The present invention relates to embedding watermarks in Context-based Adaptive Binary Arithmetic Coding (CABAC) video streams.
Today, the demand for digital watermarking as an antipiracy technology is strong. To make it more difficult for pirates to circumvent watermarks it is important for many potential watermarks to be proposed and used. However, it is important for watermarks to not interfere with the intended viewing experience for the intended audience. As such, a need exists for more efficient watermarking techniques. As such, a goal of this invention is to generate a list of possible changes generally associated with watermarking that are CABAC/AVC compliant, and yet do not create visible artifacts, thereby ultimately providing an efficacious method for embedding watermarks in a CABAC video stream.
A method of providing CABAC compliant changes such as watermarks comprises accessing encoded data such as video data which comprise at least two blocks; creating or accessing a list of changes to the encoded data that include a direct change to a block; determining motion character or motion vector differentials of non-immediate blocks, non-immediate blocks being adjacent to an immediate block that are immediately adjacent to the block; determining change to the immediate block based on original motion character of the block and the non-immediate block and the motion character of the block that would result from the application of the change; storing the change to the list if the change does not cause a difference to the immediate block; and evaluating other potential changes if other potential changes are available, wherein the other potential changes are subjected to the same process steps as the direct change. Blocks can be any size or number of a collection elements or pixels, immediate blocks can be blocks that share some finite border with the block, and non-immediate blocks can be blocks that share some finite border with the immediate blocks, but not with the block. The method can include decoding the encoded data and coding the encoded data which can include using changes that have been stored in the list. The method can include determining a context for the encoded data for at least one of the blocks and/or determining a context index for the encoded data, wherein the context index (ctxIdx) is the sum of an initial value (ctxIdxOffset) and an increment (ctxIdxInc). The method can further include calculating or computing an original increment for the immediate block and a new increment for the immediate block associated with the direct change to the block; and using the original and new increments as a criterion for determining the difference in the storing step, wherein a motion vector differential of the non-immediate block and the original and new motion vector differentials for the block can be used as a criterion for determining the difference in the storing step. If the original increment for the immediate block and a new increment for the immediate block are different for a change, the change can be removed from the list in the storing step. The method can further include the step of calculating or computing an original increment for another immediate block and a new increment for the another immediate block associated with the direct change to the block, if the original and immediate increments are not different for the immediate block; and using the original and new increments for the another immediate block as an additional criterion for determining the difference in the storing step. A motion vector differential for the another non-immediate block that is adjacent to the another immediate block and the original and the new motion vector differentials for the block can be used as an additional criterion for determining the difference in the storing step. Additionally, the change can be added to the list in the storing step if the original and new increments for the another immediate block are not different.
The invention will now be described by way of example with reference to accompanying drawings.
Watermark embedding is herein performed by changing data bytes in a coded video stream such as CABAC (Context-based Adaptive Binary Arithmetic Coding) coded video stream. It is important to point out that CABAC coding in the context of H.264/AVC video encoders is mentioned throughout to emphasize that the disclosed invention will be generally applicable to such CABAC coded video streams. However, the scope of the invention is generally applicable to other coded data streams and other types of encoders.
With the understanding that watermarking can cause an AVC decoder failure, crash or the like, embodiments are presented that are directed to avoid negative events.
Of importance is the understanding that portions of encoded video bitstreams, such as those conforming to the H.264/AVC video coding standard, cannot be blindly modified while maintaining CABAC compliance. In a system that allows modifications under the constraint of maintaining CABAC compliance, changes may be intended to mark or watermark the video data. Such changes can be designed to identify modifications that can be made such that the modifications do not affect any CABAC contexts. However, a special case can arise in which a modification can influence the contexts used for the encoding/decoding of subsequent syntax elements. In particular, a modification might not change the current context of an encoder, but might nonetheless change the context that is selected at a decoder. When this occurs, the CABAC decoder will apply the wrong context, typically causing a decoding error. This application describes one or more embodiments that identify the special cases in which this error may occur and avoid modifications that would cause such errors.
