CN105915912A - Fragile watermark embedding and detection method of MPEG-2 compressed domain video - Google Patents

Abstract

The present invention discloses a fragile watermark embedding and detection method of an MPEG-2 compressed domain video. An MPEG-2 decoder is employed to analyze the MPEG-2 video code flow according to the MPEG-2 video coding standard, the nonzero coefficient number of a brightness block Y component, a chroma block U component and a V component through DCT transformation and quantification are subjected to statistics according to the macro-block data of the I-frame video, the energy difference statistical value of the brightness block Y component is calculated according to the mapping relation; and the binary system watermarking information is generated through the Arnold transformation to embed the video code flow. The fragile watermark embedding and detection method of an MPEG-2 compressed domain video generates watermarking according to the video macro block to provide reliable watermarking embedding and detection for MPEG-2 digital products and is able perceive tiny attack; the watermarking embedding detection based on the compressed domain has no influence on the video quality, the calculation complexity is low, and the timeliness is good; and the fragile watermark embedding and the detection method of MPEG-2 compressed domain video performs watermarking detection with no need for original video and belongs to a blind watermarking method.

Description

A kind of MPEG-2 compressed domain video fragile watermark embeds and detection method

Technical field

The present invention relates to a kind of video compress territory fragile watermark, in particular a kind of MPEG-2 compressed domain video fragile watermark embeds and detection method.

Background technology

Video Fragile Watermarking Technique is an important branch in Investigation of Information Hiding Technology field.The fast development of Computer Multimedia Technology, facilitates storage, edits, transmits the multimedia messages such as video, also caused transmission safety and the Copyright Protection of multimedia messages simultaneously.How to prevent multimedia messages from illegally distorting, forging, ensure data integrity so that video fragile watermark becomes the study hotspot of field of multi-media information safety.

MPEG-2 is the video organized to set up by MPEG and audio frequency lossy compression method standard, it is particularly well-suited to coding and the transmission of the DTV of broadcasting-quality, it is widely used in wireless digital TV-set, DVB (Digital Video Broadcasting, DVB), digital satellite television, DVD (Digital Video Disk, Digital video disc) etc..

The video vulnerable watermark method having pointed out at present has:

(1) uncompressed domain fragile watermark, watermark is directly embedded in original video image by the program, forms the original video containing watermark information, carries out Video coding the most again.The program may utilize existing great amount of images Processing Algorithm and embeds watermark, and algorithm easily realizes.

(2) encoding domain fragile watermark, the program utilizes the DCT coefficient of amendment video to embed watermark when coding.Owing to watermark embedding operation is carried out in an encoding process, the bit rate of video flowing is affected less by the program, and is prone to utilize corresponding video encoding standard characteristic to embed watermark.

(3) compression domain fragile watermark, the program directly embeds or detection watermark in compression domain.Embedding watermark in video in encoded need not decoding and recompile, algorithm complex is relatively low.

But uncompressed domain fragile watermark and encoding domain fragile watermark are required to first be decoded for the video information compressed, recompile after embedding watermark, not only reduce video quality, and degree of calculating is complicated, be unsatisfactory for the real-time of video code flow transmission.Existing compression domain fragile watermark mainly embeds watermark by modulation movement vector or medium and low frequency DCT coefficient, have modified video information, there is real-time and video quality such as damages at the problem.

Summary of the invention

It is an object of the invention to overcome the deficiencies in the prior art, it is provided that a kind of MPEG-2 compressed domain video fragile watermark embeds and detection method, provide reliable watermark to embed and detection for MPEG-2 digital product.

The present invention is to be achieved through the following technical solutions:

A kind of MPEG-2 compressed domain video fragile watermark embedding grammar, it comprises the following steps:

(1) utilizing MPEG-2 decoder to resolve MPEG-2 video code flow according to MPEG-2 video encoding standard, setting I-frame video has N number of macro block, then adding up luminance block Y-component nonzero coefficient number after dct transform and quantization in N number of macro block is M_Y；

(2) chrominance block U component nonzero coefficient number M after dct transform and quantization in N number of macro block is added up_U；

(3) chrominance block V component nonzero coefficient number M after dct transform and quantization in N number of macro block is added up_V；

(4) luminance block Y-component energy difference statistical value M is added up according to mapping relations_E；

(5) by M_Y、M_U、M_VAnd M_ECarry out 16 binarizations respectively to process, and form 64 binary sequences, encrypt through Arnold conversion, obtain binary system encrypted watermark M_K；

(6) by binary system encrypted watermark M_KIt is embedded into current I-frame video bit stream data afterbody, obtains the video flowing containing watermark.

