CN107908647A - A kind of scalable video search method based on digital watermarking - Google Patents
A kind of scalable video search method based on digital watermarking Download PDFInfo
- Publication number
- CN107908647A CN107908647A CN201710937101.5A CN201710937101A CN107908647A CN 107908647 A CN107908647 A CN 107908647A CN 201710937101 A CN201710937101 A CN 201710937101A CN 107908647 A CN107908647 A CN 107908647A
- Authority
- CN
- China
- Prior art keywords
- video
- code
- retrieval
- watermark
- index
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/71—Indexing; Data structures therefor; Storage structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/73—Querying
- G06F16/732—Query formulation
- G06F16/7335—Graphical querying, e.g. query-by-region, query-by-sketch, query-by-trajectory, GUIs for designating a person/face/object as a query predicate
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Human Computer Interaction (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Image Processing (AREA)
Abstract
A kind of scalable video search method based on digital watermarking, the search method comprise the following steps：Watermark index code is generated by application watermark generator, while for constructing index data base；Retrieval code is serially formed using the characteristic information applied in information retrieval based on contents as aspect indexing code, both watermark index code and aspect indexing code；Using strong robust digital watermark algorithm by the sightless embedded video of watermark index code；The retrieval code and index data base of foundation design, propose the retrieval system of grading search：That is, the video Fast Classification that watermark index code is used to be retrieved improves retrieval rate to retrieval classification；Aspect indexing code, is from the feature for the video extraction that is retrieved, using characteristic matching principle, is matched in classification, precise search to purpose video.The present invention realizes classification retrieval by window using the digital watermark technology and Feature Extraction Technology of strong robust, reduces the complexity of calculating, reaches strong robust and quick-searching purpose.
Description
Technical field
The present invention relates to multimedia information retrieval field, more particularly to a kind of scalable video retrieval side based on digital watermarking
Method.
Background technology
People have been enter into a fast-developing informationized society, with the development of multimedia technology and network technology, people
To create and propagate media product more convenient, thus generate a large amount of multimedia messages.These information are probably picture, sound
Sound or video etc..Particularly number of videos increases especially rapid, because video data is vivid, can carry a large amount of letters
Breath, it is easier to be humanly perceptible, video, which follows a well mapped-out plan, is used widely.How that carries out video data effectively managing and examining
Rope has become the hot spot of current multimedia technology research.
Existing video retrieval technology can be divided into two classes, as follows：
1st, text based video retrieval method：
By carrying out manual analysis to video data, label character is then carried out to video and establishes keyword, so that complete
Text into video describes.When being inquired about, only video matching need to be carried out using text matches mode.Due to the text of video
Content and label character are completed by manual analysis, and processing procedure is cumbersome and efficiency is low；Meanwhile text description carries subjectivity,
Video content cannot be fully described.
2nd, content based video retrieval system：
Machine is carried out to video content to automatically analyze, including：Analysis, video lens segmentation and the key frame of video structure carry
Taking, the characteristic information extracted according to video content establishes index, and is matched by the similarity between characteristic information, from
And realize the retrieval of video.But this method is computationally intensive, retrieval rate is slow.
With the development of digital watermark technology, the method for occurring for digital watermarking being used for retrieving.Wherein numeral regards
Frequency can add watermark as video index in the production phase, by retrieval and inquisition information and the watermark extracted from video progress
Match somebody with somebody, obtain corresponding retrieval result, reach the function of quick-searching.But its anti-attack ability is limited, after under attack, retrieval effect
Fruit can substantially reduce, its robustness is poor.
The content of the invention
The present invention provides a kind of scalable video search method based on digital watermarking, the present invention utilizes the numeral of strong robust
Digital watermark and Feature Extraction Technology realize classification retrieval by window, reduce the complexity of calculating, reach strong robust and quick inspection
Rope purpose, it is described below：
A kind of scalable video search method based on digital watermarking, the search method comprise the following steps：
Watermark index code is generated by application watermark generator, while for constructing index data base；Apply based on
Characteristic information in content retrieval serially forms retrieval code as aspect indexing code, both watermark index code and aspect indexing code；
Using strong robust digital watermark algorithm by the sightless embedded video of watermark index code；
The retrieval code and index data base of foundation design, propose the retrieval system of grading search：
That is, the video Fast Classification that watermark index code is used to be retrieved improves retrieval rate to retrieval classification；
Aspect indexing code, is from the feature for the video extraction that is retrieved, using characteristic matching principle, is carried out in classification
Match somebody with somebody, precise search to purpose video.
The beneficial effect of technical solution provided by the invention is：
1st, the weak robustness of video frequency searching is avoided, a variety of strong robust video watermarking algorithms can be applied, i.e., according to application
The requirement of scene and application purpose etc., selects suitable strong robust video watermarking algorithms, strengthens the adaptability of searching system.