The context of the invention is applicable to the operation of H.264/AVC CABAC coding algorithms, wherein an arithmetic coding system is used to improve entropy coding. CABAC achieves good compression performance through (a) selecting probability models for each syntax element according to the element's context; (b) adapting probabilities estimates based on local statistics; and (c) using arithmetic coding.
A more detailed view is provided in
In the regular coding mode, a context modeling step 206 is performed, wherein a probability model is selected such that the corresponding choice of a context depends on previously encoded syntax elements or bins. After the assignment of a context mode, the bin value 207 and its associated model 208 are passed to the regular coding engine 209, where the final stage of arithmetic coding and a subsequent model updating are processed.
In the bypass coding mode, to speed up the whole coding process, a simplified non-adaptive arithmetic coding engine or bypass coding engine 210 without the use of probability estimation and an update process is applied.
A final step of H.264/AVC compression can be entropy encoding, wherein at least one of the entropy encoding methods supported by that standard is CABAC. CABAC is a binary arithmetic coding technique in which the coding of a particular AVC syntax element is performed with respect to a specific probability model which is determined by a set of variables, collectively called a context. The CABAC encoder maintains many contexts and selects one for the encoding of each syntax element or even for the encoding of each bit at different bit positions of the same syntax element. In many cases, the encoding process results in a modification of the context variables. Each context maintains a set of context variables. The encoding process can change the variables inside that context.
A first step of H.264/AVC decompression is entropy decoding. When the stream is CABAC-encoded, the decoding is CABAC decoding. The CABAC decoder maintains many contexts and uses different contexts to decode different syntax elements or even bits at different bit positions of the same syntax element. Each context maintains a set of context variables. In many cases, the decoding process results in a modification of the context variables of the corresponding context. Different contexts are identified by a context index. It is important that the data be coded in such a way so that the decoder and encoder remain synchronized. The decoder needs to use an appropriate context for any particular element (i.e., which context was used to encode that element). The variables of that context, or the state of that context, need to be as predicted in the encoder.
It is possible and useful to modify certain CABAC-encoded syntax elements in a bitstream without modifying any of the context states. However, it is possible that a special case can occur in which the modifications result in the wrong context being used for decoding of a subsequent syntax element.
Now, the binarization of a motion vector differential (MVD) can consist of a prefix, and if the MVD is larger than 9, a suffix, wherein the binarized MVD can be denoted or defined as MVDBIN. The CABAC context used to decode the prefix of the MVDBIN of an inter-predicted block is determined by the context index “ctxIdx” as specified in the H.264/AVC standard. This index is calculated from the sum of two variables, ctxIdxOffset and ctxIdxInc, an initial index value and an increment, respectively, as shown in the following equation:
ctxIdx=ctxIdxOffset+ctxIdxInc
The ctxIdxOffset is specified in the standard. Specifically, ctxIdxOffset is 40 for the horizontal component of the MVDBIN and 47 for the vertical component of the MVDBIN. The ctxIdxInc is different for different bit positions of the MVDBIN prefix. For bits other than the first bit, the ctxIdxInc is determined by bit position. Specifically, ctxIdxInc is 3, 4, 5, 6, 6, 6 at bit positions 1, 2, 3, 4, 5, 6 (or greater), respectively. For the first bit (bit position 0) of the MVD prefix, the CABAC encoder and decoder determine the value of the increment, ctxIdxInc, by examining the absolute values of the MVDs (the entire MVD) of the neighboring blocks, specifically, the neighbor to the left (A) and the neighbor above (B), as shown in
To encode/decode the first bit of a prefix of the MVDBIN of an intra-predicted block, there are three possible contexts that can be used, as explained below. The first context is indexed by ctxIdxOffset, the second by ctxIdxOffset+1, and the third by ctxIdxOffset+2. Thus, the variable ctxIdxInc can take the values 0, 1, or 2 and this variable is used to select from the three available contexts. The first of these contexts is used for cases in which the current MVD is expected to be small, the second is used when the value of the current MVD is expected to be moderate, and the third is used when the current MVD is expected to be large. The expectation is based on the neighboring MVD values as follows:
If (absMvd_A+absMvd_B) is less than 3, ctxIdxInc is set equal to 0.
