CN103975605A - Watermark extraction based on tentative watermarks - Google Patents

Watermark extraction based on tentative watermarks Download PDF

Info

Publication number
CN103975605A
CN103975605A CN201280059704.5A CN201280059704A CN103975605A CN 103975605 A CN103975605 A CN 103975605A CN 201280059704 A CN201280059704 A CN 201280059704A CN 103975605 A CN103975605 A CN 103975605A
Authority
CN
China
Prior art keywords
watermark
distortion
predistortion
template
tentative
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201280059704.5A
Other languages
Chinese (zh)
Other versions
CN103975605B (en
Inventor
拉德·彼得罗维奇
文卡特拉曼·阿蒂
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Verance Corp
Original Assignee
Verance Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from US13/288,958 external-priority patent/US8533481B2/en
Priority claimed from US13/288,952 external-priority patent/US8682026B2/en
Priority claimed from US13/288,937 external-priority patent/US8615104B2/en
Priority claimed from US13/288,945 external-priority patent/US8923548B2/en
Application filed by Verance Corp filed Critical Verance Corp
Publication of CN103975605A publication Critical patent/CN103975605A/en
Application granted granted Critical
Publication of CN103975605B publication Critical patent/CN103975605B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/80—Generation or processing of content or additional data by content creator independently of the distribution process; Content per se
    • H04N21/83—Generation or processing of protective or descriptive data associated with content; Content structuring
    • H04N21/835—Generation of protective data, e.g. certificates
    • H04N21/8358—Generation of protective data, e.g. certificates involving watermark
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T1/00—General purpose image data processing
    • G06T1/0021—Image watermarking
    • G06T1/005—Robust watermarking, e.g. average attack or collusion attack resistant
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/018—Audio watermarking, i.e. embedding inaudible data in the audio signal
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
    • H04N1/32—Circuits or arrangements for control or supervision between transmitter and receiver or between image input and image output device, e.g. between a still-image camera and its memory or between a still-image camera and a printer device
    • H04N1/32101—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title
    • H04N1/32144—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title embedded in the image data, i.e. enclosed or integrated in the image, e.g. watermark, super-imposed logo or stamp
    • H04N1/32352—Controlling detectability or arrangements to facilitate detection or retrieval of the embedded information, e.g. using markers
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2201/00—General purpose image data processing
    • G06T2201/005—Image watermarking
    • G06T2201/0065—Extraction of an embedded watermark; Reliable detection

Abstract

Methods, devices and computer program products facilitate the extraction of embedded watermarks in the presence of content distortions. Pre-distorted synchronization templates are used to detect synchronization portions of embedded watermark frames. A pre-distorted synchronization template that best matches the synchronization portion of the embedded watermark frame produces an estimation of one or more distortions that are present in the content. The remainder of watermark frame can be evaluated based on the outcome of the comparison.