2nd, the mentality of designing retrieved using the scalable video based on digital watermarking, is constructed by watermark index and aspect indexing
The retrieval code of composition.The primary positioning of watermark index and the secondary positioning of aspect indexing, substantially increase retrieval rate.
3rd, the cumbersome low with efficiency of video processing procedure is avoided, with reference in watermark generator design database and retrieval code
Watermark index part, reduces computation complexity, improves treatment effeciency.
Brief description of the drawings
Fig. 1 is index data base construction flow chart；
Fig. 2 is the concrete structure diagram of retrieval code；
Fig. 3 is a kind of flow chart of the scalable video search method based on digital watermarking；
Fig. 4 is 14 test video sequence samples.
Embodiment
To make the object, technical solutions and advantages of the present invention clearer, embodiment of the present invention is made below further
It is described in detail on ground.
Embodiment 1
A kind of scalable video search method based on digital watermarking, referring to Fig. 1-Fig. 3, which includes following step
Suddenly：
101：In retrieval structurally, watermark index code is generated by application watermark generator, while is also used for construction index
Database；
102：Using the characteristic information applied in information retrieval based on contents as aspect indexing code, watermark index code and feature rope
Draw both codes and serially form retrieval code；
103：It is using strong robust digital watermark algorithm (known to one of skill in the art) that watermark index code is sightless
In embedded video；
104：The retrieval code and index data base of foundation design, propose the retrieval system of grading search, watermark index code energy
By the video Fast Classification that need to be retrieved to retrieval classification, retrieval rate is improved；Aspect indexing code, is from the video extraction that is retrieved
Feature, using characteristic matching principle, matched in classification, precise search to purpose video.
In conclusion the embodiment of the present invention realizes the digital watermarking using strong robust by above-mentioned steps 201- steps 204
Technology and Feature Extraction Technology realize classification retrieval by window, reduce the complexity of calculating, reach strong robust and quick-searching mesh
's.
Embodiment 2
The scheme in embodiment 1 is further introduced with reference to specific example, Fig. 1-Fig. 3, it is as detailed below
Description：
201：Construct index data base；
Wherein, the construction part of database is as shown in Figure 1.Using watermark generator, the purpose sum number of video frequency searching is considered
Situations such as according to storehouse number of videos, content, according to reflection video properties (such as keyword of video production business or representing video content
Deng) information, generate respective watermark index.
Video set can be classified according to different directions according to watermark index and make index, foundation is similar to document
Index bibliographic data storehouse.Simple example, represents the video file from Hunan Satellite TV, " 002 " represents and comes from Zhejiang with " 001 "
The video file of satellite TV；" 003 " represents the video file from Jiangsu satellite TV, and so on.
Index data base only need to be divided all videos roughly, and process makes every effort to succinct, be not required to expend the excessive time.
202：Construct video frequency searching code；
Wherein, video frequency searching code is serially made of watermark index and aspect indexing two parts, can have fixed capacity per part,
8bit information is accounted for as watermark indexes, aspect indexing occupies 32bit information.The specific configuration of retrieval code is as shown in Figure 2.
Wherein, watermark index is that video passes through obtained by watermark generator, the Classification And Index data of video generic
The structure in storehouse is related.Watermark generator is flexible and changeable, not uniquely, is determined according to the application scenario of video frequency searching and purpose etc.
It is fixed.The sorting technique of video in existing database can be used.
Wherein, aspect indexing is made of the video feature vector that video feature extraction algorithm extracts.Video feature extraction is calculated
The research of method also comparative maturity, feature extraction algorithm is not also unique herein, as long as video properties can be represented more comprehensively, simple easily behaviour
Make, all can be applicable in this step.The hierarchy of video frequency searching code determines that the retrieval mode of video, and quick-searching regard
The major reason of frequency.
203：Video watermark retrieval code is hidden in inside carrier video using strong robust digital watermark and obtains embedded video,
As shown in the watermark retrieval code embedded part of Fig. 1；
This step is still flexibly to use robust digital watermark algorithm, as long as can meet robustness requirement, can resist rotation, pressure
The noise that contracting and A/D, D/A conversion are introduced so that searching system has strong robust.It is not limited to use any watermarking algorithm.
The embodiment of the present invention realizes the insertion, attack and extraction test of watermark using following steps herein.
1) embedded part is to be embedded in using constraint template into row information, realizes that information carries.The realization pair of frequency constraint template
The constraint of frequency, can be that frequency suppresses be frequency enhancing, as long as can shape frequency, then can realize that information carries.
Formula is as follows：
T (u, v)=P (u, v) * F (u, v)
Wherein, (u, v) represents frequency domain coordinates, and T represents output frequency video, and P represents frequency constraint template, and F represents input
Frequency video.