If (absMvd_A+absMvd_B) is in the range of 3 to 32, inclusive, ctxIdxInc is set equal to 1.
If (absMvd_A+absMvd_B) is greater than 32, ctxIdxInc is set equal to 2.
In short, the (absMvd_A+absMvd_B) can fall into three regions: [0, 3), [3, 32] and (32, +∞). (Notations are order theory notations: right-open interval, closed interval and open interval, respectively.)
One or more of the embodiments in the attached application can possibly result in modifications to MVDs of some blocks while taking care not to cause any context states to change. However, if the MVD of block A or that of block B change or if both change due to the modifications in such an embodiment, the new value of (absMvd_A+absMvd_B) could fall into a different region than the original one. In that case, the ctxIdxInc will be changed and the wrong context will be used for the decoding of the prefix of the MVDBIN of the current block. Thus, the CABAC decoder will behave differently in decoding the MVD of block Cur and decoding errors can result. A first solution to these decoding errors pertaining to potential modifications can be made that does not induce any changes in context states and is illustrated in
This evaluation can be extended by examining each potential modification to determine if it would induce a change in ctxIdxInc of any other block, thus possibly causing a decoding error. Potential modifications that do induce a change in ctxIdxInc of another block are, in at least one implementation, discarded.
For a potential modification in block A that will affect the MVD of block A, the MVDs of its two neighboring blocks, R (Right) and D (Down), might be affected as shown in
To check if the modification will cause a decoding error, this solution checks whether this modification of the MVD in block A will cause a change in the ctxIdxInc of the prefix of the MVD in block R or the prefix in block D. This leads to an algorithm described in
Note that the ctxIdxInc for block R is dependent, not only on the MVD of block A, but also on the MVD of block M, which is immediately adjacent to block R. The ctxIdxInc for block R is determined by the range in which the (absMvd_A+absMvd_M) falls. Similarly, the ctxIdxInc of block D depends on both the MVD of block A and the MVD of block N, an adjacent block to D, as it is determined by the range in which the (absMvd_A+absMvd_N) falls.
The algorithm of
There is a case in which the algorithm described above and shown in
A second solution and a third solution are illustrated in
Regarding the second solution, essentially, for each alternative MVD in block A, the effects on the ctxIdxInc of blocks D and R in
Specifically,
With further reference to
Additionally with reference to
If there are no additional block N alternative MVDs in decision step 670, the solution advances to step 685 in which the alternative MVD or MVDs in block A are saved, followed by advancing to decision step 645, in which additional block A alternative MVDs are accessed. If there are no more alternative block A MVDs, then the second solution is complete, which is represented by step 680. On the other hand, if there are additional alternative block A MVDs, the solution advances to step 650, in which the next alternative MVD in block A is accessed, followed by looping to step 610 and to the steps that followed step 610 as mentioned above.
This solution does discard any alternative MVD for block A that has a combination with any of the alternative MVDs for blocks M or N such that the combination will cause the ctxIdxInc of blocks D or R to change.
In some cases, removing the alternative MVD for block A may not be preferable or may need to be avoided. As such, a third solution is illustrated in
There are many approaches that can be taken to process the discard list. Here are a few:
In contrast to the first three solutions, a fourth solution can be considered “Block C centric” or “Block Cur centric” in that a potential change in block Cur is evaluated and block A or block B are checked to determine if either block A or B has changes that can impact the ctxIdxInc of block Cur. This forth solution is illustrated in
The first step 705 is to create a list of potential modifications that apply to the blocks in the current slice. It should be pointed out that embodiments of the inventions are intended to include the feature of accessing and/or modifying an existing list. In this regard, a list can be compiled, accessed or generated having all alternative MVDs for a slice.
Next, in step 710, for a given inter-predicted block, all possible MVDs are considered. All inter-predicted blocks will have at least one MVD; however, for some, there are potential changes that have been identified. Each potential change will result in a different MVD for the current block, block C, wherein the MVD is referred to as MVDc. For the current block, Block C, there are two neighbors of interest, block A to the left and block B above as shown in
An alternative implementation examines a previously created list and extracts the subset of those that apply to blocks in the current slice. The method then considers every inter-predicted block in the current slice, one at a time.