Description

Watermark extracting based on tentative watermark
related application
The application's case advocates from the 13/288th, No. 937, the 13/288th, and No. 945, the 13/288th, No. 952 and the 13/288th, the priority of No. 958 U.S. patent application case, described patent application case is all filed an application on November 3rd, 2011.The full content of mentioned patent application case is incorporated as a part for the disclosure of the application's case with way of reference before.
Technical field
In general the application's case relates to Content Management field.More particularly, the embodiment that discloses relates to embedding and the extraction from the watermark of media content.
Background technology
These chapters and sections intend to provide background or the context of the embodiment that discloses stating in claims.Explanation herein can comprise can carry out but may not be the concept of the concept that previously dreamed up or carried out.Therefore, unless separately there is indication herein, content described in these chapters and sections is not prior art to the explanation in the application's case and claims, and do not think and can form as prior art because comprising these chapters and sections.
Watermark be essentially be embedded in content host can not perceptual signal.Content host can be any one in the following: audio frequency, rest image, video maybe can be stored on physical medium or from a point, be transmitted or broadcast to arbitrary other content of another point.Watermark, and does not affect in fact the fidelity of content host or does not disturb the normal use of content host with carrying supplementary through design.For this reason, sometimes with watermark, carry out covert communications, wherein focus on and hide through hiding in the existence of signal.In addition, other extensive use of watermark comprises: prevent the undelegated use (for example, copy, play and propagate) of the content of multimedia of copyright, proof of ownership, checking, distort detection, content integrity check, broadcast monitors, transaction is followed the tracks of, audience measurement, triggers secondary activity (for example interactive with software program or nextport hardware component NextPort), transmits supplementary about content (for example title text, full name and artist name or about how buying the instruction of content) etc.Above list of application is not intended tool exhaustive, because many other current and systems in future can be benefited from the common channels transmit of main and supplementary.
(design watermaking system need to reach watermarked transparency (not sentience), watermarked robustness, the ability of having a mind to and being not intended to distorted signals is born in watermark) with the security requirement (that is, the watermarked degree of escaping detections, deletion and/or being handled by unauthorized side) of system between appropriate balance.Must carry out this balance play and limit average number and/or the maximum number that operates (that is, processing load) and memory use lower than the processing of a specific order of conventionally forcing for actual software and/or the hardware implementation scheme of water mark embedding device and/or watermark extracting device simultaneously.
Summary of the invention
These chapters and sections are intended to provide to the summary of particular exemplary embodiment and are not intended to limit the scope of the embodiment disclosing in the application's case.
An aspect of the embodiment that discloses relates to a kind of method, and it comprises: receive the watermarked frame that comprises sync section and packet portion, described watermarked frame has the content of watermark to obtain from embedding; Described received sync section and one or more predistortion synchronization template are compared; And the watermarked frame that receives described at least one outcome evaluation based on described comparison.
In an example embodiments, by following operation, produce described one or more predistortion synchronization template at least in part: produce a plurality of navigation content; Described a plurality of navigation content is embedded and have the watermark that comprises sync section; By one or more distortion, make the distortion of described a plurality of embedding navigation content; And obtain candidate's predistortion synchronization template from described a plurality of distortion navigation content.In a kind of exemplary version, produce described one or more synchronization template and be further included in by receiving described a plurality of embedding navigation content of a subgroup after the transmitting of transmitting media, wherein said transmitting media are incorporated into one or more distortion in described a plurality of embedding navigation content of described subgroup.In addition, producing described one or more synchronization template also comprises: each in the sync section of described a plurality of navigation content of described received subgroup and described candidate's predistortion synchronization template is compared, and be chosen in one or more candidate's predistortion synchronization template that the described sync section of the interior described a plurality of navigation content with described received subgroup of predefine tolerance limit mates.
In an example embodiments, described each the error count relatively producing in described one or more predistortion synchronization template, described error count is indicated the number of the not matching symbols between described the received sync section of described watermarked frame and each in described one or more predistortion synchronization template.In another example embodiments, the correlation of each in described one or more predistortion synchronization template of described relatively generation, described correlation indicates the described sync section of described watermarked frame to what extent to mate each in described one or more predistortion synchronization template.
According to another embodiment, method referred to above further comprises: described relatively after, the predistortion synchronization template of described the received sync section of watermarked frame described in identification optimum Match, and selection represents to be present in the described distortion in described content with described one or more type of distortion and the amount distortion being associated through identification predistortion synchronization template.In one embodiment, method referred to above further comprises for the distortion of each type and/or the distortion order that combines to presort and organizes described candidate's predistortion synchronization template.In this embodiment, the order of presorting described in make to have represent real content distortion compared with first candidate's predistortion synchronization template rank of high likelihood higher than second candidate's predistortion synchronization template with the less possibility that represents real content distortion.
In another embodiment, described one or more predistortion synchronization template is corresponding to the specific search granularity in distortion space.In a kind of version of above embodiment, select the number of described one or more predistortion synchronization template to reduce the one or both in the following: false watermark detection probability and/or assess the described required computational resource of watermarked frame that received.
According to another embodiment, described assessment produces the indication to the existence of the one in the following: tentative watermark, and its expression has the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting; Or concluding watermark, it represents to have after testing the watermark that is less than or equal to the false watermark detection probability of being associated of wanted false watermark detection probability.In another embodiment, described assessment identification is present in one or more type of distortion and the amount distortion in described content.
In another embodiment, described content comprises at least first and second watermark message, and described the first watermark message has the pay(useful) load that is less than described the second watermark message.In this embodiment, assess described received watermarked frame and cause described the first watermark message to detect as tentative watermark, and use the content distortion estimation obtaining for described tentative watermark to extract described the second watermark message.
The embodiment that discloses relates to a kind of device on the other hand, and it comprises receiver, and described receiver is configured to receive the watermarked frame that comprises sync section and packet portion, and described watermarked frame has had the content of watermark to obtain from embedding.Described device also comprises: comparator, and it is configured to described received sync section and one or more predistortion synchronization template to compare; And evaluator, it is configured to the watermarked frame that receives described at least one outcome evaluation based on described comparator.
Another embodiment relates to a kind of device, and it comprises: water mark embedding device, and it is configured so that a plurality of navigation content embeds the watermark that comprises sync section; Distortion processing components, it is configured to make the distortion of described a plurality of embedding navigation content by one or more distortion; And watermark extracting device, it is configured to produce candidate's predistortion synchronization template from described a plurality of distortion navigation content.In one embodiment, this device further comprises: receiver, it is configured to receive described a plurality of embedding navigation content of a subgroup after by the transmitting of transmitting media, and wherein said transmitting media are incorporated into one or more distortion in described a plurality of embedding navigation content of described subgroup; And template matches assembly, it is configured to each in the sync section of described a plurality of navigation content of described received subgroup and described candidate's predistortion synchronization template to compare, and is chosen in predefine tolerance limit one or more candidate's predistortion synchronization template mating with the described sync section of described a plurality of navigation content of described received subgroup.
The embodiment that discloses relates to a kind of device on the other hand, and it comprises processor and memory, and described memory comprises processor executable code.Described processor executable code configures described device to receive the watermarked frame that comprises sync section and packet portion when being carried out by described processor, and described watermarked frame has the content of watermark to obtain from embedding; Described received sync section and one or more predistortion synchronization template are compared; And the watermarked frame that receives described at least one outcome evaluation based on described comparison.
The embodiment that discloses relates to a kind of computer program being embodied on nonvolatile computer-readable media on the other hand, it comprises for receiving the program code of the watermarked frame that comprises sync section and packet portion, and described watermarked frame has had the content of watermark to obtain from embedding.Described computer program also comprises program code for described received sync section and one or more predistortion synchronization template are compared and for the program code of the watermarked frame that receives described at least one outcome evaluation based on described comparison.
The embodiment that discloses relates to a kind of method on the other hand, and it comprises from embedding content host and extracts a plurality of tentative watermarks, and wherein each tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.The method also comprises: use the described estimated distortion information that at least both acquisitions are associated with one or more distortion being present in described embedding content host in extracting tentative watermark; And based on described estimated distortion information, obtain one or more predistortion watermark template.Described method further comprises uses described one or more predistortion watermark template described at least one in extracting tentative watermark of reappraising.
In one embodiment, described in, reappraise to cause detecting with the false watermark detection probability that is associated with described each in extracting tentative watermark and compare and have through improving the watermark of false watermark detection probability.In another embodiment, described in, reappraise to cause detecting and there is the concluding watermark that is less than or equal to the false watermark detection probability of being associated of described wanted false watermark detection probability.In an embodiment again, each expression in described predistortion watermark template can be polluted particular type and the specified quantitative of distortion of at least one type of described embedding content host.According to another embodiment, described in reappraise cause to be present in distortion in described embedding content host through improved estimator.
In one embodiment, before method referred to above is further included in and obtains one or more predistortion watermark template, determine that whether described estimated distortion information is corresponding to the amount distortion that surpasses certain distortion threshold value.In this embodiment, only in the situation that described estimated distortion information obtains described one or more predistortion watermark template corresponding to the amount distortion that is no more than described certain distortion threshold value.
In another embodiment, described in use at least two through extracting two or more sync sections of tentative watermark, obtain described estimated distortion information.In this embodiment, by described described spacing at least between the two in extracting tentative watermark is compared and obtains described estimated distortion information with the nominal clearance that is associated with the watermark of not distortion content.In another embodiment, the described extraction of each in described tentative watermark comprises the sync section of extraction of the packet portion of the described tentative watermark of triggering that detects described tentative watermark.
The embodiment that discloses relates to a kind of device on the other hand, it comprises extractor, described extractor is configured to extract a plurality of tentative watermarks from embedding content host, and wherein each tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.This device also comprises distortion estimator, described distortion estimator be configured to use described in extracting tentative watermark at least both obtain the estimated distortion information being associated with one or more distortion being present in described embedding content host and obtain one or more predistortion watermark template.This device further comprises evaluator, and described evaluator is configured to use described one or more predistortion watermark template described at least one in extracting tentative watermark of reappraising.
The embodiment that discloses relates to a kind of device on the other hand, and it comprises processor and memory, and described memory comprises processor executable code.The configuration when being carried out by described processor of described processor executable code is stated device to extract a plurality of tentative watermarks from embedding content host, and wherein each tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.Described processor executable code also configures described device to use the described estimated distortion information that at least both acquisitions are associated with one or more distortion being present in described embedding content host in extracting tentative watermark when being carried out by described processor; And based on described estimated distortion information, obtain one or more predistortion watermark template.Described processor executable code when being carried out by described processor further the described device of configuration to use described one or more predistortion watermark template described at least one in extracting tentative watermark of reappraising.
The embodiment that discloses relates to a kind of computer program being embodied on nonvolatile computer-readable media on the other hand, it comprises for extract the program code of a plurality of tentative watermarks from embedding content host, and wherein each tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.Described computer program also comprises for using described through extracting the program code of the estimated distortion information that at least both acquisitions are associated with one or more distortion being present in described embedding content host of tentative watermark, and for obtain the program code of one or more predistortion watermark template based on described estimated distortion information.Described computer program further comprises for using described one or more predistortion watermark template to reappraise described through extracting at least one program code of tentative watermark.
The embodiment that discloses relates to a kind of method on the other hand, and it comprises from embedding content host and extracts tentative watermark, and wherein said tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.The method also comprises: use describedly through extracting tentative watermark, to obtain the estimated distortion information that indication is present in the coarse value of certain distortion in described embedding content host or distortion combination; And selection is corresponding near one or more supplementary predistortion watermark template of one or more distortion value described coarse value.The method further comprises uses described selected one or more supplementary predistortion watermark template to reappraise described through extracting tentative watermark.
In one embodiment, obtaining described estimated distortion information comprises: a plurality of rough predistortion watermark template with coarse-grained selection corresponding to the scope of the distortion value of certain distortion or distortion combination; Described each in extracting tentative watermark and described a plurality of rough predistortion watermark template is compared; And through extracting the first rough predistortion watermark template of tentative watermark, obtain described estimated distortion information described in identification optimum Match.In a kind of version of this embodiment, from predistortion watermark template, compared with random in the middle of big collection, select described a plurality of rough predistortion watermark templates.In another version of this embodiment, select corresponding to the described a plurality of rough predistortion watermark template that may be present in one or more type of the distortion in described embedding content host.
According to another embodiment, select described one or more supplementary predistortion watermark template to realize near the distortion space described coarse value with fine granulation search.In an embodiment again, described in reappraise cause to be present in distortion in described embedding content host through improved estimator.In a kind of version, utilize and describedly through improved estimator, from described embedding content host, extract extra watermark.In one embodiment, described in, reappraise to cause detecting and describedly through extracting the false watermark detection probability of tentative watermark, compare and have through improving the watermark of false watermark detection probability with being associated with.In another embodiment, described in, reappraise to cause detecting and there is the concluding watermark that is less than or equal to the false watermark detection probability of being associated of wanted false watermark detection probability.
In another embodiment, described in, reappraise and comprise: by described one or more supplementary predistortion watermark template and describedly compare through extracting tentative watermark; Produce the error count be relatively associated with each, wherein said error count represents that the supplementary predistortion watermark template of described correspondence to what extent mates described through extracting tentative watermark; And selection is corresponding to the described supplementary predistortion watermark template of the minimum error count that produces.In another embodiment, with repetitive mode, select one at a time described one or more supplementary predistortion watermark template.
The embodiment that discloses relates to a kind of device on the other hand, it comprises extractor, described extractor is configured to extract tentative watermark from embedding content host, and wherein said tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.This device further comprises distortion estimator, described distortion estimator is configured to use and describedly through extracting tentative watermark, obtains the estimated distortion information that indication is present in the coarse value of certain distortion in described embedding content host or distortion combination, and selects one or more the supplementary predistortion watermark template corresponding near one or more distortion value described coarse value.This device comprises that evaluator, described evaluator are configured to use described selected one or more supplementary predistortion watermark template to reappraise described through extracting tentative watermark in addition.
According to disclosed embodiment on the other hand, provide a kind of device, it comprises processor and memory, and described memory comprises processor executable code.Described processor executable code configures described device to extract tentative watermark from embedding content host when being carried out by described processor, and wherein said tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.Described processor executable code also configures described device and describedly through extracting tentative watermark, obtains the estimated distortion information that indication is present in the coarse value of certain distortion in described embedding content host or distortion combination to use when being carried out by described processor, and selects one or more the supplementary predistortion watermark template corresponding near one or more distortion value described coarse value.Described processor executable code further configures described device to use described selected one or more supplementary predistortion watermark template to reappraise described through extracting tentative watermark when being carried out by described processor.
The embodiment that discloses relates to a kind of computer program being embodied on nonvolatile computer-readable media on the other hand, it comprises for extracting the program code of tentative watermark from embedding content host, and wherein said tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.Described computer program also comprises for using the described program code that is present in the estimated distortion information of the certain distortion of described embedding content host or the coarse value of distortion combination through extracting tentative watermark acquisition indication, and for selecting the program code corresponding near one or more supplementary predistortion watermark template of one or more distortion value described coarse value.Described computer program further comprises for using described selected one or more supplementary predistortion watermark template to reappraise described through extracting the program code of tentative watermark.
The embodiment that discloses relates to a kind of method on the other hand, and it comprises from embedding content host and extracts tentative watermark, and wherein said tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.Described method also comprises and by acquisition, is positioned described embedding content host inherence and through extracting the symbol of potential watermarked frame of the predefine position of tentative watermark, forms one or more through extrapolation watermark with respect to described, and determines through extrapolation watermark, whether to meet the false watermark detection probability of wanting with described after testing described in when tentative watermark is evaluated jointly.In one embodiment, the described symbol location of described potential watermarked frame is located through extracting time predefined and/or the locus of tentative watermark with respect to described in described embedding content host inherence.
In another embodiment, determine and describedly through extrapolation watermark, whether meet the false watermark detection probability of wanting and comprise: the number of determining the mismark in described one or many person in the watermark of extrapolating, based on the described of mismark, through ascertaining the number, to described described one or many person in extrapolation watermark, assign weight at least in part, and through weighting, through extrapolation watermark, whether meet described wanted false watermark detection probability described in determining when with described tentative watermark combination after testing.In this embodiment, can determine the often described number of the mismark in extrapolation watermark by one or many person in the following: described through extrapolation watermark symbol and described through extracting the comparison of the symbol of tentative watermark, carry out the described error-correcting code decoding through extrapolation watermark symbol, and/or the described comparison through extrapolation watermark and one or more predistortion watermark template.
In one embodiment, forming described each in extrapolation watermark comprises: the number that obtains the mismark in one or more segmentation, the described number of the mismark based in each in described one or more segmentation is assigned weight to each in described one or more segmentation at least in part, combination is associated with described one or more segmentation described through assign weight with produce one or more through weighting through the watermark section of extrapolating, and through weighting, through extrapolation watermark section, whether meet described wanted false watermark detection probability described in determining when with described tentative watermark combination after testing.In this embodiment, the described number that obtains the mismark in each in described one or more segmentation comprises: described symbol and one or more predistortion watermark template of the described potential watermarked frame in each in described one or more segmentation are compared; And producing each the counting in described one or more segmentation, described counting represents the described symbol of described potential watermarked frame and the number of the not matching symbols between described one or more predistortion watermark template in described one or more segmentation.In addition, in a kind of version of this embodiment, the one or many person based in the following determines the number of the described segmentation in watermark through extrapolation: described through extract tentative watermark scope, be present in the amount distortion in described embedding content host and/or be present in the type of distortion in described embedding content host.In addition, in a kind of version of this embodiment, the expection weight that does not add watermark segmentation based on equal particular fragments in length is assigned the described weight of described particular fragments at least in part.According to another embodiment, based on one or many person in the following, determine at least in part the scope of each segmentation: amount distortion and expection that expection is present in described embedding content host are present in the type of distortion in described embedding content host.In another embodiment, for described one or more segmentation that jointly strides across the scope less than described tentative watermark, carry out described in when with described tentative watermark combination after testing through extrapolation watermark section, whether to meet the described definite of the false watermark detection probability of described institute through weighting.