Scheme, described step (1) are added up luminance block Y-component nonzero coefficient number M after dct transform and quantization further in N number of macro block_YMethod comprise the following steps:

(1) luminance block Y-component nonzero coefficient number Y after dct transform and quantization in i-th 8x8 macro block is first added up_i, wherein, i=1,2 ..., N；

(2) further according to Y_iCalculate luminance block Y-component nonzero coefficient number M after dct transform and quantization in N number of macro block_Y:

$M_Y=\underset{i=1}{\overset{N}{\Σ}}{Y}_i$

Scheme, described step (2) are added up chrominance block U component nonzero coefficient number M after dct transform and quantization further in N number of macro block_UMethod comprise the following steps:

(1) chrominance block U component nonzero coefficient number U after dct transform and quantization in i-th 8x8 macro block is first added up_i, wherein, i=1,2 ..., N；

(2) further according to U_iCalculate chrominance block U component nonzero coefficient number M after dct transform and quantization in N number of macro block_U:

$M_U=\underset{i=1}{\overset{N}{\Σ}}{U}_i$

Scheme, described step (3) are added up chrominance block V component nonzero coefficient number M after dct transform and quantization further in N number of macro block_VMethod comprise the following steps:

(1) chrominance block V component nonzero coefficient number V after dct transform and quantization in i-th 8x8 macro block is first added up_i, wherein, i=1,2 ..., N；

(2) further according to V_i, calculate chrominance block V component nonzero coefficient number M after dct transform and quantization in N number of macro block_V:

$M_V=\underset{i=1}{\overset{N}{\Σ}}{V}_i$

Scheme further, adds up luminance block Y-component energy difference statistical value M according to mapping relations in described step (4)_EMethod comprise the following steps:

(1) according to luminance block Y-component dct transform coefficient in i-th 8x8 macro block, the ENERGY E of luminance block Y-component in i-th 8x8 macro block is calculated_i:

$E_i=\underset{b=0}{\overset{7}{\Σ}}\underset{b=0}{\overset{7}{\Σ}}|D(a,b){|}^2$

In formula, i=1,2 ..., N, a, b=0,1 ..., 7, D (a b) is 8x8 luminance block Y-component coefficient value at a row b row that dct transform obtains；

(2) ENERGY E of luminance block Y-component in i-th 8x8 macro block is compared_iWith the ENERGY E of luminance block Y-component in the N-i+1 8x8 macro block_N-i+1If, E_iMore than E_N-i+1, luminance block Y-component energy difference statistical value M in the most N number of macro block_EAdd 1, M_EInitial value is 0.

Another goal of the invention of the present invention is to provide the detection method after embedding above-mentioned a kind of MPEG-2 compressed domain video fragile watermark, and it comprises the following steps:

(1) utilize MPEG-2 decoder that the MPEG-2 video code flow containing watermark is resolved according to MPEG-2 video encoding standard, set I-frame video and have N number of macro block data, then add up luminance block Y-component nonzero coefficient number M ' after dct transform and quantization in N number of macro block_Y；

(2) chrominance block U component nonzero coefficient number M ' after dct transform and quantization in N number of macro block is added up_U；

(3) chrominance block V component nonzero coefficient number M ' after dct transform and quantization in N number of macro block is added up_V；

(4) luminance block Y-component energy difference statistical value M ' is added up according to mapping relations_E；

(5) by M '_Y、M′_U、M′_VWith M '_ECarry out 16 binarizations respectively to process, form 64 binary sequences, encrypt through Arnold conversion, obtain binary system encryption authenticating water-mark M '_K；

(6) encryption authenticating water-mark M ' is compared_KThe encrypted watermark M embedded with current I-frame video code stream afterbody_KIf, M '_KEqual to M_K, then current I-frame video content is distorted；Otherwise, current I-frame video content is distorted.

The present invention has the advantage that compared to existing technology

1, the present invention generates watermark according to video macro block content, and trickle attack is the most discernable, and real-time and video quality on video information the most do not affect；

2, the present invention is that on video quality without impact, computation complexity is low, and real-time is good based on compression domain to the detection that watermark embeds；It addition, detection watermark need not original video, belong to blind watermark method.

Detailed description of the invention

Elaborating embodiments of the invention below, the present embodiment is implemented under premised on technical solution of the present invention, gives detailed embodiment and concrete operating process, but protection scope of the present invention is not limited to following embodiment.