Several of the implementations and features described in this application can be used in the context of the H.264/MPEG-4 AVC (AVC) standard. However, these implementations and features can be used in the context of another standard (existing or future), or in a context that does not involve a standard. Thus, provided herein is one or more implementations having particular features and aspects. However, features and aspects of described implementations can also be adapted for other implementations.
The implementations described herein can be implemented in, for example, a method or process, an apparatus, a software program, a datastream, or a signal. Even if only discussed in the context of a single form of implementation such as only being mentioned as a method, the implementation or features discussed can also be implemented in other forms such as an apparatus or program. An apparatus can be implemented in, for example, appropriate hardware, software, and firmware. The methods can be implemented in, for example, an apparatus such as, for example, a computer or other processing device. Additionally, the methods can be implemented by instructions being performed by a processing device or other apparatus, and such instructions may be stored on a computer readable medium such as a CD, or other computer readable storage device, or an integrated circuit. Further, a computer readable medium can store the data values produced by an implementation.
As should be evident to one of skill in the art, implementations may also produce a signal formatted to carry information that can, for example, be stored or transmitted. The information may include, for example, instructions for performing a method, or data produced by one of the described implementations. For example, a signal can be formatted to carry a watermarked stream, an unwatermarked stream, or watermarking information.
Additionally, many implementations can be implemented in one or more of an encoder, a decoder, a post-processor processing output from a decoder, or a pre-processor providing input to an encoder. Further, other implementations are contemplated by this disclosure. For example, additional implementations can be created by combining, deleting, modifying, or supplementing various features of the disclosed implementations.
This application claims the benefit, under 35 U.S.C. §365 of International Application PCT/US2009/04713, filed Aug. 17, 2009, which was published in accordance with PCT Article 21(2) on Feb. 25, 2010 in English and which claims the benefit of United States provisional patent application No. 61/189,372, filed Aug. 19, 2008.
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/US2009/004713 | 8/17/2009 | WO | 00 | 2/11/2011 |
Publishing Document | Publishing Date | Country | Kind |
---|---|---|---|
WO2010/021699 | 2/25/2010 | WO | A |
Number | Name | Date | Kind |
---|---|---|---|
5867109 | Wiedeman | Feb 1999 | A |
6009176 | Gennaro et al. | Dec 1999 | A |
6064661 | Benn | May 2000 | A |
6341350 | Miyahara et al. | Jan 2002 | B1 |
6373960 | Conover et al. | Apr 2002 | B1 |
6894628 | Marpe et al. | May 2005 | B2 |
6900748 | Marpe et al. | May 2005 | B2 |
7113612 | Sugahara et al. | Sep 2006 | B2 |
7197164 | Levy | Mar 2007 | B2 |
7286710 | Marpe et al. | Oct 2007 | B2 |
7646881 | Zarrabizadeh | Jan 2010 | B2 |
7839312 | Tanaka et al. | Nov 2010 | B2 |
7865034 | So | Jan 2011 | B2 |
7974714 | Hoffberg | Jul 2011 | B2 |
8121341 | Tapson et al. | Feb 2012 | B2 |
