated by measuring a difference between the first field image and the second field image;
The system of claim 1. A system for decoding stereoscopic digital video,
A plurality of headers each having data indicating whether or not a delta image derived from a difference between the first field image and the second field image exists, and a plurality of the first fields each having the corresponding header. Receiving means for receiving one or more video streams having one field-of-view image;
Reading the header with the data to determine whether the one or more video streams also have the delta image derived from a difference between the first field image and the second field image; Means for constructing the second field image from the delta image and the first field image if the data in the header indicates that a delta image is present;
Video output means for outputting stereoscopic digital video by the first field image and the second field image;
A system comprising: The video output means outputs the stereoscopic digital video to a high-definition display or a stereoscopic vision system;
The system of claim 12. The constructing means reads the header and the first field-of-view image from a first video stream, and reads the delta image from a predetermined data structure;
The system of claim 12. The constructing means constructs the second field image by adding the delta image to the first field image;
The system of claim 12. The delta image has a P-type image or a B-type image;
The system of claim 12. The one or more video streams are MPEG compatible video streams;
The system of claim 12. The one or more video streams comprise a plurality of video streams;
The system of claim 12. The delta image is stored in a predetermined structure specific to the delta image;
The system of claim 12. The first visual field image is a P-type image;
The system of claim 12. The constructing means is executed by a digital signal processor;
The system of claim 12. The delta image includes a plurality of motion vectors calculated by measuring a difference between the first field image and the second field image;
The system of claim 1. A method for decoding stereoscopic digital video, comprising:
A plurality of headers each having data indicating whether or not a delta image derived from a difference between the first field image and the second field image exists, and a plurality of the first fields each having the corresponding header. Receiving one or more video streams having one field of view image;
Reading the header with the data to determine whether the one or more video streams also have the delta image derived from a difference between the first field image and the second field image; Configuring the second field image from the delta image and the first field image if the data in the header indicates that a delta image exists; and
Outputting a stereoscopic digital video based on the first field image and the second field image;
Including methods. The stereoscopic digital video is output to a high definition display or stereoscopic vision system;
24. The method of claim 23. The header and the first field-of-view image are read from a first video stream, and the delta image is read from a predetermined data structure;
24. The method of claim 23. The delta image is added to the first field image to form the second field image,
24. The method of claim 23. The one or more video streams are MPEG compatible video streams;
24. The method of claim 23. The delta image has a P-type image or a B-type image;
28. The method of claim 27. The one or more video streams comprise a plurality of video streams;
24. The method of claim 23. The delta image is stored in a predetermined structure specific to the delta image;
24. The method of claim 23. The first visual field image is a P-type image;
24. The method of claim 23. The delta image includes a plurality of motion vectors calculated by measuring a difference between the first field image and the second field image;
24. The method of claim 23.

Images