RU2420912C1 - Method of distributing and transcoding video content - Google Patents

Abstract

FIELD: information technology. ^ SUBSTANCE: modified time-based inverse discrete cosine transform is used, where said transform is performed by one operation for multiplying matrices of all coefficients obtained through reverse scanning on all planes of the cube, by single common requantisation matrix, elements of which are formed from elements of dequantisation matrices for time-based inverse discrete cosine transform and MPEG quantisation. ^ EFFECT: efficient distribution and transcoding of video data for generating video sequences containing Intra-frames from 3D discrete cosine transform video programs owing to time-based modified inverse cosine transform. ^ 2 dwg

Description

The proposed technical solution relates to the field of television and digital video processing, in particular to methods for decoding compressed video data, and is intended for the design of decoding systems based on three-dimensional discrete cosine transform (hereinafter DCT-3D) video data.

Known and widely used is the method of encoding and decoding MPEG-4 video data, based on a two-dimensional discrete cosine transform to eliminate spatial frame redundancy and use complex motion prediction and compensation algorithms to eliminate temporal redundancy of video sequences [J. Richardson. World of digital processing. Video coding. N. 264 and MPEG 4 are the standards of the new generation. - M .: Technosphere, 2006. - 113 p.].

In the known method, at the decoding stage, video data recovery is carried out only for programs encoded by MPEG-4, and is characterized by high implementation complexity, which complicates its hardware implementation when creating video systems on a chip.

Closest to the claimed in its technical essence is a method of transcoding MPEG AVC - MPEG-2 (Cisco) for receiving MPEG AVC signals and converting them to MPEG-2 format (www.cisco.com or www.cisco.ru).

The main disadvantage of the prototype method is the inability to transcode programs encoded DCT-3D.

The technical result of the proposed method is to ensure the effective distribution and transcoding of video data to form video sequences containing Intra frames from video programs encoded by DCT-3D by performing a modified inverse cosine transformation in time.

This is achieved in that the method for distributing and transcoding video content using a three-dimensional discrete cosine transform is characterized in that in the inventive method, after receiving an elementary stream encoded by a three-dimensional discrete cosine transform, entropy decoding, a modified inverse discrete cosine transformation in time, direct scanning and entropy coding of the obtained coefficients for the formation of an elementary stream with MPEG standard, moreover, the modified inverse discrete cosine transformation in time is performed by one operation of multiplying the matrix of all coefficients obtained by inverse scanning on all planes of the cube by a once-generated common quantization matrix, the elements of which are formed by multiplying the elements of the dequantization matrix by the elements of the matrix of the inverse discrete cosine transformation in time and then dividing the resulting products into elements of the MPEG quantization matrix.

The presented drawing explains the essence of the proposed technical solution.

The drawings show a structural diagram of a system for distributing and transcoding compressed video data based on a three-dimensional discrete cosine transform, with which this method is implemented.

The proposed method can be implemented as follows.

The transcoding system for compressed video data based on DCT-3D comprises (Fig. 1a) a transcoder device 10 including a series-connected entropy decoding unit 11, a modified reverse DCT-1D unit (ODKP-1D) 12, a direct MPEG scanning unit 13, an entropy encoding unit 14 and a unit 15 for generating elementary streams, and a block 12 (Fig. 1b) includes serially connected block 121 for reverse scanning and dequantization, block 122 ODKP-1D in time and block 123 for quantizing MPEG.

The distribution and transcoding system of compressed video data based on DCT-3D works as follows.

The DKP-3D elementary stream arriving at the input of the transcoder block 10 is subjected to entropy decoding, modified by ODKP-1D, direct scanning, entropy coding, and the formation of an MPEG elementary stream in blocks 11, 12, 13, 14, and 15, respectively.

Further, to describe the essence of the method, it is necessary to disclose the operating principle of block 12 of the modified ODKP-1D, which allows transcoding video data from programs encoded by the DCT-3D method into quantized transformants to form an elementary MPEG stream. Block 12 operates as follows.

For the distribution and transcoding of video content using three-dimensional discrete cosine transform, it is necessary to perform the three most expensive operations:

a) first, all F * _DKP-3D transformants (x, y, t) on all planes of the cube are backscanned and dequantized in block 121 by multiplying by the elements of the _DKP-3D quantization matrix q (x, y, t):

where F _DCT-3D (x, y, t) are the dequantized coefficients of DCT-3D; x, y - frequency-spatial coordinates, at - time-frequency coordinate of the transform F *;

b) perform the reverse DCT-1D in time block 122:

, где

where

;

;

c) the obtained new values of the coefficients are again quantized by block 123, but now according to the MPEG method, by dividing by the corresponding and pre-calculated coefficients of the q matrix _MPEG (x, y):

.

.

In the proposed method, these three operations are combined (i.e., a, b and c) to obtain the formula for a simpler modified inverse DCT-1D in time with quantization by block 12 (Fig. 1a):

.

.

Features of the implementation of the proposed method are:

- reduced traffic in global networks for the distribution of data and the distribution of video content and, consequently, the flow rate at the transcoder input by 1.5 ... 2 times for standard definition video and 2 ... 4 times for high definition video compared to existing methods for transcoding video content;

- ensuring the redistribution of traffic between two types of video data transmission networks in which traffic saving is required (global Internet, satellite broadcasting, IP-television), and networks in which this restriction is not required;

- ease of implementation of quantization, which consists in replacing the three matrices of dequantization DKP-3D, inverse cosine transformation and quantization of MPEG into one matrix, the elements of which are calculated once when changing the quantization parameter by the formula (1);

- improving the quality of the video by decoding DCT-3D and the formation of Intra frames (with the least loss) from them in accordance with the MPEG standard.

Most decoders used in practice can receive video signals of programs encoded only by the MPEG format, and the proposed method allows you to receive video signals of programs encoded by the DCT-3D method, which makes it popular and universal for delivering high-definition video content via satellite channels with MPEG video compression technology.

Claims (1)

A method for distributing and transcoding video content using a three-dimensional discrete cosine transform, characterized in that in the inventive method, after receiving an elementary stream encoded by a three-dimensional discrete cosine transform, entropy decoding, modified inverse discrete cosine time conversion, forward scanning and entropy encoding of the obtained coefficients are performed sequentially for the formation of an elementary stream of the MPEG standard, and The differential inverse discrete cosine time transformation is performed by one operation of multiplying the matrix of all coefficients obtained by reverse scanning on all planes of the cube by a once-generated common quantization matrix, the elements of which are formed by multiplying the elements of the dequantization matrix by the elements of the matrix of the inverse discrete cosine transformation in time and then dividing the resulting works on the elements of the MPEG quantization matrix.

Images