20020136428 | Sugahara et al. | Sep 2002 | A1 |
20030070075 | Deguillaume et al. | Apr 2003 | A1 |
20040017852 | Garrido et al. | Jan 2004 | A1 |
20040168110 | Fuldseth et al. | Aug 2004 | A1 |
20040247154 | Bodo et al. | Dec 2004 | A1 |
20050044411 | Somin et al. | Feb 2005 | A1 |
20050069169 | Zarrabizadeh | Mar 2005 | A1 |
20050123207 | Marpe et al. | Jun 2005 | A1 |
20050207499 | Hwang et al. | Sep 2005 | A1 |
20060078292 | Huang et al. | Apr 2006 | A1 |
20060236130 | Ito et al. | Oct 2006 | A1 |
20060269096 | Kumar et al. | Nov 2006 | A1 |
20070053438 | Boyce et al. | Mar 2007 | A1 |
20070110033 | Tu et al. | May 2007 | A1 |
20070242862 | Watson et al. | Oct 2007 | A1 |
20080009272 | Toledano | Jan 2008 | A1 |
20080063071 | Suzuki | Mar 2008 | A1 |
20080165849 | Moriya et al. | Jul 2008 | A1 |
20090290750 | Tapson et al. | Nov 2009 | A1 |
20110129116 | Thorwirth | Jun 2011 | A1 |
20110176610 | He et al. | Jul 2011 | A1 |
20110222723 | He et al. | Sep 2011 | A1 |
20110293016 | Suzuki | Dec 2011 | A1 |
20120237078 | Watson et al. | Sep 2012 | A1 |
Number | Date | Country |
---|---|---|
101218830 | Jul 2008 | CN |
1515506 | Mar 2005 | EP |
1909508 | Apr 2008 | EP |
11341450 | Dec 1999 | JP |
2001119557 | Apr 2001 | JP |
2003134329 | May 2003 | JP |
2003179740 | Jun 2003 | JP |
2004221715 | Aug 2004 | JP |
2005533410 | Nov 2005 | JP |
2006279992 | Oct 2006 | JP |
2006303580 | Nov 2006 | JP |
2007053687 | Mar 2007 | JP |
2007525074 | Aug 2007 | JP |
WO2004066206 | Aug 2004 | WO |
WO2007067168 | Jun 2007 | WO |
WO2008065814 | Jun 2008 | WO |
WO2008118145 | Oct 2008 | WO |
WO2008154041 | Dec 2008 | WO |
Entry |
---|
International Search report dated Oct. 16, 2009. |
Nguyen et al., “A Fast Watermarking System for H.264/AVC Video,” 2006 IEEE, Dept. of Electronic Engineering, La Trobe University, Bundoora, Australia, pp. 81-84. |
Seo et al., “Low-Complexity Watermarking Based on Entropy Coding in H.264/AVC,” IEICE Trans. Fundamentals, vol.E91-A, No. 8, Aug. 2008. |
Noorkami, “Secure and Robust Compressed-Domain Video Watermarking for H.264,” A Thesis Presented to The Academic Faculty at Georgia Institute of Technology, 124 pages, Aug. 2007. |
Zou et al., “H.264/AVC Stream Replacement Technique for Video Watermarking”, ICASSP, 2008 IEEE, 2008. |
Mobasseri et al., “Authentication of H.264 Streams by Direct Watermrking of CACLC Blocks”, Security, Steganography, and Watermarking of Multimedia Contents IX, Proceedings of SPIE-IS&T Electronic Imaging, SPIE, vol. 6505, 65951W, 2007. |
Profrock et al, “H.264/AVC Video Authentication using Skipped Macroblocks for an Erasable Watermark”, Visual Communications and Image Processing, 2005 SPIE, Bellinigham, WA 2005. |
Hu, “Information Hiding Based on Intra Predictioin Modes for H.264 AVC”, Multimedia and Expo, 2007 IEEE, International Conference, IEEE PI, Jul. 1, 2007, pp. 1231-1234. |
Winkler, “Preceptual Quality Assessments for Video Watermarking”, Proceedings from the IEEE Conference on Standardizaton and Innovation in Information Technology, Oct. 18, 2002, pp. 90-94. |
Liu et al., “Data Hiding in Inter and Intra Prediction Modes of h.264/AVC”, IEEE Int'l. Symposium on Circuits and Systems, 2008 (ISCAS 2008), May 18, 2008, pp. 3025-3028. |
Song et al., “A Data Embedded Video Coding Scheme for Error-Prone Channels”, IEEE Transactions on Multimedia, vol. 3, No. 4, Dec. 1, 2001, pp. 415-423. |
Number | Date | Country | |
---|---|---|---|
20110135143 A1 | Jun 2011 | US |
Number | Date | Country | |
---|---|---|---|
61189372 | Aug 2008 | US |