According to another embodiment, described wanted false watermark detection probability is corresponding to the detection of concluding watermark.In another embodiment, method referred to above further comprises report to being present in distortion or the amount of distortion combination and the estimation of type in described embedding content host.
The embodiment that discloses relates to a kind of device on the other hand, it comprises extractor, described extractor is configured to extract tentative watermark from embedding content host, and wherein said tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.Described device also comprises: watermark extrapolator, and it is configured to be positioned described embedding content host inherence by acquisition and through extracting the symbol of potential watermarked frame of the predefine position of tentative watermark, forms one or more through extrapolation watermark with respect to described; And evaluator, it is configured to determine through extrapolation watermark, whether to meet the false watermark detection probability of wanting with described after testing described in when tentative watermark is evaluated jointly.
The embodiment that discloses relates to a kind of device on the other hand, and it comprises processor and memory, and described memory comprises processor executable code.Described processor executable code configures described device to extract tentative watermark from embedding content host when being carried out by described processor, and wherein said tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.Described processor executable code also configures described device to be positioned in described embedding content host forming one or more through extrapolation watermark with respect to described through extracting the symbol of potential watermarked frame of the predefine position of tentative watermark by acquisitions when being carried out by described processor, and determine with described after testing when tentative watermark is evaluated jointly described in through the whether satisfied false watermark detection probability of wanting of extrapolation watermark.
The embodiment that discloses relates to a kind of computer program being embodied on nonvolatile computer-readable media on the other hand, it comprises for extracting the program code of tentative watermark from embedding content host, and wherein said tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting.Described computer program also comprises for be positioned described embedding content host inherence by acquisition and through extracting the symbol of potential watermarked frame of the predefine position of tentative watermark, forms one or more through the program code of extrapolation watermark with respect to described, and through extrapolation watermark whether is meeting the program code of want false watermark detection probability with described after testing described in when tentative watermark is evaluated jointly for determining.
The embodiment that discloses relates to a kind of method on the other hand, and it comprises: from embedding the potential watermarked frame of contents extraction that has one or more watermark, wherein each watermarked a plurality of symbol that form watermarked frame that comprise; Described watermarked frame is divided into a plurality of segmentations, and each in wherein said a plurality of segmentations comprises both in more watermark symbol; To each segmentation, assign weight; Combination is associated with two or more segmentations described through appointment weight to produce one or more through weighting watermarked frame partially or completely; And determine described one or more through weighting wholly or in part watermarked frame whether meet the false watermark detection probability of wanting.
In one embodiment, according to the scope that is present in wrong each segmentation of type selecting in described content.In another embodiment, the expection weight that does not add watermark segmentation based on equal particular fragments in length is assigned the described weight of described particular fragments at least in part.
The embodiment that discloses relates to a kind of device on the other hand, and it comprises extractor, and described extractor is configured to the potential watermarked frame of contents extraction that has one or more watermark from embedding, wherein each watermarked a plurality of symbol that form watermarked frame that comprise.Described device also comprises evaluator, and described evaluator is configured to described watermarked frame to be divided into a plurality of segmentations, and each in wherein said a plurality of segmentations comprises both in more watermark symbol; To each segmentation, assign weight; Combination is associated with two or more segmentations described through appointment weight to produce one or more through weighting watermarked frame partially or completely; And determine described one or more through weighting wholly or in part watermarked frame whether meet the false watermark detection probability of wanting.
The embodiment that discloses relates to a kind of device on the other hand, and it comprises processor and memory, and described memory comprises processor executable code.Described processor executable code configures described device to have the potential watermarked frame of contents extraction of one or more watermark, wherein each watermarked a plurality of symbol that form watermarked frame that comprise from embedding when being carried out by described processor; Described watermarked frame is divided into a plurality of segmentations, and each in wherein said a plurality of segmentations comprises both in more watermark symbol; To each segmentation, assign weight; Combination is associated with two or more segmentations described through appointment weight to produce one or more through weighting watermarked frame partially or completely; And determine described one or more through weighting wholly or in part watermarked frame whether meet the false watermark detection probability of wanting.
The embodiment that discloses relates to a kind of computer program being embodied on nonvolatile computer-readable media on the other hand, it comprises: for have the program code of the potential watermarked frame of contents extraction of one or more watermark, wherein each watermarked a plurality of symbol that form watermarked frame that comprise from embedding; For described watermarked frame being divided into the program code of a plurality of segmentations, each in wherein said a plurality of segmentations comprises both in more watermark symbol; For assign the program code of weight to each segmentation; For combining be associated with two or more segmentations described through assigning weight to produce one or more through the weighting program code of watermarked frame partially or completely; And for determine described one or more through weighting wholly or in part watermarked frame whether meet the program code of the false watermark detection probability of wanting.
Accompanying drawing explanation
Fig. 1 is the block diagram that can adapt to the watermaking system of disclosed embodiment.
Fig. 2 is the block diagram that can adapt to the watermark extracting apparatus of disclosed embodiment.
Fig. 3 A shows the exemplary watermarked frame with synchronous (SYNCH) and PACKET part.
Fig. 3 B is illustrated in success and the compilation of the exemplary PACKET after unsuccessful compilation of the SYNCH part of watermarked frame and attempts.
Fig. 3 C graphic extension is according to the PACKET compilation operation of the use predistortion SYNCH template of one exemplary embodiment.
Fig. 4 shows according to the predistortion SYNCH template of one exemplary embodiment.
One group of operation that Fig. 5 graphic extension is carried out according to one exemplary embodiment.
Fig. 6 is according to the block diagram of the specific components of the some or all of operations in the operation that can be used for execution graph 5 of one exemplary embodiment.
Fig. 7 graphic extension can be through carrying out to produce one group of operation of one group of predistortion watermark template according to one exemplary embodiment.
Fig. 8 is the block diagram that can be used for the specific components of the some or all of operations in the operation of execution graph 7.
Fig. 9 graphic extension can be through carrying out to promote one group of operation of watermarked extraction according to one exemplary embodiment.
Figure 10 graphic extension can be through carrying out with from one group of watermarked operation of distortion contents extraction according to one exemplary embodiment.
Figure 11 A shows the exemplary watermarked frame with synchronous (SYNCH) and PACKET part.
Figure 11 B is illustrated in success and the compilation of the exemplary PACKET after unsuccessful compilation of the SYNCH part of watermarked frame and attempts.
Figure 11 C graphic extension is according to the PACKET compilation operation after detecting tentative watermark of one exemplary embodiment.
Figure 11 D graphic extension according to one exemplary embodiment after detecting tentative watermark and use the PACKET compilation of watermark segmentation to operate.
Figure 12 graphic extension is the exemplary plot to watermark weight threshold according to the false watermark detection probability that separate watermark search and sparse granular search and fine granulation search are compared of one exemplary embodiment.
Figure 13 graphic extension can be through carrying out one group of operation 1300 with the watermark of extrapolating on by segmentation basis according to one exemplary embodiment.
Figure 14 is the block diagram of specific components that can be used for carrying out the some or all of operations in the operation of Figure 13 and/or Figure 18 according to one exemplary embodiment.
Figure 15 shows according to the curve chart of the expection weight that does not add watermark segmentation of one exemplary embodiment.
Figure 16 graphic extension can be implemented the reduced graph of the device of various disclosed embodiment within it.
Figure 17 graphic extension is according to SYNCH compilation and the predistortion SYNCH template operation of one exemplary embodiment.
Figure 18 graphic extension can be through carrying out one group of operation with extrapolation watermark according to one exemplary embodiment.
Embodiment
In the following description, the unrestriced object for explanation, statement details and explanation are to provide the thorough understanding to disclosed embodiment.Yet, it will be apparent to those skilled in the art that, can in deviating from other embodiment of these details and explanation, put into practice the present invention.
In addition, in the present invention's explanation, use word " exemplary " to mean to serve as example, example or graphic extension.The arbitrary embodiment or the design that are described as in this article " exemplary " may not be considered as than other embodiment or design preferred or favourable.But, intend in concrete mode, to present concept with word is exemplary.
Can use various watermark embedded technologies that watermark is embedded in content host by following operation: for instance, in time domain or frequency domain, handle the least significant bit of host signal, use spread-spectrum, phase place, amplitude or frequency modulating technology to insert watermark by Independent Carrier Wave signal, and use the interdependent carrier signal of main frame (for example feature modulation and the embedded technology of knowing the inside story) to insert watermark.Most embedded technologies utilize the psycho-visual of content host or psychologic acoustics (or both) to analyze optimal location and the amplitude of determining insertion watermark.This analyzes identification host signal conventionally can hide or shelter the watermarked degree as people's perception.
Conventionally use various storages and/or the storage of transmitting channel to embed content host and/or embedding content host is transmitted into another location.These channels are characterized by intrinsic noise and distortion, for example, owing to polluting the cut of the data on optical media and the mistake that fingerprint causes, noise in the radio broadcasting of audio frequency-vision content, abandons etc. via internet or from the bag in the flowing of the content of multimedia of media server.The extra infringement that can affect watermarked fidelity may for example, cause due to the various signal processing operations (lossy compression method, convergent-divergent, rotation, mould/number conversion etc.) of conventionally content of multimedia being carried out.In most digital watermarkings application, watermarkedly must under this noise like and distortion condition, maintain its integrality.
Watermarked fail safe be watermaking system on the other hand.For example, in application-specific (proof of ownership, source-verify, tracing pirate, to there being the access control of the content of copyright), watermarked must resist be intended to detect watermark existence, decipher by data, the modification of those watermark carryings or insert illegal value (forgery) and/or remove watermarked having a mind to and handle.
Another consideration in design watermaking system is watermark payload capacity.The application-specific of watermaking system is depended in this requirement.Typical case's application was arrived in the scope of the supplementary that needs tens per second between exist (that is, the needing single state watermark) of detecting only watermark from needs.In rear a kind of situation, can come carrying identification and time sequence information with embedding position, for example sequence number and timestamp, metadata (metadata), such as title, artist name, purchase information etc.
Another factor assessing the cost for embedding and/or extraction unit in design actual watermark system.This factor is to becoming more and more important by consumer electronics device or the software utilization of limited silicon area or calculation requirement enforcement.This factor can be closely related with the set purposes of watermaking system.For instance, the watermark of following the tracks of for the court of pirate channel (for example different code being embedded in to those watermarks of each copy of content of scattering via internet) needs simple embedded device and tolerable is complicated and expensive court's extractor.On the other hand, for instance, through design with prevent from offseting the content of multimedia in expense type electronic installation undelegated access the exquisite embedded device of copy control system tolerable but need simple and efficient extractor.
Another key factor in design watermaking system is false watermark detection probability.False watermark can produce when from unlabelled content detection to watermark, or may cause owing to the watermark value of the watermark value that is different from initial embedding being detected.The false watermark detection level of wanting also can be depending on the set application of watermaking system and changes.For instance, at copy, control in application, false detection probability must be very low by (for example, approximately 10 -12), because for example, certainly will make user dejected and device producer and/or content provider had to negative influence owing to the false watermark detection of the content of legal purchase being carried out to restriction (, stopping the playback of content).On the other hand, for (for instance), follow the tracks of the number of times that following song broadcasted and monitor application, the relatively high false verification and measurement ratio of tolerable (for example, approximately 10 to produce the broadcast of royalty or pop roll -6), because several false existence detecting can have seldom impact to the end product of counting.
The selection that another key factor that affects the overall performance of watermaking system is the particular technology for watermark embedding and extraction.In view of application-specific on hand, between requirement referred to above, making optimal compromise is the task that very tool is challenged.
The embodiment that discloses significantly improves embedding and the extraction of watermark, and the various system requirements of balance, comprise watermark transparency, robustness, false positive detection, fail safe, payload capacity and process load simultaneously.In certain embodiments, watermark extracting based on tentative watermark detection as the intermediate steps towards concluding watermark extracting.
Can promote with following conversion the understanding of the concept relevant to disclosed embodiment.However, it should be understood that can be with other label and represent to adapt to the watermaking system of the embodiment of the application's case.Watermaking system can be by tuple { O, W, K, E k, D k, C tdescribe, wherein O is one group of whole origin host content (that is, not having one or more watermarked content host), and W is one group of whole watermark, and K is one group of whole stego-key, E kfor embedding computing, D kfor extracting computing and C tfor comparison operation.Embed computing E kand extraction computing D kcan by equation (1) and (2), be represented respectively.
E K：O×W×K→O w(1)。
D K：O w×K→W′ (2)。
In equation (1) and (2), O wembed content and W ' be from described group of embedding content O for described group wdescribed group of candidate's watermark of extracting.The comparison operation C being represented by equation (3) tto compare with described group of whole watermark W through extracting watermark W '.
C T：<W′，W>→{0，1} (3)。
It should be noted that in certain embodiments, can be not for whole watermarks but carry out described comparison for a subgroup of whole watermarks.For instance, this subgroup can comprise there is the watermark of specific range of values, for embedding the watermark etc. of the content of particular type.After obtaining successfully mating between W ' and W, described comparison operation produces the first value (for example, " 1 ") at once, and after unsuccessful coupling, comparison operation produces the second value (for example, " 0 ") at once.
Fig. 1 shows according to the simplified block diagram of the watermaking system 100 of one exemplary embodiment.Origin host content 102, c oand watermark 110W is the input to water mark embedding device 104.Watermark 110 embeds stego-key 106 by using, and the watermark design module 108 of K and origin host content 102 produces.The watermark that produces 110 is embedded in content host 102 to produce embedding content host 112, c w.By embedding computing E kthe generation of embedding content host 112 can be represented by following equation (4).
c w＝E k(c o，W) (4)。
Embed content host 112 and use one or more 114 storages of transmitting/medium and/or transmitting.Transmitting/medium 114 can be unidirectional or two-way transmitting media.In utilizing the embodiment of two-way transmitting channel, from the information of watermark extracting device, can be back to the embedded components of watermaking system 100.This feedback can be including but not limited to the transmitting of identification, checking and/or authorization message, the success operating about watermark extracting or failed indication, the report of the various statistics relevant to extractor operation etc.
Back with reference to figure 1, the content by 114 transmittings of transmitting/medium can comprise because intrinsic transmitting/medium 114 causes and/or for to embedding the distortion of having a mind to the result of attacking of content host 112.This embeds content host 116, c ' through revising wfor the input to watermark extracting device 118, watermark extracting device 118 uses and extracts stego-key 120, and K and (possibly) origin host content 102 produce one or more candidate's watermark 122, W '.By extracting computing D kthe generation of candidate's watermark 122 of carrying out can be represented by following equation (5).
W′＝D K(c′ w，c o) (5)。
Note, it is watermarked that many watermark detectors do not need origin host content 102 to extract, and therefore, in those situations (being commonly referred to blind Detecting), in equation (5), extract computing D kto the dependence of origin host content 102, can decline.Be further noted that in certain embodiments, embed stego-key 106 identical with extraction stego-key 120.Yet, in other embodiments, embed stego-key 106 and be different from extraction stego-key 120.
Candidate's watermark 122 is in general due to possibility content manipulation or because imperfect embedding program step is different from the watermark 110 of selecting for embedding.By 124 pairs of candidate's watermarks of comparator 122 and one or more possibility watermark 126, W icompare.Described comparator produce indicate whether to produce candidate's watermark 122 with may watermark 126 in one between the watermark detector mating export 128.For instance, can produce watermark detector output 128 according to equation (3).Watermark detector output 128 indications that can further comprise about value, state and/or the pattern of watermark after testing.The comparator 124 that it should be noted that Fig. 1 only graphic extension for the evaluate candidate watermark W ' upper exemplary mechanism that represents the true watermark of content whether really.For evaluating the existence of watermark and/or other exemplary technique of value, including but not limited to mistake in using, detect and/or error-correcting code.For example, can be by the candidate's watermark W ' decoding through error-correcting code (ECC) coding to determine whether to recover the wrong number of faultless watermark bag and/or acquisition pollution specific watermark bag.
In view of wanted false watermark extracting probability P fproduce watermark detector output 128.False watermark detection can occur by two kinds of different modes.First, for arbitrary watermark extracting system, be present in not add and in watermark content, detect mistakenly the little of watermark but limited probability (being sometimes also called false positive detects).For instance, Secure Digital Music Initiative proposal (SDMI) and DVD audio request regulation are less than 10 -12the false positive watermark detection probability of/15 seconds assay intervals.This means, is necessary for the average time between false positive watermark detection not adding watermark content continuous monitoring 4,760,000 years.The vacation of Second Type detects when inspection has the content of watermark and the watermark value of the watermark value that is different from initial embedding detected and occurs.The vacation of this type detects (so-called error detection) can produce undesirable result.For instance, error detection can cause detecting " not allowing playback " watermark but not " not allowing copy " watermark potentially, and therefore, causes the playback of the content of legal purchase to stop.For the watermaking system of appropriate design, false detection rate should be probably the order of magnitude identical with false positive verification and measurement ratio as described above.
In some systems, with threshold value T, determine the output of the comparison operation being represented by equation (3).That is, comparator output can be represented by following equation (6).
C T ( W &prime; , W ) = 1 , m > T 0 , m &le; T - - - ( 6 ) .
In equation (6), m is that comparative result (and/or correlation) makes along with m increases false watermark detection probability P freduce monotonously.
Selected watermarking algorithm is depended in the calculating of comparator computing and false watermark detection probability.For instance, many watermarking algorithms based on spread spectrum are used the calculation of correlation between W ' and W to measure as a comparison m.In particular, estimate the coefficient correlation m between candidate watermark W ' and the pseudorandom pattern W of use when watermarked.Next, if estimated coefficient correlation m is greater than specific threshold T, report so successful watermark extracting.In the situation of unmarked content, through extracting the coefficient correlation of computing between false watermark pattern and long pseudorandom pattern, roughly following normal distribution, this means can use the false watermark detection probability P of well-known error function computing f.Along with coefficient correlation m increases, false watermark detection probability P freduce monotonously.Therefore, threshold value T and P fthere is mapping one to one.
Or, can be also bit string by candidate's watermark extracting, and then can mistake in using detection/correction code or with the number e of the direct pattern computing bit-errors of the position pattern matching of potential watermark.Then, can use the false watermark detection probability P of binomial distribution computing f.For instance, if watermark is N position, number long and bit-errors is less than or equal to e, can user's formula (7) calculate false detection probability so.
P f ( N , e ) = 2 - N &Sigma; j = 0 e N j - - - ( 7 ) .
If watermark is comprised of N=64 position and detector only finds e=5 bit-errors, so false watermark detection probability P fbe less than 10 -12.This P fcondition is enough to carry out concluding watermark extracting and (for instance) copyright enforcement completely conventionally.The false watermark detection probability P obtaining according to equation (7) falong with, the number m of the position of coupling (that is N-e) increases and reduce monotonously.
Comparison operation as described above only produces two results.In the first result, extract and report that watermark (for example) is to implement copyright rule.This result is sometimes referred to as and strong watermark detected.In the second result, do not find watermark and do not start other action.Except previously mentioned concluding watermark detection and no-watermark detection, the appointment that also can establish comparison program step detects the 3rd result of " tentative " watermark.For instance, the 3rd result can be specified wherein relatively to produce for strong watermark detection and is more than or equal to threshold value T but is less than the extraction situation of the value (for example, the number e of the vicious position of tool) of another threshold value t.Three outputs of comparison operation can be represented by following equation (8).
C T , t ( w &prime; , w ) = 1 , e < T 0.5 , T &le; e < t 0 , e &GreaterEqual; t - - - ( 8 ) .
Again consider that wherein watermark is comprised of bit string and situation that comparator output counting is comprised of the number of bit-errors.In above example, in 64 long strings, find nearly 5 mistakes can be enough to carry out concluding watermark extracting.Yet if detector finds 20 bit-errors, so false watermark detection probability is less than 0.2%, this makes watermark have possibility but is not enough to determine its use in extractor outside of proof.Therefore, can use and surpass 5 (that is, T=6) but be less than 21 (that is, arbitrary error count t=21) is by tentative watermark detection sender.If error count is more than or equal to 21 mistakes, so false watermark detection probability is greater than 0.2%, and therefore, can ignore watermark.