A kind of MPEG-2 compressed domain video fragile watermark embedding grammar, it comprises the following steps:

(1) utilizing MPEG-2 decoder to resolve MPEG-2 video code flow according to MPEG-2 video encoding standard, setting I-frame video has N number of macro block, the most first luminance block Y-component nonzero coefficient number Y after dct transform and quantization in statistics i-th 8x8 macro block_i, wherein, i=1,2 ..., N；Further according to Y_iCalculate luminance block Y-component nonzero coefficient number M after dct transform and quantization in N number of macro block_Y:

$M_Y=\underset{i=1}{\overset{N}{\Σ}}{Y}_i$

(2) chrominance block U component nonzero coefficient number M after dct transform and quantization in N number of macro block is added up_U；

First chrominance block U component nonzero coefficient number U after dct transform and quantization in statistics i-th 8x8 macro block_i, wherein, i=1,2 ..., N；Further according to U_iCalculate chrominance block U component nonzero coefficient number M after dct transform and quantization in N number of macro block_U:

$M_U=\underset{i=1}{\overset{N}{\Σ}}{U}_i$

(3) chrominance block V component nonzero coefficient number M after dct transform and quantization in N number of macro block is added up_V；

The most first chrominance block V component nonzero coefficient number V after dct transform and quantization in statistics i-th 8x8 macro block_i, wherein, i=1,2 ..., N；Further according to V_i, calculate chrominance block V component nonzero coefficient number M after dct transform and quantization in N number of macro block_V:

$M_V=\underset{i=1}{\overset{N}{\Σ}}{V}_i$

(4) luminance block Y-component energy difference statistical value M is added up according to mapping relations_E；

I.e. first according to luminance block Y-component dct transform coefficient in i-th 8x8 macro block, calculate the ENERGY E of luminance block Y-component in i-th 8x8 macro block_i:

$E_i=\underset{b=0}{\overset{7}{\Σ}}\underset{b=0}{\overset{7}{\Σ}}|D(a,b){|}^2$

In formula, i=1,2 ..., N, a, b=0,1 ..., 7, D (a b) is 8x8 luminance block Y-component coefficient value at a row b row that dct transform obtains；

Then the ENERGY E of luminance block Y-component in i-th 8x8 macro block is compared_iWith the ENERGY E of luminance block Y-component in the N-i+1 8x8 macro block_N-i+1If, E_iMore than E_N-i+1, luminance block Y-component energy difference statistical value M in the most N number of macro block_EAdd 1, M_EInitial value is 0；

(5) by M_Y、M_U、M_VAnd M_ECarry out 16 binarizations respectively to process, and form 64 binary sequences, according to 8 bit a line, generate 8x8 matrix W_8x8, obtain binary system encrypted watermark M through Arnold conversion_K, number of transitions is m, and one time Arnold conversion step is as follows:

X '=(x+y) mod 8

Y '=(x+2y) mod 8

In formula, x, y are matrix W_8x8Element subscript, x ', y ' they are the new subscript of the matrix after Arnold conversion, x, y=0,1 ..., 7, mod is complementation；

(6) by binary system encrypted watermark M_KIt is embedded into current I-frame video bit stream data afterbody, obtains the video flowing containing watermark.

To the detection method after MPEG-2 compressed domain video fragile watermark embeds, it comprises the following steps:

(1) to the MPEG-2 video code flow containing watermark, MPEG-2 decoder is utilized to resolve according to MPEG-2 video encoding standard, according to N number of macro block data of the I-frame video parsed, add up luminance block Y-component nonzero coefficient number M ' after dct transform and quantization in N number of macro block_Y；

(2) chrominance block U component nonzero coefficient number M ' after dct transform and quantization in N number of macro block is added up_U；

(3) chrominance block V component nonzero coefficient number M ' after dct transform and quantization in N number of macro block is added up_V；

(4) luminance block Y-component energy difference statistical value M ' is added up according to mapping relations_E；

(5) by M '_Y、M′_U、M′_VWith M '_ECarry out 16 binarizations respectively to process, form 64 binary sequences, encrypt through m Arnold conversion, obtain binary system encryption authenticating water-mark M '_K；

(6) encryption authenticating water-mark M ' is compared_KThe encrypted watermark M embedded with current I-frame video code stream afterbody_KIf, M '_KEqual to M_K, then current I-frame video content is distorted；Otherwise, current I-frame video content is distorted.

Video I frame each 8x8 macro block brightness of the present invention and chroma data generate watermark, and abundant information and independence, trickle attack is the most discernable；Watermark telescopiny does not revise video data, on video quality without impact；By video code stream analyzing data genaration watermark, embedding and detect and all carry out in compression domain, process is simple, and computation complexity is low, it is not necessary to parameter selects and arranges；Watermark detection process need not original video, belongs to blind watermatking.

The foregoing is only presently preferred embodiments of the present invention, not in order to limit the present invention, all any amendment, equivalent and improvement etc. made within the spirit and principles in the present invention, should be included within the scope of the present invention.