Fig. 2 shows the simplified block diagram of the watermark extracting apparatus 200 that can adapt to disclosed embodiment.Watermark extracting apparatus 200 can be the assembly that self-contained unit maybe can be incorporated as larger content disposal plant (for example content playback apparatus, content relay, content recording apparatus, Set Top Box etc.).At receiver 204 places, receive to import into through revising and embed content 202.In Fig. 2, receiver 204 is depicted as to the independent assembly in watermark extracting apparatus 200.However, it should be understood that receiver 204 can be compared with the assembly in the different sub-systems of media giant disposal plant maybe can be incorporated as a part for another assembly in watermark extracting apparatus 200.Owing to can being various forms and can comprising several audio frequency, video, multimedia or data-signal through revise embedding content 202, thus receiver 204 by through revise embed content 202 be suitably adjusted to can be can be by the appropriate form of other assembly identification of watermark extracting apparatus 200 necessary.Described adjusting can comprise signal processing operations, for example rectification, decompression, release of an interleave, deciphering, descrambling, resample, A/D conversion, reformatting, filtering etc.
Stego-key select assembly 208 then from by watermark extracting engine 206 for extracting at least one stego-key of Resource selection of watermarked stego-key.In certain embodiments, stego-key is stored in stego-key memory module 210.Watermark extracting result 216 is produced by watermark extracting engine 206.Watermark extracting result 216 can be indicated through recovering existence and/or value and the indication to the reliability through recovery watermark and out of Memory of watermark.Can predetermined time interval (for example,, to specific user or device) report watermark extracting result 216.In certain embodiments, distortion estimation assembly 212 produces being present in the estimation of type and the amount of one or more distortion in received content.Estimated distortion information can be in the situation that there is content distortion by watermark extracting engine 206 for promoting watermarked detection.Estimated distortion information also can be delivered to evaluator assembly 214, and the watermarked frame that evaluator assembly 214 is processed by watermark extracting engine 206 by assessment promotes watermarked detection.In certain embodiments, evaluator assembly 214 can be carried out for example operation of the following: to assigning weight through extracting potential watermark or watermark segmentation; Combination is through assigning weight; Assessment is through extracting watermark or whether meeting through accumulation watermark segmentation the false watermark detection probability etc. of wanting.In certain embodiments, watermark extracting result 216 can be by evaluator assembly 214 but not 206 outputs of watermark extracting engine.Distortion estimation assembly 212, evaluator assembly 214 and watermark extracting engine 206 can for example, be communicated by letter with the storage device 218 that contains predistortion watermark template and out of Memory (, database).To in following chapters and sections, discuss in detail predistortion watermark template.Storage device 218 can be the physical memory cell identical or different with stego-key memory module 210.It should be noted that as promoting the understanding to basic concept, the independent assembly in Fig. 2 graphic extension watermark extracting apparatus 200.However, it should be understood that in some embodiments, can be by both or both above the combining in described assembly.For instance, in some embodiments, one or many person in distortion estimation assembly 212, evaluator assembly 214 and storage device 218 can be incorporated as to a part for watermark extracting engine 206.
In some applications, if 216 indications of watermark extracting result detect at least one concluding watermark, so can be at outside other coercion action (such as stopping content playback, stop content record, showing alert message etc.) of carrying out of watermark extracting apparatus 200.If 216 indications of watermark extracting result detect at least one tentative watermark, through extracting the use of tentative watermark, can be limited in watermark extracting apparatus 200 with (for instance) the initial various adjustment to detector operation so.As early mentioned, can be based on the tentative watermark of threshold test, described threshold value depends on will trade off between tentative watermark use and processing load and robustness gain.
Many attacks intentionally or even conventional contents processing to digital watermarking operate the watermark distortion that causes changing watermark location, size, orientation and other parameter.For instance, record the film being shown on screen and can cause picture rotation, cutting in time domain or spatial domain, picture stretching, angle modification etc.Similarly, audio frequency watermark can stand time shift (for example, be connected in series and cause due to audio frequency cutting or with other audio frequency), time-scaling, pitch-shift, tremble and vibration etc.Can and make the anti-phase most distortions that compensate in described distortion of content distortion by Detection of content distortion before watermark extracting.Or, in certain embodiments, can be by adjusting watermark search utility step so that relevant to predistortion watermark template and completely or partially alleviate the impact of this type of distortion through extracting candidate's watermark.
Distortion detection and compensation can be manually or automatically, and it can be in the situation that have or do not have the original contents at detector place and realize.Clearly, the most desirable method is not used original contents to simplify watermark detection program step for providing automatic distortion to detect.Conventionally, this can realize in synchronous by introducing (or registration) watermark.
Fig. 3 A shows the exemplary watermarked frame of the length L that comprises sync section (SYNCH) and packet portion (PACKET).In Fig. 3 A exemplary graphic, in order to simplify, describe a dimensional view of watermarked frame (for example,, corresponding to audio content).However, it should be understood that similar principles for example, applicable to the watermark (being embedded in the watermark in image, video sequence etc.) with two or more dimensions.Synchronous and the packet portion of watermarked frame can comprise a plurality of symbols (that is, binary system or non-binary symbols) separately.In certain embodiments, the sync section of watermarked frame can be special style or the sequence being embedded in content host.So, the sync section of watermarked frame is sometimes referred to as synchronization watermarking, synchronous pattern, sync mark, synchronizing sequence etc.One in the function of sync section is for helping the watermarked frame in positioning host content.For instance, watermark extracting device can search content (for example, the sample of contents processing) to find the known synchronous pattern of the beginning of mark watermarked frame.Successfully detecting after sync section, watermark extracting device can be followed and at once attempt recovering the watermark bag symbol after sync section.In certain embodiments, the detection of synchronous pattern is with coarse-grained by the beginning sender of watermarked frame, and this can further trigger with fine granulation and carry out extra process to increase the reliability of watermark detection.In some systems, watermarked frame can be containing being useful on (for instance) with the position of different grain size level establishment watermarked frame and/or by two or more sync sections of the beginning of watermarked frame, end and/or other section sender.
In exemplary, extract in operation element flow process, after the successful compilation and detection of SYNCH, compilation PACKET.Yet, when there is the distortion of significant quantity in content, can be difficult to maybe can not detect SYNCH.Fig. 3 B is illustrated in successfully the PACKET#1 that collects after the SYNCH part of beginning of certification mark PACKET#1.Yet, due to lead in PACKET#2 and #3 each the distortion of SYNCH part and do not collect PACKET#2 and #3.For the further extraction of graphic extension watermark, consider that wherein audio content embeds and has watermark and subsequently owing to having a mind to or being not intended to contents processing operation and in time by the application of compression (for example,, by moving on the tone that is associated) linearly.Depend on specific watermark embedded technology, through the SYNCH of watermarked frame part, can before it detects unsuccessfully, tolerate the time-scaling of specified quantitative.In content, suffered in some situations of linear session convergent-divergent, if can estimated time the amount of convergent-divergent, can improve the immunity to time-scaling by the opposite direction content was carried out to convergent-divergent before attempting extracting SYNCH pattern so.In this way, make audio frequency time-scaling compression anti-phase (that is, being compensated) and watermark detection become feasible for compression infringement in a big way.
In another embodiment, alternative modified is through the content of time-scaling (or otherwise infringement), produce one or more predistortion synchronization template (for example,, through time compression synchronization template) and in the situation that not changing content itself, use it for and detect through embedding synchronous pattern.As early mentioned, with SYNCH pattern, establish the position of watermarked frame.SYNCH template is for being embedded in the predefine SYNCH pattern of the detection of the SYNCH pattern in content with promotion through formation.According to some embodiment, SYNCH trace routine step comprises processes received content to obtain candidate's SYNCH pattern and then candidate SYNCH pattern and one or more SYNCH template that can comprise a plurality of predistortion SYNCH templates to be compared.Analysis and empirical result indication produce one group of predistortion SYNCH template and provide and content carried out to convergent-divergent as much but manage the watermark detection reliability under cost in lower.
In order to promote the understanding to disclosed embodiment, consider that wherein SYNCH pattern comprises that the example of the sequence of 50 in the order section that uses specific embedding algorithm to be embedded in audio content is useful.Each of further hypothesis SYNCH pattern strides across 1 second of main frame audio content.For the audio content with 44,100 samples per second, each of SYNCH pattern is embedded on 44,100 samples of content.Detection is 44 of (1) collection received content from a kind of method of the SYNCH pattern of received content, 100 samples, (2) process described sample to extract candidate's place value, (3) repeating step (1) and (2) 49 times to be to form 50 candidate SYNCH patterns, and (4) compare to determine that by candidate SYNCH pattern and one or more SYNCH template whether candidate SYNCH pattern is really corresponding to the true SYNCH pattern with particular trusted level.
Owing to not knowing the position of SYNCH pattern in content, therefore can arrive (4) with the mode repeating step (1) of sliding window, wherein step (1) to (4) each repeatedly in, the treated sample of institute's audio reception content with previously repeatedly overlapping.In egregious cases, can arrive (4) 44 by repeating step (1), 100 times to produce the pond of 44,100 candidate SYNCH patterns with a sample searches granularity.Yet in most practical applications, this extreme method is infeasible on calculating, also to detecting, to have the embedding SYNCH pattern of rational reliability unnecessary.Therefore, conventionally for each watermark bit duration repeating step (1) to (4) several times.This is sometimes referred to as with X position granular search's content, and wherein X represents for each duration repeating step (1) to (4) or the number of times of step (1) and (2) at least.In certain embodiments, trace routine step can be roughly son position granularity (for example, 3 son positions) start, once and potential SYNCH pattern be detected and just repeat or restart to search for higher son position granularity.
In the situation that not there is not content distortion, can with the isolated candidate SYNCH of nominal range pattern place value, carry out SYNCH testing process referred to above each other by compilation.Figure 17 is the reduced graph that SYNCH collects and/or template produces that can carry out according to one exemplary embodiment.As early mentioned, pond, son position represents the candidate's place value producing with son position granularity.Nominal stencil-chosen operation in Figure 17 exemplary graphic comprises selects each the 5th son position (or more generally, separately the son position of d0).In above example, wherein SYNCH pattern is 50 bit sequences, selects 50 these type of son positions to form candidate SYNCH pattern.Then candidate SYNCH pattern and nominal (or not distortion) SYNCH template are compared to determine that it to what extent mates (how many not match bit for example, being detected).
For example, in the situation that there is distortion (time-scaling distortion), can be by forming candidate SYNCH pattern to be different from the spacing chooser position of nominal clearance.With reference to Figure 17, can for the distortion (for example, temporal extension) of the first kind with apart from d1 and for example, for the distortion (, the time shrinks) of Second Type with apart from d2 chooser position.In some situations, the candidate SYNCH pattern obtaining is compared to evaluate it with nominal SYNCH template to what extent mate based on the sub-column pitch of difference.
In certain embodiments, substitute candidate SYNCH pattern is mated with nominal SYNCH template, candidate SYNCH pattern and one or more predistortion SYNCH template that can be different from nominal SYNCH template are compared.Can obtain this type of predistortion SYNCH template by following operation: a content (or various content) is embedded SYNCH pattern, makes content suffer various distortions, and with nominal and/or through revising sub-column pitch from distortion contents extraction SYNCH pattern.For instance, with reference to Figure 17, can for example, by select N son position to obtain length N (, predistortion template #1 N=50), and can be by select the predistortion template #2 of a N son acquisition length N with d2 spacing with d1 spacing.Based on the sub-column pitch of difference, select predistortion template the most effective to the distortion of convergent-divergent type.For example, for some distortions (, acoustic propagation, intensity/Modulation and Amplitude Modulation, Video Camera are caught etc.), can be by obtaining predistortion template with nominal clearance d0 chooser position.
In general, can use many dimensions in predistortion SYNCH template development distortion space.Fig. 4 show wherein make original (or nominal) SYNCH template 402 (that is, the SYNCH template in the situation that not there is not content distortion) by various potential distortion A to K404,406,408,410 to produce one group of predistortion SYNCH template A to K414,416,418,420 example.Fig. 4 also describe can with the not distortion SYNCH template 412 of predistortion SYNCH template A to storage together with K414,416,418,420.The predistortion of template is applicable to the content of revising (comprising linearity or non-linear zoom (distortion) and rotation) in one or more dimension in frequency domain/time domain/spatial domain/amplitude domain.The non-exhaustive list that can affect the distortion that is inserted into the watermark in image comprises image rotation, in horizontal or vertical dimension, stretch, the dynamic range compression of illumination intensity signal, color balance etc.The non-exhaustive list that can affect the infringement that is inserted into the watermark in audio content comprises tone convergent-divergent, dynamic range compression, the equilibrium that linear session convergent-divergent, time-scaling, time that tone is constant are constant and resamples.
It should be noted that above discussion about SYNCH pattern can easily expand to comprises watermark bag.In addition, the principle that discloses may extend into SYNCH or the PACKET comprising in two or more dimensions that are embedded in spatial domain, time domain and/or frequency domain.This type of SYNCH or PACKET can be represented by binary system or non-binary symbols, and its detection can be based on binary system or non-binary symbols and/or parameter.For instance, can generally watermarked SYNCH be partly described as using the predefine signal pattern of relative program step extraction.
The similar PACKET compilation operation of the bag that Fig. 3 C graphic extension is identical with the bag of showing in Fig. 3 B, but operation in Fig. 3 C utilizes predistortion SYNCH template in the situation that there is content distortion, to realize the successful detection of embedding SYNCH.Treating capacity and false watermark detection probability depend on for detection of the number K that embeds the predistortion template of synchronous pattern.The number K of predistortion template depends on scope and granularity and the digital watermark sensitiveness to certain distortion of the distortion that needs adapt to.For instance, the technology that the digital watermark based on spread-spectrum is generally compared based on copying modulation time-scaling distortion is more responsive.Therefore,, when when copying modulation technique and compare, the more predistortion template of Technology Need based on spread-spectrum effectively detects watermark in the situation that of the distortion of life period convergent-divergent.
Consider that wherein service time, constant tone convergent-divergent (TIPS) was revised the situation that makes audio frequency watermark distortion.Further hypothesis predistortion watermark template is only taken into account following TIPS modification value: 3%, 6% and 9%.In this case, watermark extracting device can only detect to control oneself by about 3%, 6% or the watermark of the content of 9%TIPS infringement (plus-minus tolerance value) distortion be reported as tentative or conclusive reliably.In this exemplary situation, when audio frequency watermark is for example, during with arbitrary other TIPS ratio (, 4.5%) distortion, there is the possibility that can't detect concluding watermark.Clearly, the reliability of wanting that SYNCH detects is significantly depended in the selection of the number of predistortion template.Along with the granularity increase of predistortion template, the chance that reliable SYNCH (for example, having low dislocation miscount or high phase relation numerical value) detected also increases, but process load, also increases.Therefore, provide trading off as important between the number of predistortion synchronization template and false watermark detection probability.
In certain embodiments, the granularity of the predistortion synchronization template within the scope of certain distortion (and therefore, the number of predistortion template) through select to make to detect at least single test watermark in distortion content.For instance, even if select the number of predistortion synchronization template and spacing to make distortion in the middle of two predistortion templates (as in above example) in a mode, still exist and the follow-up good opportunity that tentative watermark then detected of SYNCH pattern detected.
Once find tentative watermark, extractor just can start the search more meticulously in distortion space.One group of operation 500 that Fig. 5 graphic extension is carried out according to example embodiments.At 502 places, receive watermarked frame.Institute's received frame can comprise synchronously (SYNCH) and watermark bag (PACKET).At 504 places, received SYNCH and a plurality of predistortion SYNCH template are compared.At 506 places, determine whether successfully to detect SYNCH, and if (that is, at 506 places, being "Yes") like this, so described group of operation 500 continues compilation PACKET at 508 places.On the other hand, if determining of 506 places do not indicate successful SYNCH to detect (that is, being "No" at 506 places), so described group of operation 500 continues at 520 places, wherein abandons PACKET.
Back, with reference to figure 5, at 508 places, PACKET compilation can comprise various operations, and for example release of an interleave, descrambling and other bag form operation.At 510 places, determine the wrong number in compilation PACKET.For instance, the operation at 510 places can comprise that mistake in using correction/error detecting code is decoded, template matches etc.At 512 places, determine whether wrong number in compilation PACKET is less than the first threshold that the detection with strong (that is, concluding) watermark is associated, and if so (, at 512 places, be "Yes"), in the declaration of 514 places, concluding watermark detected so.If 512 places really definiteness show be greater than first threshold improper value (, at 512 places, be "No"), so described group of operation 500 continues at 516 places, wherein determines whether the wrong number in compilation PACKET is less than the Second Threshold being associated with the detection of tentative watermark.If 516 places are defined as "Yes", so described group of operation 500 continues at 518 places, and wherein declaration detects tentative watermark.On the other hand, if 516 places be designated as "No", so described group of operation 500 continues at 520 places, and abandons PACKET.
Fig. 6 is the reduced graph that graphic extension can be used for the assembly 600 of the some or all of operations in the operation described in execution graph 5.Watermarked frame receiver 602 is configured to receive SYNCH and the PACKET watermarked frame partly that comprises watermarked frame.Comparator 604 (a) be configured to receive SYNCH and by itself and predistortion synchronization template and not one or many person in distortion SYNCH template compare.The result of described comparison can be by distortion estimator 604 (b) for obtaining being present in the estimation of one or more distortion of content.For instance, distortion estimator can be identified the predistortion synchronization template of the optimum Match of generation and the SYNCH that receives, and selects to represent to be present in the distortion in content be associated through identification predistortion SYNCH template one or more type of distortion and/or amount distortion.In certain embodiments, comparator 604 (a) and distortion estimator 604 (b) are combined into single component (for example, comparator 604 (a)).
Template can be stored in storage 606 assemblies.PACKET assembly components 608 be configured to receive PACKET at least and about comparator 604 (a) obtain the SYNCH that receives and predistortion synchronization template or the not success of mating between one or many person in distortion SYNCH template or failed indication.After SYNCH being detected successfully, at once the collect PACKET of watermarked frame of PACKET assembly components 608, described PACKET determines that by PACKET mistake assembly 610 is for generation of error count.The error count that produces by the evaluator 612 that comprises decision logic assembly for determining that error count is to arrest limit in specified packet error range or higher or lower than described specified packet error range.For instance, evaluator 612 can determine whether to detect concluding or tentative watermark (and the output of indicating it is provided).Evaluator 612 also can produce and will detect SYNCH it fails to match or abandon due to comparator 604 (a) and/or distortion estimator 604 (b) indication of PACKET when can't detect tentative or concluding watermark.The output of evaluator 612 can be used for triggering operation bidirectional, for example, when tentative watermark being detected, trigger expanding packet search.The output of evaluator 612 can comprise extraneous information, for example watermark value (, watermark state, watermark pattern etc. after testing), the indication of the credible level in this type of watermark detection (for example, bag error count, timing error counting etc.), be present in the amount of one or more distortion in content and type etc.
In conjunction with Fig. 5 and 6 described multistage methods, realize the overall process load and the efficient balance of false positive watermark detection probability that is associated in watermark extracting device, and further improve overall robustness and the fail safe of watermark extracting.
In certain embodiments, the two steps search of can a mode implementing distortion space trigger mating between predistortion SYNCH template and detected candidate SYNCH pattern to expand extractor to operate immediately, for example, expand position and granularity to the search of synchronous pattern and/or watermark bag.Yet this of extractor operation is expanded immediately and may be created in the optimal compromise of processing load and false positive watermark detection probability aspect.For instance, suppose to wish improvement concluding watermark extracting performance in constant time-scaling (PITS) distortion range of-20% to+20% tone.Must carefully select to search for granularity to insert rightly predistortion SYNCH template database 606.For example, if select fine search granularity (, thering is 1%PITS spacing), can utilize so a large amount of predistortion templates (for example, 40 predistortion SYNCH templates and not distortion SYNCH template) to detect reliably SYNCH pattern.In this situation, described a large amount of predistortion SYNCH templates can negatively affect processes load and false watermark detection probability.In addition, if predistortion template matches has high false SYNCH detection probability, each SYNCH detects and can trigger the operation of expansion extractor so, and this will increase processes load and false watermark detection probability.On the other hand, for example, if select rough search granularity (, thering is 5%PITS spacing), exist so and can't detect the possibility that embeds SYNCH and corresponding watermark PACKET.In addition,, if predistortion template matches has too low false SYNCH detection probability, the many watermarks in distortion content may not be detected so.Yet this defect can compensate by take the number that extra process load is that cost increases predistortion template.Therefore, producing a finite population predistortion synchronization template and the mode of take in watermark extracting device utilizes it to make the balance of improving true watermark detection, processing load and false watermark detection probability as favourable.
Fig. 7 graphic extension can be through carrying out to produce one group of operation 700 of one group of predistortion watermark template according to one exemplary embodiment.It should be noted that described group in describing Fig. 7 operates at 700 o'clock, with reference to predistortion watermark template.However, it should be understood that and can produce this class template for the one or both in synchronous pattern and known watermark ladle sample formula.For instance, having the watermark of two pay(useful) loads can be only by the sender that exists of the one in four different conditions.In this system, unique symbol pattern can be associated with each in watermark state, and for each unique symbol pattern, can carry out the described group of operation 700 of describing in conjunction with Fig. 7.
With reference to figure 7, at 702 places, produce one or more pilot signal.These signals can be corresponding to specific function, for example pulse, SIN function, square-wave, plane picture, edge image etc.It should be noted that use term " signal " to pass on can numeral or analog form transmitting, receive and the content of the particular type of storage.This type of signal can suffer signal processing operations, and for example modulating/demodulating system, A/D and D/A change, compress, resample etc.At 702 places, produce (or otherwise obtaining) once pilot signal embed and have synchronous pattern and/or watermark bag and distortion (or launching by noisy communication channel) just can provide about the characteristic of distortion and transmitting channel and this type of distortion and/or transmitting how to affect watermarked seeing clearly.In alternate embodiment, substitute pilot signal, at 702 places, produce or obtain additionally or alternati representative content (for example, the serial connection of dissimilar film).
At 704 places, pilot signal is embedded watermark.At 704 places, embed pilot signal and experience one or more distortion.These distortions can be including but not limited to various time-scaling distortions, A/D and D/A distortion, cutting, rotation, convergent-divergent, space and/or time shift etc.At 706 places, can use one or more distortion parameter that can stride across particular range that described distortion is applied to main frame pilot signal.For instance, can be according to linear session convergent-divergent (LTS) the percentage parameter application LTS distortion in the scope between-10% to+10% with 1% increment.At 708 places, from distortion pilot signal, extract watermark.At 710 places, for each successful watermark extracting, produce corresponding to predistortion candidate's watermark template with the certain distortion of certain distortion parameter.At 712 places, from field, receive one group of distortion watermark.In an example, after main frame pilot signal has experienced the real world transmitting by noisy communication channel, after embedding by the watermark at 704 places, from main frame pilot signal extraction watermark, produce 712 described group of distortion watermark.For example, contain watermarked main frame pilot signal playback at the cinema, use Video Camera to catch and be input to watermark extracting device to obtain distortion watermark.
At 714 places, received distortion watermark and candidate's predistortion watermark template are compared.This which one that relatively can disclose in candidate's predistortion watermark template is optimum Match or relevant.At 716 places, the predistortion of being correlated with watermark template classification (for example,, to be correlated with or the order of optimum Match) and being stored in database.In certain embodiments, described database can comprise the described predistortion watermark template of a subgroup that produces tentative watermark or strong watermark.Note, can when the new distortion watermark obtaining from field, by new template, upgrade and/or supplement predistortion template database.
Described in Fig. 7 exemplary graphic, can be for each distortion in particular range and the program step that repeats the experimental selection of predistortion watermark template for the combination of distortion.In certain embodiments, for the number of the distortion that pilot signal is carried out is limited in actual limit, first identify critical distortion vector (distortion for example, causing due to the time/space in particular range/frequency convergent-divergent is attacked, Video Camera is caught, low bit rate perception compression etc.).Then, determine predistortion watermark template and the combination thereof of critical distortion vector.Can contrast all other distortion vectors of predistortion template test (for example, non-critical distortion vector) of critical distortion vector to determine whether to need extra predistortion watermark template.
Fig. 8 is the reduced graph that can be used for the assembly of the some or all of operations in the operation 700 of execution graph 7.Embedded equipment 802 is for embedding the pilot signal of the various distortions of being used subsequently distortion processing components 804.The distortion being produced by distortion processing components 804 embeds pilot signal and is imported into extractor assembly 806, wherein produces corresponding to the predistortion watermark template with the certain distortion of certain distortion parameter.In Fig. 8, these watermark templates are denoted as to " WM template 1 ", " WM template 2 " etc.Fig. 8 also describes from what be input to that the field of extractor assembly 808 obtains to embed navigation content through revising.In certain embodiments, extractor assembly 806 and extractor assembly 808 are same physical device.(for example use therein in the embodiment of distributing processing) in other embodiments, extractor assembly 806 and extractor assembly 808 can be the different physical entities that can be positioned at different physical locations.
The further graphic extension of Fig. 8 compares and makes its template matches of mating and classification component 810 by the distortion watermark from field with the predistortion candidate's watermark template being produced by extractor assembly 806.The coupling of various distortions and the result store of sort operation are in database 812.
In certain embodiments, in order to reduce to process load and false watermark detection probability, extractor can be skipped one or many person in the predistortion watermark template at arbitrary candidate's watermark location place.In following chapters and sections, the method is sometimes referred to as the sparse search in distortion space.A shortcoming skipping the predistortion watermark of a subgroup is the opportunity that extractor can lose watermark extracting.Yet if the number on opportunity is fully large, the general impacts on the opportunity of missing so can be insignificant.In addition, in certain embodiments, for example, if extractor (was located in another stage of watermark extracting program step, under difference extraction opportunity) find tentative watermark, extractor can be configured to some opportunitys in opportunity that (for instance) tentative watermark extrapolation exploitation based on below discussing misses so.In a kind of version, only access has the opportunity of missing of high successful watermark detection probability and is not all opportunitys of missing.This realizes with those opportunitys of (one or some) distortion parameter that tentative watermark is associated after testing by again accessing only coupling.
During sparse search in design in distortion space, the embodiment that discloses can utilize following two features.First, random (or pseudorandom) selects the predistortion watermark template of a subgroup.If this decision-making is deterministic, assailant can experiment method or is found certainty pattern by reverse-engineering so, and adjust subsequently distortion pattern with the coupling institute search volume of being skipped with escape watermark detection.
Secondly, if extractor is known the possibility of the certain distortion of existence, be favourable.For instance, if " cinema projection " state in watermark conveys audio-visual content, so typical pirate situation relates to recorded content.For this, attack situation, for example, for example, to be applicable to the predistortion template template of cutting, stretching, rotation etc. (, corresponding to) of type than the high probability selection of other predistortion template (, time/frequency convergent-divergent etc.).Similarly, if being digital encryption form, watermark conveys instruction content " trusts and originates " state, so the predistortion template of perception code conversion for attacking and select with high probability to be associated with perception compression.In general, the probability of predistortion stencil-chosen is proportional with the probability that is present in the certain distortion in content, as long as can estimate this probability.
In certain embodiments, based on two or more different watermarks, carry out distortion computing.Two or more different watermarks can comprise the combination of independent SYNCH pattern, routine tests watermark or SYNCH and tentative watermark.When those watermarks have predefine correlation and described pass at embedded device place, be extractor when known, can be by any these quality and quantity that deviate to come computing distortion of relation therewith.For instance, if along two synchronous patterns in horizontal line embedded images and its make after testing to connect its line and horizontal line at angle, can suppose that so image is rotated detected angle.Similarly, if periodically embed audio frequency watermark and extractor finds the isolated audio frequency watermark with P ' with cycle P, can suppose that so content is by time-scaling factor P '/P.
Distortion computing based on independent SYNCH pattern can have specific advantages, the processing load for example reducing.Yet service test watermark but not SYNCH pattern can provide superior function, because tentative watermark has lower false detection probability and better position resolution conventionally.As early mentioned, when comparing with the detection of watermark bag, the detection of SYNCH pattern is characterized by higher false detection probability conventionally.Doing is like this processing load and the robustness requirement for balance extraction procedure step.Therefore, the distortion computing based on independent embedding SYNCH pattern can produce incorrect or unreliable distortion estimation.Yet if tentative watermark detected, it provides the existence of watermark after testing and the extra confidence level of position so.
Once based on two or more watermark computing amount distortion, whether next step is real for assessing distortion so.For instance, suppose that the image rotation during video record is safe by the time compression that is not more than 20 degree (with respect to horizontal or vertical baseline) or audio track by being not more than 50%.The arbitrary amount distortion that is greater than those predefine limits all may make image or the audio frequency can not perception, and therefore, the distortion computation that produces the value that exceeds those limits may be owing to false watermark detection.
When real amount distortion being detected, then can select and/or produce the extra predistortion watermark template corresponding to the distortion in the neighborhood through computing distortion, and described extra predistortion watermark template is for other watermark search.Predistortion watermark template can be corresponding to the SYNCH part of watermarked frame, the SYNCH through combination S YNCH and PACKET part or watermarked frame of the PACKET part of watermarked frame, watermarked frame or PACKET sub-fraction partly.Can repeatedly make extra predistortion watermark template to relevant to increase the reliability of watermark detection and the reliability of distortion computing through extracting watermark.For instance, by the watermark extracting of new watermark template, can produce and have than few wrong of the tentative watermark of previous detection or than the watermark of its high correlation (and so than its high reliability).Then, can be by the rerun amount of estimated distortion of the watermark after testing with higher reliability.For example, in some application in real time (, when occurring during the real-time playback of watermark extracting in content), can be difficult to carry out the repeat search to the same segmentation of content.Therefore, in certain embodiments, extra predistortion watermark template is only for subsequent content segmentation.
Fig. 9 graphic extension can be through carrying out to promote one group of operation 900 of the search of distortion according to example embodiments.At 902 places, extract tentative watermark.At 904 places, the detection distortion estimator amount based on two or more tentative watermarks.For instance, as early mentioned, can with two after testing the spacing between watermark estimate the amount of the linear session convergent-divergent in audio content.At 906 places, determine that whether estimated amount distortion is lower than specific threshold.The operation at 906 places provides in order to determine whether estimated distortion is real mechanism.If estimated distortion, higher than threshold value, abandons distortion estimation at 912 places so.Now, described group of operation 900 turns back to 902 to wait for and can be used for the distortion estimation at repetition 904 places until the feasible tentative watermark of another group becomes, or with another group can with tentative watermark produce another distortion estimation.
If distortion estimation is lower than threshold value (that is, being "Yes" at 906 places), at 908 places, select and/or produce extra predistortion watermark template so.Can use and allow to produce the algorithm of new predistortion watermark template and " generation " of the predistortion watermark template that associated parameter realizes 908 places.Or or in addition, can select suitable predistortion watermark template from database.At 910 places, by extra predistortion watermark template, detect and there is the watermark of higher reliability and/or improve the estimation to content distortion.
In distortion content, tentative watermark band is relevant for the uncertainty of its accurate location and parameter.Therefore, distortion computing is also approximate.As mentioned in conjunction with Fig. 9, can select at 908 places extra predistortion watermark template to improve watermark detection and/or distortion estimation.In certain embodiments, in a mode, select extra predistortion watermark template to realize more efficient and reliable watermark detection and/or distortion estimation.In particular, when producing predistortion watermark template, in predefine mode, make a plurality of content distortions that have watermark, detect tentative watermark, based on tentative watermark estimation distortion after testing, and collect and the wrong statistics being associated through computing distortion.For example, when (selecting/produce extra predistortion watermark template at watermark extracting device place, operation 908 in Fig. 9) time, not only based on estimated distortion (for example,, from operating 904 acquisitions) and the distortion based in the scope of anticipation error, select/produce this type of extra predistortion watermark template.Therefore, for each estimated amount distortion, an optional majority predistortion watermark template, all described predistortion watermark templates are near estimated distortion.In this way, significantly improve the chance of extracting the watermark with high reliability, and significantly do not increase the number that template matches is attempted.
In certain embodiments, use single test watermark to carry out the estimation to distortion.As early mentioned, in order to reduce processing load and the false watermark detection probability of extractor, can carry out watermarked initial ranging by the predistortion watermark template of a subgroup only.The template of this subgroup realizes the rough search in distortion space.When using the coarse in distortion space single test watermark to be detected, if near the distortion parameter of detection that triggers tentative watermark with compared with fine granulation search distortion space, there is so the good opportunity that can find more reliable watermark.
Figure 10 graphic extension can be through carrying out with from one group of watermarked operation 1000 of distortion contents extraction according to one exemplary embodiment.At 1002 places, receive watermarked frame, and at 1004 places, carry out the multi-step search to tentative watermark.In an example, use described group of operation described in conjunction with Figure 5 to carry out the operation at 1002 and 1004 places.At 1006 places, determine whether to detect tentative watermark.If tentative watermark do not detected, operate so 1000 and continue at 1018 places, wherein next tentative watermark of system wait.If 1006 places really definiteness show and tentative watermark detected, operate so 1000 and continue at 1008 places, wherein distortion estimator information.For instance, to the estimation of distortion information, can comprise the amount that is present in the certain distortion (or distortion combination) in received content of determining.In one embodiment, can carry out this estimation by following operation: one or more SYNCH part of compilation watermarked frame (or otherwise obtaining the part through compilation SYNCH); And itself and one or more predistortion SYNCH template corresponding to the certain distortion (or distortion combination) of various amounts are compared.As early mentioned, in certain embodiments, according to the rough search in distortion space, carry out the detection to the estimation of amount distortion and/or tentative watermark.The estimated distortion information of the certain distortion that the watermark detection producing according to this rough search provides indication to be present in to embed in content host or the coarse value of distortion combination.
At 1010 places, select and/or produce extra (or supplementing) predistortion watermark template with carry out distortion space compared with fine search.The template of selecting in addition/producing is corresponding to the distortion value in the neighborhood of (roughly) distortion value obtaining at 1008 places.The TIPS that considers audio content revises example, and wherein the predistortion watermark template of initial set allows the rough or sparse search in the time-scaling space in 3%, 6% and 9% value.If make audio content distortion with 3.5% in this example during the rough search in distortion space, can use so the predistortion watermark template being associated with 3%TIPS to obtain the tentative watermark bag compilation of the follow-up then success of best SYNCH compilation.For this example, select additional distortion watermark template (that is, the operation at 1010 places) to comprise to select or be created in for example, predistortion watermark template in the neighborhood of 3%TIPS distortion (, with 0.1% stepping 2.5% to 3.5%).
At 1012 places, the number of the bag mistake of using newly the predistortion watermark template of selecting/producing to reappraise to be associated with tentative watermark.In one embodiment, PACKET candidate being recovered compares to produce with the some or all of extra predistortion watermark template in extra predistortion watermark template the error count that is associated.So, the predistortion PACKET template of new a group can be used for to wherein detect tentative watermark same content segmentation repeat search or for the search of subsequent content segmentation.
At 1014 places, select to produce bag or the bag template of minimal error number.Selected bag/bag template is corresponding to representing to be present in more accurately the amount distortion of the true amount distortion in received content.Therefore,, in some situations, the bag mistake through reappraising can be indicated watermarked the existing (and more accurate location) with higher reliability.For instance, the operation at 1014 places can cause detecting strong or concluding watermark.At 1016 places, based on produce the distortion information of the bag/template renewal of best improper value for collect following bag and/or synchronous pattern at 1014 places.
Although it should be noted that the operation at 1014 places is described as selecting to produce least number of errors object bag/template, the operation at 1012 and 1014 places also should produce the selection corresponding to low false positive detection probability.In alternate embodiment, by repetitive process, realize extraction and the estimation to distortion of watermark.In this embodiment, select one at a time/be created in the supplementary predistortion watermark template in the neighborhood of distortion of previous estimation.For instance, the first supplementary predistortion watermark template can be corresponding to the less slightly content distortion of the predistortion watermark template of the tentative watermark detection than for 1004 places.If first supplements predistortion watermark template, produce less bits mistake or higher relevant to candidate's watermark, by second, supplement predistortion watermark stencil-chosen for the content distortion corresponding to being less than the first supplementary predistortion watermark template so.On the other hand, if first supplements the correlation that predistortion watermark template causes larger bit-errors number or reduces, by second, supplement predistortion watermark stencil-chosen for the content distortion corresponding to being greater than the distortion being associated with the predistortion watermark template of tentative watermark detection for 1004 places so.Iteration can continue until the one in predistortion watermark template provides best watermark detection and the best estimate to content distortion.In some instances, can use numerical method (for example dichotomy or Newton method) to promote the convergence of iteration.Iterative process as described above obtains the most reliably estimating of watermark after testing more efficiently and with the low false detection probability of a plurality of search than under fine granulation.
In certain embodiments, after tentative watermark being detected, at once with extrapolation technique, improve the reliability of watermark detection.In distortion content, although the effort that indivedual watermark can be corrupted to distortion computing as described above and compensation is possible degree without concluding watermark detection.In these cases, extractor can continue the search to the follow-up watermark in content near tentative watermark after testing.Yet, if find one or more extra tentative watermark in subsequent searches, associating " weight " that so can two or more watermarks of computing and existing with its watermark of further reappraising.
In certain embodiments, can adopt and depend on the alternative method that watermark embedding pattern is predicted position, the Symbol Style and the watermark parameters of adjacent watermark.Conventionally, expection spread all over content same watermark repeat embed, but more complicated embedding pattern is also possible, and in some cases, for favourable.For instance, in some situations, adjacent watermark can be in a known way (for example, when counter that the watermarked embedding comprising with each extra watermark bag increases progressively, or when with different scrambling key during by the pay(useful) load scrambling of adjacent watermark bag) differs from one another.In the situation of the embedding pattern of subsequently known and/or previous watermark, the detection of the tentative watermark of can extrapolating is with other watermarked the existing and be worth in the expectation position in evaluation content therein.In certain embodiments, use tentative watermark after testing to carry out the existence of other watermark pattern in predictive content.Test subsequently this prediction (for example, using hypothesis test statistics method), this can cause refusal or the confirmation of the feasibility of watermark detection.
Figure 11 A shows to have SYNCH and the PACKET exemplary watermarked frame sequence partly that is similar to those parts that present in Fig. 3 A.After successfully collecting and detecting SYNCH, compilation PACKET.When the distortion of significant quantity is applied to content, can be difficult to maybe can not detect the SYNCH part of watermarked frame.Figure 11 B is illustrated in the PACKET#1 that collects after previous SYNCH that successfully collects.Yet, do not collect PACKET#2 and #3 due to the serious distortion of the corresponding SYNCH part of watermarked frame #2 and #3.Figure 11 C displaying wherein, after tentative watermark being detected, is crossed over the example that watermarked frame #2 and #3 carry out (expection comprises PACKET#2 and #3) extrapolation of tentative watermark.Typical case, extrapolate in situation, the associated symbols that expection some Symbol matchings in the symbol of extrapolation watermark are not extrapolated in (that is, tentative) watermark.At compilation, after the symbol of extrapolation watermark, at once count symbol error, estimate false watermark detection probability and to assigning weight through extrapolation watermark.Can add up through assigning weight jointly to evaluate existing and being worth of watermark.
The extrapolation technique of the embodiment that discloses comprises to improve and detects watermarked possibility and make false watermark detection probability maintain the feature in desirable limit simultaneously.In particular, in certain embodiments, when extrapolating to search for extra watermark, do not need to find extra SYNCH pattern to collect ensuing watermark bag symbol.In distortion content, many SYNCH patterns may be damaged and cannot identification.Yet even if the SYNCH part of watermarked frame do not detected for all potential watermarked frame, the extrapolation technique of the application's case is still realized compilation and the use of bag symbol.
In addition, in some extrapolation techniques in the extrapolation technique of disclosed embodiment, can be low to 1 in order to detect the number of attempting through the extraction of extrapolation watermark.Therefore, do not need to adjust to being wanted false detection probability to maintain through the weight of extrapolation watermark symbol appointment.Yet, in a particular embodiment, in order to take into account the uncertainty being associated with distortion estimation and the true place of tentative watermark, can use several start skew and several distortion estimation execution watermark extrapolation trials.The method promotes to realize the optimal compromise between robustness and false detection probability.Even in this situation, the number of trial still significantly triggers in the cumulative system of weight lower than the independent detection based on watermarked therein.In this watermarked independent detection, need to the be associated detection of SYNCH part of the detection of each watermark (for example, each tentative watermark), as illustrated in Fig. 3 B and 11B.
Figure 12 shows the exemplary plot of false watermark detection probability to watermark weight threshold.The experimental result that exemplary plot in Figure 12 obtains corresponding to the watermarked frame for having binary character, described binary character has 32 SYNCH parts and 64 the PACKET parts that are embedded in audio content.Curve 1220 is corresponding to the sparse granular search in distortion space, and the fine granulation that curve 1240 obtains corresponding to the watermark extrapolation technique that uses the embodiment that discloses is searched for.Figure 12 also shows corresponding to the independent recovery operation based on watermark to realize the 3rd curve 1260 of the false watermark detection probability of the cumulative extractor of weight.Note, for about 5 * 10 -9false watermark detection probability, use the watermark weight threshold of the tentative watermark extrapolation technique that discloses (sparse and fine search both) to be less than the threshold value that independent weight adds up and uses in technology.So, compare with the cumulative technology of weight of independent detection based on watermarked, the extrapolation of the tentative watermark based on previous detection the extrapolation technique that discloses under lower computing load, produce similar or better false positive performance.Meanwhile, relative with the detection of concluding watermark, the detection of the extrapolation of the tentative watermark based on previous detection need to be compared with low watermark detection threshold for the detection of tentative watermark.These features are greatly improved the robustness performance of watermark extracting device.
Figure 18 graphic extension can be through carrying out one group of operation 1800 with extrapolation watermark according to one exemplary embodiment.At 1802 places, extract tentative watermark.For instance, the operation at 1802 places can comprise in conjunction with at least some operations in Fig. 5,9 and 10 operations of discussing.At 1804 places, form one or more through extrapolation watermark.For instance, can be positioned to embed in content host with respect to realizing this bag and form through extracting the symbol of potential watermarked frame of the predefine position of tentative watermark by acquisition.At 1806 places, common evaluation through extrapolation watermark and the tentative watermark of having extracted.At 1808 places, determine whether the common evaluation of watermark causes meeting the false watermark detection probability of wanting.For instance, this wants the false watermark detection probability can be corresponding to the detection probability being associated with concluding watermark.