Claims (6)

1. a MPEG-2 compressed domain video fragile watermark embedding grammar, it is characterised in that: comprise the following steps:

(1) utilizing MPEG-2 decoder to resolve MPEG-2 video code flow according to MPEG-2 video encoding standard, setting I-frame video has N number of macro block, then adding up luminance block Y-component nonzero coefficient number after dct transform and quantization in N number of macro block is M_Y；

(2) chrominance block U component nonzero coefficient number M after dct transform and quantization in N number of macro block is added up_U；

(3) chrominance block V component nonzero coefficient number M after dct transform and quantization in N number of macro block is added up_V；

(4) luminance block Y-component energy difference statistical value M is added up according to mapping relations_E；

(5) by M_Y、M_U、M_VAnd M_ECarry out 16 binarizations respectively to process, and form 64 binary sequences, encrypt through Arnold conversion, obtain binary system encrypted watermark M_K；

(6) by binary system encrypted watermark M_KIt is embedded into current I-frame video bit stream data afterbody, obtains the video flowing containing watermark.

A kind of MPEG-2 compressed domain video fragile watermark embedding grammar the most according to claim 1, it is characterised in that: described step (1) is added up luminance block Y-component nonzero coefficient number M after dct transform and quantization in N number of macro block_YMethod comprise the following steps:

(1) luminance block Y-component nonzero coefficient number Y after dct transform and quantization in i-th 8x8 macro block is first added up_i, wherein, i=1,2 ..., N；

(2) further according to Y_iCalculate luminance block Y-component nonzero coefficient number M after dct transform and quantization in N number of macro block_Y:

A kind of MPEG-2 compressed domain video fragile watermark embedding grammar the most according to claim 1, it is characterised in that: described step (2) is added up chrominance block U component nonzero coefficient number M after dct transform and quantization in N number of macro block_UMethod comprise the following steps:

(1) chrominance block U component nonzero coefficient number U after dct transform and quantization in i-th 8x8 macro block is first added up_i, wherein, i=1,2 ..., N；

(2) further according to U_iCalculate chrominance block U component nonzero coefficient number M after dct transform and quantization in N number of macro block_U:

A kind of MPEG-2 compressed domain video fragile watermark embedding grammar the most according to claim 1, it is characterised in that: described step (3) is added up chrominance block V component nonzero coefficient number M after dct transform and quantization in N number of macro block_VMethod comprise the following steps:

(1) chrominance block V component nonzero coefficient number V after dct transform and quantization in i-th 8x8 macro block is first added up_i, wherein, i=1,2 ..., N；

(2) further according to V_i, calculate chrominance block V component nonzero coefficient number M after dct transform and quantization in N number of macro block_V:

A kind of MPEG-2 compressed domain video fragile watermark embedding grammar the most according to claim 1, it is characterised in that: described step (4) adds up luminance block Y-component energy difference statistical value M according to mapping relations_EMethod comprise the following steps:

(1) according to luminance block Y-component dct transform coefficient in i-th 8x8 macro block, the ENERGY E of luminance block Y-component in i-th 8x8 macro block is calculated_i:

In formula, i=1,2 ..., N, a, b=0,1 ..., 7, D (a b) is 8x8 luminance block Y-component coefficient value at a row b row that dct transform obtains；

(2) ENERGY E of luminance block Y-component in i-th 8x8 macro block is compared_iWith the ENERGY E of luminance block Y-component in the N-i+1 8x8 macro block_N-i+1If, E_iMore than E_N-i+1, luminance block Y-component energy difference statistical value M in the most N number of macro block_EAdd 1, M_EInitial value is 0.

Detection method after a kind of MPEG-2 compressed domain video fragile watermark the most as claimed in claim 1 embedding, it is characterised in that: comprise the following steps:

(1) utilize MPEG-2 decoder that the MPEG-2 video code flow containing watermark is resolved according to MPEG-2 video encoding standard, set I-frame video and have N number of macro block data, then add up luminance block Y-component nonzero coefficient number M ' after dct transform and quantization in N number of macro block_Y；

(2) chrominance block U component nonzero coefficient number M ' after dct transform and quantization in N number of macro block is added up_U；

(3) chrominance block V component nonzero coefficient number M ' after dct transform and quantization in N number of macro block is added up_V；

(4) luminance block Y-component energy difference statistical value M ' is added up according to mapping relations_E；

(5) by M '_Y、M′_U、M′_VWith M '_ECarry out 16 binarizations respectively to process, form 64 binary sequences, encrypt through Arnold conversion, obtain binary system encryption authenticating water-mark M '_K；

(6) encryption authenticating water-mark M ' is compared_KThe encrypted watermark M embedded with current I-frame video code stream afterbody_KIf, M '_KEqual to M_K, then current I-frame video content is distorted；Otherwise, current I-frame video content is distorted.