The amount of the watermark extrapolation of carrying out according to disclosed embodiment depends on the essence of distortion and the accuracy of distortion estimation.For instance, in the situation of the audio frequency watermark of the distortion due to the acoustic propagation of content host, when background noise and reverberation appear to be main distortion sources, time-scaling distortion is generally insignificant.In this environment, assumption period ground is watermarked, may stride across the successful extrapolation of several watermark length.On the other hand, if content is attacked by time-scaling, can limit the accuracy that convergent-divergent distortion is associated with estimated time so, and therefore, only a peanut watermark extrapolation can be on each time orientation feasible.
In some cases, distortion changes on time or space because content converts or attacks intentionally.For instance, known audio frequency tape player is trembled or trembles illusion because the variation of the speed of the motor of tape player produces.Therefore the variation of the time-scaling factor, producing based on motor speed and changing in time.Assailant can adopt similar techniques to disturb appropriate estimation and the compensation of the time-scaling distortion in the content that has watermark.Yet warning is for too soon and/or the distortion that changes of the too large time feasibility that conventionally produces offensive perception illusion and reduce to attack.Therefore, exploitation exist in the situation that the distortion that changes of the time slowly changing in the watermark duration increase the more effectively antagonism of the robustness of watermark extracting miss can be useful.
The experiment of being undertaken by inventor is indicated, and while slowly changing in the length of distortion in watermark, conventionally by the general predistortion SYNCH template that is applicable to certain within the scope of distortion variations, realizes tentative watermark detection.Yet, within the duration of watermark PACKET that is conventionally being greater than SYNCH part, the distortion value outside the detection range of predistortion watermark template that can fluctuate.This means that predistortion bag template can only mate the actual distortion of the limited section of watermarked frame, causes whereby remarkable symbol error rate.
According to some embodiment, (for example know in advance watermarked the Symbol Style therein, by the tentative watermark pay(useful) load of previous detection, by considering only one group of limited pay(useful) load etc.) situation in, whole watermark interval is divided into some segmentations and carries out watermark extracting trial for each segmentation.Can understand better this operation with reference to figure 11D.Note, the horizontal scale in Figure 11 D can be different from Figure 11 A to those horizontal scales in 11C.With the distortion estimation that the SYNCH from PACKET#1 detects and/or tentative watermark detection obtains, use the extraction of attempting the first watermark segmentation SEG1 corresponding to one or many person in the predistortion watermark template of estimated distortion value.In certain embodiments, all or part of after PACKET#1 of the expection that SEG1 comprises SYNCH.For example, if watermarked frame is comprised of 20 SYNCH and 100 PACKET, so in an example, each segmentation can comprise 20 positions (that is, 6 segmentation/frames).In another example, each segmentation can comprise 30 positions (that is, 4 segmentation/frames).In a further example, each segmentation can comprise 10 positions (that is, 12 segmentation/frames).In an example again, can skip SYNCH part, and can only carry out the segmentation coupling of the PACKET part of watermarked frame.Above example does not intend to provide the exhaustive list of different fragment size and segmented configuration, but can needs and ability based on watermark detection system and target application select different fragment size and configuration through providing with graphic extension.
In certain embodiments, based on for example, select the number of segmentation through extracting the scope (, duration, spatial dimension etc.) of tentative watermark.For instance, the watermark with less pay(useful) load can be segmented into the segmentation fewer than the watermark with larger pay(useful) load.The other factors of number of determining segmentation is (or expecting to exist) amount distortion in embedding content host and/or have (or expecting to exist) type of distortion in embedding content host for existing.The selection of the number of segmentation can dynamically change based on (for instance) amount distortion and type, or can carry out statically or semi-statically, and wherein the number of segmentation is fixing.In addition the expectation amount distortion in, can be content-based and/or the length of each segmentation of type selecting.
Back with reference to figure 11D, in certain embodiments, the predistortion watermark template that approaches estimated distortion value is most for SEG1 and detecting through extrapolation corresponding near one or more predistortion watermark template of the distortion estimated distortion value.No matter which coupling attempts producing optimum, and the distortion levels that is associated is as the beginning distortion estimation of next segmentation SEG2.Operation referred to above for through extrapolation watermark other segmentation SEG2 ..., SEG K continues.In this way, can effectively follow the tracks of and alleviate by utilizing appropriate predistortion watermark template the distortion of slow variation.
Figure 13 graphic extension can be through carrying out one group of operation 1300 with the watermark of extrapolating on by segmentation basis according to one exemplary embodiment.At 1302 places, receive watermarked frame, and at 1304 places, carry out the multi-step search to tentative watermark.In an example, use described group of operation described in conjunction with Figure 5 to carry out the operation at 1302 and 1304 places.At 1306 places, determine whether to detect tentative watermark.If tentative watermark do not detected, so described group of operation 1300 continues at 1320 places, wherein system wait next frame, and the operation that then repetition is carried out at 1302 to 1306 places.If 1306 places really definiteness show and tentative watermark (that is, being "Yes" at 1306 places) detected, so described group of operation 1300 continues at 1308 places, wherein distortion estimator information.For instance, to the estimation of distortion information, can comprise and determine amount and the type that is present in the distortion (or combination of distortion) in received content.In one embodiment, can carry out this estimation by following operation: the SYNCH part of compilation watermarked frame (or obtaining the SYNCH part having collected); And itself and one or more predistortion SYNCH template corresponding to one (or some) certain distortion of various amounts are compared.As early mentioned, conventionally according to the rough search in distortion space, carry out the detection to the estimation of amount distortion and/or tentative watermark.
At 1310 places, determine the granularity of search.Granularity based on particular default selection search (for example at first, the number of predistortion bag and the spacing in distortion space thereof), can for granularity described in initial fragment design develop and/or described granularity, can the mistake based on by segmentation basis minimize criterion based on estimated distortion information.Can select described criterion based on wanted false watermark detection probability, robustness and computational complexity.In certain embodiments, in response to observed or expect distortion and adjust described criterion.Based on mistake, minimize criterion, the predistortion watermark template at segmentation stage place (it can reside in watermark extracting device place or remotely reside in database place in this locality) is rearranged and reclassify to meet mistake and minimize criterion.In addition, renewable for selecting the selective rule of predistortion watermark template so that watermark extrapolator can be selected one group of optimum pre-distortion watermark segmentation template.In certain embodiments, mistake minimizes the selection that criterion further triggers the new stego-key that can be used for extracting follow-up watermark.Can a mode carry out stego-key selection to optimize at least one in robustness, fail safe and the computational efficiency of the extraction that is embedded in the watermark in content host.In addition, the selection of one or more stego-key can be suitable for expecting the conversion of content host.For instance, described conversion can be changed the outward appearance of at least one watermark embedding with the first embedding stego-key, and described at least one watermarked seeming with the second embedding stego-key is embedded.
Back, with reference to Figure 13, at 1312 places, carry out the watermark extrapolation of the segmentation of considering.In certain embodiments, the operation at 1312 places comprises one or more predistortion watermark template at the symbol of considered segmentation and segmentation stage place is compared.For instance, can use different "or" by turn (XOR) computing between the position of fragment bit and predistortion watermark template to carry out this relatively.Granularity and estimated distortion information based on search, can carry out one or more extrapolation trial that (for instance) produces one or more improper value.The operation at 1312 places further comprises selects best matching result (for example, producing least number of errors object result).At 1314 places, optionally determine whether processed all segmentations.Can by by the number of processed so far segmentation with based on comprising the length of tentative watermark bag after testing, the factor of the amount distortion of existence and type etc. expects to comprise that the number of the segmentation of watermark bag compares makes this and determine.In certain embodiments, needn't process all segmentations.So, can for example, outer the postponing that jointly strides across one or more segmentation of the scope less than tentative watermark (, duration, spatial dimension etc.), at once carry out subsequent operation (that is, the operation at 1316 and 1318 places).In these embodiments, the operation at 1314 places can comprise about whether processed 1 definite to the predetermined number segmentation in the scope of k.In certain embodiments, k is the number that expectation comprises the segmentation of a watermark bag.In other embodiments, k is the number of the segmentation of watermark bag more than expecting to comprise.Treated segmentation therein strides across in the embodiment of the scope less than tentative watermark jointly, if the operation at 1318 places can not produce wanted result, can process and assess so additionally through extrapolation segmentation.
Back with reference to Figure 13 and hypothesis, carried out the operation at 1314 places, if 1314 places really definiteness show and must process extra segmentation (, at 1314 places, be "No"), so described group of operation 1300 turns back to 1308, and the result of the extrapolation based on previous segmentation obtains the new estimation of distortion information.For instance, at 1308 places, estimation that can refinement distortion space (for example,, based on produce the predistortion watermark template of best matching result at 1312 places).
If 1314 places really definiteness show processed all segmentations (being "Yes" at 1314 places), so described group of operation 1300 continues at 1316 places, wherein to each segmentation, assigns weight.In an example, according to the equation providing (9) and (10), assign weight in following chapters and sections.At 1318 places, existence and the value of the common assessment watermark based on one or more segmentation.For instance, the weight of assigning at 1316 places can be added together and itself and the weight threshold of detection that represents to have the watermark of specific reliable level are compared.It should be noted that and can carry out the described group of operation 1300 of describing in Figure 13 by different order order.For instance, can carry out immediately to described segmentation and assign weight simultaneously or after the watermark extrapolation of carrying out each segmentation.Be further noted that the program step that distortion as described above is followed the tracks of also increases false watermark detection probability.Therefore, must be carefully to guarantee that false conclusion watermark detection probability remains in predefine limit.
Figure 14 is the reduced graph of specific components that can be used for carrying out the some or all of operations in the operation of Figure 13 and/or Figure 18 according to disclosed embodiment.The assembly of showing in Figure 14 can reside at least in part can (for instance) to be play, record, copy, transmits or otherwise in the content disposal plant of access content.Content receiving assembly 1402 is configured to receive through revising and embeds content.Through revise to embed content can (for instance) for containing watermarked and having suffered to have a mind to or be not intended to the content of distortion.Reception is provided to multi-step searching component 1404 through revising embedding content with appropriate form, and multi-step searching component 1404 is configured to produce one or more tentative watermark and the segmentation of being extrapolated.Multi-step searching component 1404 can be utilized the stego-key of being selected assembly 1410 supplies by stego-key.Stego-key can be stored on stego-key memory module 1408.Watermark extrapolator assembly 1406 is configured to carry out watermark extrapolation for whole watermarked frame or on by segmentation basis.In certain embodiments, stencil-chosen assembly 1418 is configured to fully or receives/retrieve predistortion watermark template on by segmentation basis.For instance, stencil-chosen assembly 1418 can and make it can be used for watermark extrapolator assembly 1406 from database 1416 one or more predistortion watermark templates of retrieval.
Back, with reference to Figure 14, mistake minimizes criterion and determines that assembly 1412 is configured to receive the result being produced by watermark extrapolator assembly 1406.Mistake minimizes criterion and determines that assembly 1412 can further be configured to from user or allow the wrong criterion that minimizes to determine that definite larger system or the entity in device that will search for granularity of assembly 1412 receives extra input.Mistake minimizes criterion and determines that assembly 1412 is further configured to select assembly 1410, stencil-chosen assembly 1418 and template to rearrange assembly 1414 with stego-key and communicates by letter with (for instance) reception and registration new search granularity, upgrades stencil-chosen rule etc.Template rearranges assembly 1414 and is configured to rearrange predistortion watermark template based on minimized the information that criterion determines that assembly 1412 provides by mistake.Template rearranges assembly 1414 can and store predistortion watermark template into database 1416 from database 1416 retrieval predistortion watermark templates.Evaluator assembly 1420 can be carried out various watermark evaluations operation, for example to through extrapolation segmentation or watermark, assign weight, by through extrapolation watermark or its segmentation with predistortion template compares, determine whether common evaluation through extrapolation watermark or watermark segmentation meets wanted false watermark detection probability, report the amount distortion that is present in content host and type etc.In certain embodiments, evaluator assembly 1420 is incorporated in watermark extrapolator assembly 1406.
According to some embodiment, depend in can polluting and perhaps make the wrong essence of content distortion promote watermark extracting.In particular, contents processing and/or have a mind to content and attack and random or burst error can be injected in the watermark being embedded in content.For instance, burst error in the video image of being caught by Video Camera can be moved and be caused by unexpected Video Camera, and the burst error in institute's capturing audio can cause by (for instance) undesirable sound (for example people's s cough) in background.According to some embodiment, in the situation that there is burst error, if for watermark segmentation but not for the false watermark detection probability of whole watermark estimation, can strengthen so the performance of watermark extracting system.In an example embodiments, for realizing this enhancing in watermark extracting and reducing to process load, substitute for whole watermark self computing weight the weight to whole watermark computing cumulative indivedual segmentations.
How following instance embodiment graphic extension estimates the weight of indivedual watermark segmentations.Suppose that watermark segmentation shown by the string list of n position, and extractor finds that particular fragments contains e bit-errors.For instance, can be by the position that comprises segmentation be compared to the wrong e that ascertains the number with the template that represents the actual bit in described watermark segmentation.The weights W (n, e) that segmentation is associated therewith can be represented by following equation (9).
W ( n , e ) = - log 10 ( 2 - n &Sigma; i = 0 e n i ) - w u ( n ) - - - ( 9 ) .
In equation (9), w u(n) represent the expection weight that does not add watermark segmentation of length n, it can represent according to following equation (10).
w u ( n ) = - &Sigma; j = 0 n 2 - n n j log 10 ( 2 - n &Sigma; i = 0 j n i ) - - - ( 10 ) .
Figure 15 shows the section length n of the expection weight Sui Yiweiwei unit that does not add watermark segmentation producing based on equation (10) and the exemplary plot that becomes.For the utilization of the curve chart in graphic extension equation (9) and (10) and Figure 15, consider wherein 20 vicious 64 candidate's watermark PACKET in position.According to equation (9), the weight that bag is associated is therewith 2.3462.Yet if just cause 16 bit-errors in the first segmentation and 4 bit-errors in the second segmentation if watermark is divided into two 32 segmentations and error burst, two segmentations is 4.52 through weight accumulation so.Therefore,, by checking individually two segmentations and following combined result, can establish the existence of the watermark with higher confidence level.In certain embodiments, in a mode, select section length (or more generally, as the segmentation limit applicable to multidimensional watermark) to promote the detection of the watermark in (or optimization) different distortion environment.For instance, if embed content (or it suffers the distortion of particular type) by producing the communication channel transmitting of burst error, can select so the section length (for example, 4≤n≤10) in particular range.In other example, can use for example, some or all of section lengths in possibility section length (, 2≤n≤(watermarked frame length/2)) to assess potential watermark to select to produce the highest section length through weight accumulation by repetitive mode.
In many situations, a plurality of watermarks of the different pay(useful) loads of carrying are separately embedded in same content.For instance, simultaneously carrying of content is set for copying control, content recognition, content owner's identification, allocator and/or recipient's the watermark message of identification, thereby simultaneous operation with applications etc. is provided.In these cases, the different performance that different watermarks can have about pay(useful) load size, robustness, fail safe, false detection probability, processing load requires and other requirement.In these systems, can conventionally embed independently and extract each watermark message.
When a plurality of watermark message are embedded in content, can greatly strengthen by extracting the one or both in watermarked message with coordination mode the extraction of described watermark message.In particular, in certain embodiments, can reuse the extraction that the distortion estimation of carrying out for the first watermark message promotes to have the second watermark message of the performance requirement different from the first watermark message.Conventionally, for example, compare from the watermark message (content recognition watermark and/or for distinguishing court's watermark of the different copies of content) with larger pay(useful) load size, can use more efficiently the watermark message (for example copy is controlled watermark) with lower pay(useful) load size to estimate content distortion.For instance, the watermark message that has a less pay(useful) load of band can realize the use of stronger error correction capability.In addition, the watermark message with less area occupied can more not be subject to the distortion effect of time/spatial variations and can greater number embed, and improves whereby the robustness of watermark extracting.In addition, there is the use that little pay(useful) load can realize the more effective search algorithm of the template matches based on relative with error correcting/decoding process.Therefore, in certain embodiments, the watermark message based on having less pay(useful) load size is carried out distortion estimation, and is undertaken extracting and attempting for the target with the watermark message of larger pay(useful) load size by result.
According to some embodiment, when there is a plurality of watermark message in content, by the only extraction of the initial watermark with larger pay(useful) load (and lower robustness) and further improve watermarked extraction when the tentative watermark with less pay(useful) load being detected.In this way, can be owing to preventing not adding the realizing remarkable calculating without efficient search and save compared with large pay(useful) load watermark in watermark content.In addition, even, in markd content, still eliminate a plurality of search to larger pay(useful) load watermark in distortion space.Must carry out less search and also reduce the false watermark detection probability being associated with larger pay(useful) load watermark.The experiment of being undertaken by inventor discloses, in some situations, make the insignificant improvement to the detection of the lower pay(useful) load watermark message of the search of larger pay(useful) load watermark message and concluding associated generation watermark extracting efficiency and false positive rate, but can significantly reduce the robustness of larger pay(useful) load watermark message.That is, trigger continually not the extraction of larger pay(useful) load watermark message, and therefore waste the sufficient opportunity of the detection of larger pay(useful) load watermark.Therefore,, as early mentioned, the detection based on tentative less pay(useful) load watermark message triggers larger pay(useful) load watermark message and improves the efficiency of extraction operation and process load, and the reasonable robust detection result of larger pay(useful) load watermark message is provided simultaneously.
Should be understood that and can in the device by various hardware and/or software module and module composition, individually or jointly implement various embodiment of the present invention.Describe while disclosing embodiment, sometimes independent assembly is illustrated as and is configured to carry out one or more operation.However, it should be understood that can be by both or both above combine and/or each assembly can comprise the sub-component of not describing in this class component.In addition, with particular order order, present operation described in the various figure of the application's case to promote the understanding to basic concept.However, it should be understood that and can carry out this generic operation by different order order, and in addition, can carry out the various operations that disclose by extra or less step.
In some instances, device described in the application's case can comprise processor, memory cell, the interface connecting by correspondence each other, and can between from desktop PC and/or laptop computer for example, in the scope of consumer electronics device (media player, mobile device etc.).For instance, Figure 16 graphic extension can be implemented the block diagram of the device 1600 of various disclosed embodiment within it.Device 1600 comprises at least one processor 1602 and/or controller, at least one memory Unit 1604 of communicating by letter with processor 1602 and realizes at least one communication unit 1606 direct or and the data of other entity, device, database and network and the exchange of information indirect by communication link 1608.Communication unit 1606 can provide wired and/or wireless communication ability according to one or more communication protocol, and therefore, it can comprise appropriate emitter/receiver antenna, circuit and port and can and/or receive necessary coding/decoding ability to the appropriate transmitting of data and other information.Exemplary device 1600 depicted in figure 16 can be integrated into a part for content disposal plant, watermark embedding and/or extraction element.
In the general context that can be implemented on by computer program a method in embodiment or program step, describe various embodiment described herein, described computer program is for example embodied in, in the computer-readable media of the computer executable instructions (program code) that the computer that comprises in networked environment carries out.Computer-readable media can comprise self-mountable & dismountuble and non-removable storage device, including but not limited to read-only memory (ROM), random access memory (RAM), CD (CD), digital versatile disc (DVD), Blu-ray Disc etc.Therefore, the computer-readable media described in the application's case comprises nonvolatile medium.Conventionally, program module can comprise the routine carrying out particular task or implement particular abstract data type, program, object, assembly, data structure etc.Computer executable instructions, associated data structures and program module represent for carrying out the example of program code of the step of method disclosed herein.The particular sequence of this type of executable instruction or associated data structures represents for implementing the example of the respective action of the function described in this type of step or program step.According to the embedding of disclosed embodiment, there is the content of watermark to be stored in medium.In certain embodiments, comprise that one or more can not perception watermarked this can trigger through storage content that watermark extracting program step, associated signal are processed operation for example, during in content disposal plant (, software or the hardware media player) access by disposing watermark extracting device and by the application's case in the watermark extracting device that discloses and/or the subsequent operation of content disposal plant.
Object for diagram and explanation presents the above stated specification to embodiment.Above stated specification do not intend for exhaustive or embodiments of the invention are limited to disclosed precise forms, and can make or can obtain and revise and version from the practice of various embodiment according to above-mentioned teaching content.Select and the embodiment that discussed is herein described so as to explain the principle of various embodiment and practical application thereof and essence so that those skilled in the art can be in various embodiments and utilize the present invention in the situation that contain the various modifications that are suitable for special-purpose.Can in likely the combining of method, equipment, module, system and computer program, combine the feature of embodiment described herein.

Claims (51)

1. a method, it comprises:
Reception comprises the watermarked frame of sync section and packet portion, and described watermarked frame has the content of watermark to obtain from embedding;
Described received sync section and one or more predistortion synchronization template are compared; And
The watermarked frame that receives described at least one outcome evaluation based on described comparison.
2. method according to claim 1, wherein produces described one or more predistortion synchronization template by following operation at least in part:
Produce a plurality of navigation content;
Described a plurality of navigation content is embedded and have the watermark that comprises sync section;
By one or more distortion, make the distortion of described a plurality of embedding navigation content; And
From described a plurality of distortion navigation content, obtain candidate's predistortion synchronization template.
3. method according to claim 2, it further comprises:
The described a plurality of embedding navigation content that receive a subgroup after by the transmitting of transmitting media, wherein said transmitting media are incorporated into one or more distortion in described a plurality of embedding navigation content of described subgroup;
Each in the sync section of described a plurality of navigation content of described received subgroup and described candidate's predistortion synchronization template is compared; And
Be chosen in predefine tolerance limit one or more candidate's predistortion synchronization template mating with the described sync section of described a plurality of navigation content of described received subgroup.
4. method according to claim 1, the wherein said one or both relatively producing in the following:
The error count of each in described one or more predistortion synchronization template, it indicates the number of the not matching symbols between described the received sync section of described watermarked frame and each in described one or more predistortion synchronization template; And
The correlation of each in described one or more predistortion synchronization template, it indicates the described sync section of described watermarked frame to what extent to mate each in described one or more predistortion synchronization template.
5. method according to claim 1, it further comprises:
Described relatively after, the predistortion synchronization template of described received sync section of watermarked frame described in identification optimum Match; And
Select to represent to be present in the described distortion in described content with described one or more type of distortion and the amount distortion being associated through identification predistortion synchronization template.
6. method according to claim 1, it further comprises:
For the distortion of each type and/or the distortion order that combines to presort, organize described candidate's predistortion synchronization template, the wherein said order of presorting make to have represent real content distortion compared with first candidate's predistortion synchronization template rank of high likelihood higher than second candidate's predistortion synchronization template with the less possibility that represents real content distortion.
7. method according to claim 1, wherein said assessment produces there is the indication of the one in the following:
Tentative watermark, its expression has the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting,
Concluding watermark, it represents to have after testing the watermark that is less than or equal to the false watermark detection probability of being associated of wanted false watermark detection probability,
Be present in one or more type of distortion and amount distortion in described content.
8. method according to claim 1, wherein:
Described content comprises at least first and second watermark message;
Described the first watermark message has the pay(useful) load that is less than described the second watermark message;
Assessing described received watermarked frame causes described the first watermark message to detect as tentative watermark; And
Use the content distortion estimation obtaining for described tentative watermark to extract described the second watermark message.
9. method according to claim 1, it further comprises:
From content host, extract tentative watermark, described tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting;
By acquisition, be positioned described content inherence and through extracting the symbol of potential watermarked frame of the predefine position of tentative watermark, form one or more through extrapolation watermark with respect to described; And
Determine and through extrapolation watermark, whether meeting the false watermark detection probability of wanting with described after testing described in when tentative watermark is evaluated jointly.
10. method according to claim 9, the described symbol location of wherein said potential watermarked frame in described content inherence with respect to described through extracting time predefined and/or the place, locus of tentative watermark.
11. methods according to claim 9, wherein said definite comprising:
Determine the number of the mismark in described one or many person in extrapolation watermark;
Based on the described of mismark, through ascertaining the number, to described described one or many person in extrapolation watermark, assign weight at least in part; And
Described in determining when with described tentative watermark combination after testing, through weighting, through extrapolation watermark, whether meet described wanted false watermark detection probability.
12. methods according to claim 11, wherein by one or many person in the following, determine the often described number of the mismark in extrapolation watermark:
Described through extrapolation watermark symbol and described through extracting the comparison of the symbol of tentative watermark;
Carry out the described error-correcting code decoding through extrapolation watermark symbol; And
The described comparison through extrapolation watermark and one or more predistortion watermark template.
13. methods according to claim 9, wherein:
Form the described symbol that obtains described potential watermarked frame on extrapolation watermark is included in by segmentation basis, wherein each segmentation comprises both in more watermark symbol; And
Described definite comprising:
Obtain the number of the mismark in one or more segmentation;
The described number of the mismark based in each in described one or more segmentation is assigned weight to each in described one or more segmentation at least in part;
Combination is associated with described one or more segmentation described through assign weight with produce one or more through weighting through the watermark section of extrapolating; And
Described in determining when with described tentative watermark combination after testing, through weighting, through extrapolation watermark section, whether meet described wanted false watermark detection probability.
14. methods according to claim 13, the described number that wherein obtains the mismark in each in described one or more segmentation comprises:
Described symbol and one or more predistortion watermark template of the described potential watermarked frame in each in described one or more segmentation are compared; And
The counting of each described in producing in one or more segmentation, described counting represents the described symbol of described potential watermarked frame and the number of the not matching symbols between described one or more predistortion watermark template in described one or more segmentation.
15. methods according to claim 13, wherein one the or many person based in the following determines the number of described segmentation in extrapolation watermark:
Described through extracting the scope of tentative watermark;
Be present in the amount distortion in described content; And
Be present in the type of distortion in described content.
16. methods according to claim 13, wherein the expection weight that does not add watermark segmentation based on equal particular fragments in length is assigned the described weight of described particular fragments at least in part.
17. methods according to claim 9, wherein said wanted false watermark detection probability is corresponding to the detection of concluding watermark.
18. methods according to claim 9, it further comprises that report is present in distortion or the amount of distortion combination and the estimation of type in described embedding content host.
19. methods according to claim 13, wherein based on one or many person in the following, determine at least in part the scope of each segmentation:
Expection is present in the amount distortion in described content; And
Expection is present in the type of distortion in described content.
20. methods according to claim 1, it further comprises
From a plurality of tentative watermarks of described contents extraction, each tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting;
Use the described estimated distortion information that at least both acquisitions are associated with one or more distortion being present in described embedding content host in extracting tentative watermark;
Based on described estimated distortion information, obtain one or more predistortion watermark template; And
Use described one or more predistortion watermark template described at least one in extracting tentative watermark of reappraising.
21. methods according to claim 20, wherein said reappraising causes one or many person in the following:
Detect with the false watermark detection probability that is associated with described each in extracting tentative watermark and compare and have through improving the watermark of false watermark detection probability;
Detect and there is the concluding watermark that is less than or equal to the false watermark detection probability of being associated of described wanted false watermark detection probability; And
Be present in distortion in described embedding content host through improved estimator.
22. methods according to claim 21, described in wherein using, at least two obtain described estimated distortion information through extracting two or more sync sections of tentative watermark.
23. methods according to claim 1, it further comprises:
From the tentative watermark of described contents extraction, described tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting;
Use and describedly through extracting tentative watermark, obtain the estimated distortion information that indication is present in the coarse value of certain distortion in described embedding content host or distortion combination;
Selection is corresponding near one or more supplementary predistortion watermark template of one or more distortion value described coarse value; And
Use described selected one or more supplementary predistortion watermark template to reappraise described through extracting tentative watermark.
24. methods according to claim 23, wherein obtain described estimated distortion information and comprise:
A plurality of rough predistortion watermark template with coarse-grained selection corresponding to the scope of the distortion value of certain distortion or distortion combination;
Described each in extracting tentative watermark and described a plurality of rough predistortion watermark template is compared; And
Described in identification optimum Match, through extracting the first rough predistortion watermark template of tentative watermark, obtain described estimated distortion information.
25. methods according to claim 23, wherein said reappraise to cause detecting there is the concluding watermark that is less than or equal to the false watermark detection probability of being associated of wanted false watermark detection probability.
26. methods according to claim 23, wherein said reappraising comprises:
By described one or more supplementary predistortion watermark template and describedly compare through extracting tentative watermark;
Produce the error count be relatively associated with each, described error count represents that the supplementary predistortion watermark template of described correspondence to what extent mates described through extracting tentative watermark; And
Selection is corresponding to the described supplementary predistortion watermark template of the minimum error count that produces.
27. 1 kinds of devices, it comprises:
Receiver, it is configured to receive the watermarked frame that comprises sync section and packet portion, and described watermarked frame has the content of watermark to obtain from embedding;
Comparator, it is configured to described received sync section and one or more predistortion synchronization template to compare; And
Evaluator, it is configured to the watermarked frame that receives described at least one outcome evaluation based on described comparator.
28. devices according to claim 27, wherein said comparator is configured to produce the one or both in the following:
The error count of each in described one or more predistortion synchronization template, it indicates the number of the not matching symbols between described the received sync section of described watermarked frame and each in described one or more predistortion synchronization template; And
The correlation of each in described one or more predistortion synchronization template, it indicates the described sync section of described watermarked frame to what extent to mate each in described one or more predistortion synchronization template.
29. devices according to claim 27, wherein said comparator is configured to:
The predistortion synchronization template of described the received sync section of watermarked frame described in identification optimum Match; And
Select to represent to be present in the described distortion in described content with described one or more type of distortion and the amount distortion being associated through identification predistortion synchronization template.
30. devices according to claim 27, wherein said evaluator is configured to produce the indication to the existence of one or many person in the following:
Tentative watermark, its expression has the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting,
Concluding watermark, it represents to have after testing the watermark that is less than or equal to the false watermark detection probability of being associated of wanted false watermark detection probability, and
Be present in one or more type of distortion and amount distortion in described content.
31. devices according to claim 27, wherein:
Described content comprises at least first and second watermark message;
Described the first watermark message has the pay(useful) load that is less than described the second watermark message; And
It is the indication of tentative watermark and content distortion estimation that described evaluator is configured to produce to described the first watermark message is detected, and wherein with described content distortion estimation, extracts described the second watermark message.
32. devices according to claim 27, it further comprises
Extractor, it is configured to from the tentative watermark of described contents extraction, and described tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting; And
Watermark extrapolator, it is configured to be positioned described content inherence by acquisition and through extracting the symbol of potential watermarked frame of the predefine position of tentative watermark, forms one or more through extrapolation watermark with respect to described; Wherein
Described evaluator is further configured to determine through extrapolation watermark, whether to meet the false watermark detection probability of wanting with described after testing described in when tentative watermark is evaluated jointly.
33. devices according to claim 32, wherein said watermark extrapolator is configured to from described content inherence with respect to the described described symbol that obtains described potential watermarked frame through extracting the time predefined of tentative watermark and/or the position of locus.
34. devices according to claim 32, wherein said evaluator is configured to:
Determine the number of the mismark in described one or many person in extrapolation watermark;
Based on the described of mismark, through ascertaining the number, to described described one or many person in extrapolation watermark, assign weight at least in part; And
Described in determining when with described tentative watermark combination after testing, through weighting, through extrapolation watermark, whether meet described wanted false watermark detection probability.
35. devices according to claim 34, wherein by one or many person in the following, determine the often described number of the mismark in extrapolation watermark:
Described through extrapolation watermark symbol and described through extracting the comparison of the symbol of tentative watermark;
Carry out the described error-correcting code decoding through extrapolation watermark symbol; And
The described comparison through extrapolation watermark and one or more predistortion watermark template.
36. devices according to claim 32, wherein:
Described watermark extrapolator is configured to form describedly through extrapolation watermark by obtaining at least in part the symbol of described potential watermarked frame on by segmentation basis, and wherein each segmentation comprises both in more watermark symbol; And
Described evaluator is configured to:
Obtain the number of the mismark in one or more segmentation;
The described number of the mismark based in each in described one or more segmentation is assigned weight to each in described one or more segmentation at least in part;
Combination is associated with described one or more segmentation described through assign weight with produce one or more through weighting through the watermark section of extrapolating; And
Described in determining when with described tentative watermark combination after testing, through weighting, through extrapolation watermark section, whether meet described wanted false watermark detection probability.
37. devices according to claim 36, wherein said evaluator is configured to by following operation, obtain at least in part the described number of the mismark in each in described one or more segmentation:
Described symbol and one or more predistortion watermark template of the described potential watermarked frame in each in described one or more segmentation are compared; And
The counting of each described in producing in one or more segmentation, described counting represents the described symbol of described potential watermarked frame and the number of the not matching symbols between described one or more predistortion watermark template in described one or more segmentation.
38. devices according to claim 32, wherein said evaluator is configured to the expection weight that does not add watermark segmentation based on equal particular fragments in length at least in part and assigns described weight to described particular fragments.
39. devices according to claim 32, wherein said evaluator is configured to report and is present in distortion or the amount of distortion combination and the estimation of type in described content.
40. devices according to claim 27, it further comprises
Extractor, it is configured to from a plurality of tentative watermarks of described contents extraction, and each tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting; And
Distortion estimator, its be configured to use described in extracting tentative watermark at least both obtain the estimated distortion information being associated with one or more distortion being present in embedding content host and obtain one or more predistortion watermark template; Wherein
Evaluator is configured to use described one or more predistortion watermark template described at least one in extracting tentative watermark of reappraising.
41. according to the device described in claim 40, wherein said distortion estimator is configured to determine that whether described estimated distortion information is corresponding to the amount distortion that surpasses certain distortion threshold value, and only in the situation that described estimated distortion information obtains described one or more predistortion watermark template corresponding to the amount distortion that is no more than described certain distortion threshold value.
42. according to the device described in claim 40, and described in wherein said distortion estimator is configured to use, at least two obtain described estimated distortion information through extracting two or more sync sections of tentative watermark.
43. devices according to claim 27, it further comprises
Extractor, it is configured to from the tentative watermark of described contents extraction, and described tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting;
Distortion estimator, it is configured to use and describedly through extracting tentative watermark, obtains the estimated distortion information that indication is present in the coarse value of certain distortion in described content or distortion combination, and selects one or more the supplementary predistortion watermark template corresponding near one or more distortion value described coarse value; Wherein
Described evaluator is further configured to use described selected one or more supplementary predistortion watermark template to reappraise described through extracting tentative watermark.
44. according to the device described in claim 43, and wherein said distortion estimator is configured to:
A plurality of rough predistortion watermark template with coarse-grained selection corresponding to the scope of the distortion value of certain distortion or distortion combination;
Described each in extracting tentative watermark and described a plurality of rough predistortion watermark template is compared; And
Described in identification optimum Match, through extracting the first rough predistortion watermark template of tentative watermark, obtain described estimated distortion information.
45. according to the device described in claim 43, and wherein said evaluator is configured to:
By described one or more supplementary predistortion watermark template and describedly compare through extracting tentative watermark;
Produce the error count be relatively associated with each, described error count represents that the supplementary predistortion watermark template of described correspondence to what extent mates described through extracting tentative watermark; And
Selection is corresponding to the described supplementary predistortion watermark template of the minimum error count that produces.
46. 1 kinds of computer programs that are embodied on nonvolatile computer-readable media, it comprises:
For receiving the program code of the watermarked frame that comprises sync section and packet portion, described watermarked frame has the content of watermark to obtain from embedding;
For the program code that described received sync section and one or more predistortion synchronization template are compared; And
Program code for the watermarked frame that receives described at least one outcome evaluation based on described comparison.
47. according to the computer program described in claim 46, and it further comprises:
For the program code from the tentative watermark of described contents extraction, described tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting;
For be positioned described content inherence by acquisitions, with respect to described, through extracting the symbol of potential watermarked frame of the predefine position of tentative watermark, form one or more through the program code of the watermark of extrapolating; And
For determining, through extrapolation watermark, whether meeting the program code of the false watermark detection probability of wanting with described after testing described in when tentative watermark is evaluated jointly.
48. according to the computer program described in claim 46, and it further comprises:
For the program code from a plurality of tentative watermarks of described contents extraction, each tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting;
For use described through extract tentative watermark at least both obtain the program code of the estimated distortion information being associated with one or more distortion being present in embedding content host;
For obtain the program code of one or more predistortion watermark template based on described estimated distortion information; And
Described through extracting at least one program code of tentative watermark for using described one or more predistortion watermark template to reappraise.
49. according to the computer program described in claim 46, and it further comprises:
For the program code from the tentative watermark of described contents extraction, described tentative watermark represents to have the candidate's watermark that surpasses the false watermark detection probability of being associated of false watermark detection probability of wanting;
For using the described program code that is present in the estimated distortion information of the certain distortion of described embedding content host or the coarse value of distortion combination through extracting tentative watermark acquisition indication;
For selecting the program code corresponding near one or more supplementary predistortion watermark template of one or more distortion value described coarse value; And
For using described selected one or more supplementary predistortion watermark template to reappraise described through extracting the program code of tentative watermark.
50. 1 kinds of devices, it comprises:
Water mark embedding device, it is configured so that a plurality of navigation content embeds the watermark that comprises sync section;
Distortion processing components, it is configured to make the distortion of described a plurality of embedding navigation content by one or more distortion; And
Watermark extracting device, it is configured to produce candidate's predistortion synchronization template from described a plurality of distortion navigation content.
51. according to the device described in claim 50, and it further comprises:
Receiver, it is configured to receive described a plurality of embedding navigation content of a subgroup after by the transmitting of transmitting media, and wherein said transmitting media are incorporated into one or more distortion in described a plurality of embedding navigation content of described subgroup; And
Template matches assembly, it is configured to each in the sync section of described a plurality of navigation content of described received subgroup and described candidate's predistortion synchronization template to compare, and is chosen in predefine tolerance limit one or more candidate's predistortion synchronization template mating with the described sync section of described a plurality of navigation content of described received subgroup.
CN201280059704.5A 2011-11-03 2012-11-02 Watermark extracting based on tentative watermark Active CN103975605B (en)

Applications Claiming Priority (9)

Application Number Priority Date Filing Date Title
US13/288,958 US8533481B2 (en) 2011-11-03 2011-11-03 Extraction of embedded watermarks from a host content based on extrapolation techniques
US13/288,945 2011-11-03
US13/288,952 US8682026B2 (en) 2011-11-03 2011-11-03 Efficient extraction of embedded watermarks in the presence of host content distortions
US13/288,952 2011-11-03
US13/288,937 2011-11-03
US13/288,937 US8615104B2 (en) 2011-11-03 2011-11-03 Watermark extraction based on tentative watermarks
US13/288,945 US8923548B2 (en) 2011-11-03 2011-11-03 Extraction of embedded watermarks from a host content using a plurality of tentative watermarks
US13/288,958 2011-11-03
PCT/US2012/063431 WO2013067439A1 (en) 2011-11-03 2012-11-02 Watermark extraction based on tentative watermarks

Publications (2)

Publication Number Publication Date
CN103975605A true CN103975605A (en) 2014-08-06
CN103975605B CN103975605B (en) 2018-09-07

Family

ID=48192861

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201280059704.5A Active CN103975605B (en) 2011-11-03 2012-11-02 Watermark extracting based on tentative watermark

Country Status (5)

Country Link
EP (1) EP2774385A4 (en)
KR (1) KR20140097306A (en)
CN (1) CN103975605B (en)
HK (1) HK1197126A1 (en)
WO (1) WO2013067439A1 (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106796625A (en) * 2014-08-20 2017-05-31 凡瑞斯公司 Use the multifarious watermark detection of prediction pattern
CN109151483A (en) * 2018-09-10 2019-01-04 五邑大学 System based on image and video metadata reversible insertion and extraction
US10499120B2 (en) 2014-03-13 2019-12-03 Verance Corporation Interactive content acquisition using embedded codes
CN110827189A (en) * 2019-11-01 2020-02-21 山东浪潮人工智能研究院有限公司 Method and system for removing watermark of digital image or video
US10757537B2 (en) 2016-08-31 2020-08-25 Alibaba Group Holding Limited User positioning method, information push method, and related apparatus

Families Citing this family (28)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7644282B2 (en) 1998-05-28 2010-01-05 Verance Corporation Pre-processed information embedding system
US6737957B1 (en) 2000-02-16 2004-05-18 Verance Corporation Remote control signaling using audio watermarks
EP1552454B1 (en) 2002-10-15 2014-07-23 Verance Corporation Media monitoring, management and information system
US20060239501A1 (en) 2005-04-26 2006-10-26 Verance Corporation Security enhancements of digital watermarks for multi-media content
US8020004B2 (en) 2005-07-01 2011-09-13 Verance Corporation Forensic marking using a common customization function
US8781967B2 (en) 2005-07-07 2014-07-15 Verance Corporation Watermarking in an encrypted domain
US8259938B2 (en) 2008-06-24 2012-09-04 Verance Corporation Efficient and secure forensic marking in compressed
US10664940B2 (en) 2010-03-05 2020-05-26 Digimarc Corporation Signal encoding to reduce perceptibility of changes over time
US8477990B2 (en) 2010-03-05 2013-07-02 Digimarc Corporation Reducing watermark perceptibility and extending detection distortion tolerances
US8971567B2 (en) 2010-03-05 2015-03-03 Digimarc Corporation Reducing watermark perceptibility and extending detection distortion tolerances
US9607131B2 (en) 2010-09-16 2017-03-28 Verance Corporation Secure and efficient content screening in a networked environment
US8682026B2 (en) 2011-11-03 2014-03-25 Verance Corporation Efficient extraction of embedded watermarks in the presence of host content distortions
US8923548B2 (en) 2011-11-03 2014-12-30 Verance Corporation Extraction of embedded watermarks from a host content using a plurality of tentative watermarks
US8615104B2 (en) 2011-11-03 2013-12-24 Verance Corporation Watermark extraction based on tentative watermarks
US8533481B2 (en) 2011-11-03 2013-09-10 Verance Corporation Extraction of embedded watermarks from a host content based on extrapolation techniques
US8745403B2 (en) 2011-11-23 2014-06-03 Verance Corporation Enhanced content management based on watermark extraction records
US9323902B2 (en) 2011-12-13 2016-04-26 Verance Corporation Conditional access using embedded watermarks
US9547753B2 (en) 2011-12-13 2017-01-17 Verance Corporation Coordinated watermarking
US9571606B2 (en) 2012-08-31 2017-02-14 Verance Corporation Social media viewing system
US8869222B2 (en) 2012-09-13 2014-10-21 Verance Corporation Second screen content
US9106964B2 (en) 2012-09-13 2015-08-11 Verance Corporation Enhanced content distribution using advertisements
US9262793B2 (en) 2013-03-14 2016-02-16 Verance Corporation Transactional video marking system
US9251549B2 (en) 2013-07-23 2016-02-02 Verance Corporation Watermark extractor enhancements based on payload ranking
KR102113503B1 (en) 2013-08-26 2020-06-05 삼성전자주식회사 Electronic apparatus and method for providing contents in the electronic apparatus
US9208334B2 (en) 2013-10-25 2015-12-08 Verance Corporation Content management using multiple abstraction layers
US10504200B2 (en) 2014-03-13 2019-12-10 Verance Corporation Metadata acquisition using embedded watermarks
EP3225034A4 (en) 2014-11-25 2018-05-02 Verance Corporation Enhanced metadata and content delivery using watermarks
US9602891B2 (en) 2014-12-18 2017-03-21 Verance Corporation Service signaling recovery for multimedia content using embedded watermarks

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030174862A1 (en) * 2000-02-14 2003-09-18 Rhoads Geoffrey B. Digital watermark screening and detection strategies
US6757407B2 (en) * 1998-05-12 2004-06-29 Lucent Technologies Inc. Transform domain image watermarking method and system
US20060104477A1 (en) * 2004-11-12 2006-05-18 Kabushiki Kaisha Toshiba Digital watermark detection apparatus and digital watermark detection method
CN101751927A (en) * 2008-10-10 2010-06-23 汤姆森许可贸易公司 Method and apparatus for regaining watermark data in an original signal
WO2010135687A1 (en) * 2009-05-21 2010-11-25 Digimarc Corporation Combined watermarking and fingerprinting

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7366908B2 (en) * 1996-08-30 2008-04-29 Digimarc Corporation Digital watermarking with content dependent keys and autocorrelation properties for synchronization
US6483927B2 (en) * 2000-12-18 2002-11-19 Digimarc Corporation Synchronizing readers of hidden auxiliary data in quantization-based data hiding schemes
US7840005B2 (en) * 2002-01-22 2010-11-23 Digimarc Corporation Synchronization of media signals
US20060239501A1 (en) * 2005-04-26 2006-10-26 Verance Corporation Security enhancements of digital watermarks for multi-media content
US7668334B2 (en) * 2004-07-02 2010-02-23 Digimarc Corp Conditioning imagery to better receive steganographic encoding
WO2009031082A1 (en) * 2007-09-03 2009-03-12 Koninklijke Philips Electronics N.V. Apparatus and methods for transferring digital content

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6757407B2 (en) * 1998-05-12 2004-06-29 Lucent Technologies Inc. Transform domain image watermarking method and system
US20030174862A1 (en) * 2000-02-14 2003-09-18 Rhoads Geoffrey B. Digital watermark screening and detection strategies
US20060104477A1 (en) * 2004-11-12 2006-05-18 Kabushiki Kaisha Toshiba Digital watermark detection apparatus and digital watermark detection method
CN101751927A (en) * 2008-10-10 2010-06-23 汤姆森许可贸易公司 Method and apparatus for regaining watermark data in an original signal
WO2010135687A1 (en) * 2009-05-21 2010-11-25 Digimarc Corporation Combined watermarking and fingerprinting

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10499120B2 (en) 2014-03-13 2019-12-03 Verance Corporation Interactive content acquisition using embedded codes
CN106796625A (en) * 2014-08-20 2017-05-31 凡瑞斯公司 Use the multifarious watermark detection of prediction pattern
CN106796625B (en) * 2014-08-20 2019-09-24 凡瑞斯公司 Use the multifarious watermark detection of prediction pattern
US10445848B2 (en) 2014-08-20 2019-10-15 Verance Corporation Content management based on dither-like watermark embedding
US10757537B2 (en) 2016-08-31 2020-08-25 Alibaba Group Holding Limited User positioning method, information push method, and related apparatus
CN109151483A (en) * 2018-09-10 2019-01-04 五邑大学 System based on image and video metadata reversible insertion and extraction
CN110827189A (en) * 2019-11-01 2020-02-21 山东浪潮人工智能研究院有限公司 Method and system for removing watermark of digital image or video
CN110827189B (en) * 2019-11-01 2023-12-05 山东浪潮科学研究院有限公司 Watermark removing method and system for digital image or video

Also Published As

Publication number Publication date
WO2013067439A1 (en) 2013-05-10
HK1197126A1 (en) 2015-01-02
KR20140097306A (en) 2014-08-06
EP2774385A1 (en) 2014-09-10
CN103975605B (en) 2018-09-07
EP2774385A4 (en) 2015-05-27

Similar Documents

Publication Publication Date Title
CN103975605A (en) Watermark extraction based on tentative watermarks
US8615104B2 (en) Watermark extraction based on tentative watermarks
US8533481B2 (en) Extraction of embedded watermarks from a host content based on extrapolation techniques
US8682026B2 (en) Efficient extraction of embedded watermarks in the presence of host content distortions
US8923548B2 (en) Extraction of embedded watermarks from a host content using a plurality of tentative watermarks
US9704211B2 (en) Signal continuity assessment using embedded watermarks
EP1362485B1 (en) Generating and matching hashes of multimedia content
US6252972B1 (en) Method and arrangement for detecting a watermark using statistical characteristics of the information signal in which the watermark is embedded
KR100614425B1 (en) Watermark detection
US8112818B2 (en) System for identifying content of digital data
CN1988669B (en) Digital marking structure and verifying method in stream medium monitoring and broadcasting
CN1957396A (en) Device and method for analyzing an information signal
CN102461066A (en) Combined watermarking and fingerprinting
EP1561176A2 (en) Improved audio data fingerprint searching
US20070220265A1 (en) Searching for a scaling factor for watermark detection
Qiao et al. Adaptive steganalysis based on statistical model of quantized DCT coefficients for JPEG images
Varna et al. A decision theoretic framework for analyzing binary hash-based content identification systems
CN100405840C (en) Encoding and detecting apparatus
US7085398B2 (en) Method and system to decode image watermarks
Sharma et al. Watermark synchronization: Perspectives and a new paradigm
CN101065769B (en) Method and device for reading digital tattooing
US20070160260A1 (en) Method of scale factor retrieval
Lim et al. Content-based Authentication using Hash Function in Digital Images

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
REG Reference to a national code

Ref country code: HK

Ref legal event code: DE

Ref document number: 1197126

Country of ref document: HK

GR01 Patent grant
GR01 Patent grant
REG Reference to a national code

Ref country code: HK

Ref legal event code: WD

Ref document number: 1197126

Country of ref document: HK