2) attack part is the noise of the generations such as the conversion for introducing rotation, compression, light and A/D, D/A, specifically, design
The noise of camera collection screen video introduces environment.Output frequency video T is played out in liquid crystal display, passes through liquid crystal
The shooting of display screen obtains T1, but causes active area to account for the decline of whole frame aspect ratio, so using multi-media processing software pair
The display content of rear video T1 extraction liquid crystal displays is gathered, that is, positions picture-in-picture, obtains video T2.
3) extraction part is corresponding with embedded part, frequency is analyzed, analyze whether frequency carries out about
Beam shaping, proposes the energy statistics histogram analysis method of frequency domain.
Specifically, energy statistics histogram is obtained to the frequency domain figure picture calculating histogram of video T2, by energy statistics Nogata
Figure obtains classification and Detection device as being trained in characteristic information input support vector machines, using grader to embedded video into
Row detection, draws video watermark.
204：The grading search system of scalable video retrieval.
As shown in figure 3, retrieval code is made of two parts, so a point two-stage is retrieved.First order retrieval is to utilize step
The watermark index code of 203 extractions carries out the Similarity matching of text message with index data base, navigates to video generic, but class
Still there are many different videos under other, it is impossible to be pin-pointed to purpose video, belong to Primary search；Second level retrieval is will to utilize spy
The feature of video extraction carries out characteristic matching in the aspect indexing and generic of sign extraction algorithm extraction, takes most like video
As retrieval result, belong to secondary search.
Compared with existing content-based retrieval, watermark indexes the information for often increasing by 1 bit, can reduce one in principle
Half retrieval time.
In conclusion the embodiment of the present invention realizes the digital watermarking using strong robust by above-mentioned steps 201- steps 204
Technology and Feature Extraction Technology realize classification retrieval by window, reduce the complexity of calculating, reach strong robust and quick-searching mesh
's.
Embodiment 3
Feasibility verification is carried out to the scheme in Examples 1 and 2 with reference to specific experimental data, Fig. 4, it is as detailed below
Description；
Totally 14 test video sequences in experiment, as shown in Figure 4.All videos of this experiment all only transmit 000010011111
Watermark index, i.e. classified index in form 1.This method is according to bit-detection accuracy rate and precision ratio come measure algorithm
Can, it is defined as follows：
Wherein, Accuracy is bit-detection accuracy rate, and C represents the bit number correctly transmitted per frame, and N represents the biography per frame
Defeated bit number, frames represent all frames.
Wherein, Precision is precision ratio, NzIt is the quantity of correct retrieval object, NallIt is the number of all retrieval objects
Amount.Simple experiment result is as shown in the table.
Table 1
Feasibility of this method in terms of video frequency searching can be seen that by above-mentioned experimental data.In such complicated retrieval
Remain to reach in environment 78% or so precision ratio this absolutely prove the strong robustness of this method.
The 1bit information of database can help 2 times of retrieval rates of conventional retrieval system improving, m bit information lifting 2mTimes
Speed, so this is quick and robust a video frequency searching design.
It will be appreciated by those skilled in the art that attached drawing is the schematic diagram of a preferred embodiment, the embodiments of the present invention
Sequence number is for illustration only, does not represent the quality of embodiment.
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all the present invention spirit and
Within principle, any modification, equivalent replacement, improvement and so on, should all be included in the protection scope of the present invention.
Claims (2)
1. a kind of scalable video search method based on digital watermarking, it is characterised in that the search method comprises the following steps：
Watermark index code is generated by application watermark generator, while for constructing index data base；Applying based on content
Characteristic information in retrieval serially forms retrieval code as aspect indexing code, both watermark index code and aspect indexing code；
Using strong robust digital watermark algorithm by the sightless embedded video of watermark index code；
The retrieval code and index data base of foundation design, propose the retrieval system of grading search：
That is, the video Fast Classification that watermark index code is used to be retrieved improves retrieval rate to retrieval classification；
Aspect indexing code, is from the feature for the video extraction that is retrieved, using characteristic matching principle, is matched in classification, essence
Really retrieve purpose video.
A kind of 2. scalable video search method based on digital watermarking according to claim 1, it is characterised in that the profit
To be specially in the sightless embedded video of watermark index code with strong robust digital watermark algorithm：
1) it is embedded in using constraint template into row information, realizes that information carries；
T (u, v)=P (u, v) * F (u, v)
Wherein, (u, v) represents frequency domain coordinates, and T represents output frequency video, and P represents frequency constraint template, and F represents incoming frequency
Video；
2) display content using multi-media processing software to collection rear video extraction liquid crystal display, that is, position picture-in-picture, obtain
To video T2；
3) energy statistics histogram is obtained to the frequency domain figure picture calculating histogram of video T2, using energy statistics histogram as feature
It is trained in information input support vector machines, obtains classification and Detection device, embedded video is detected using grader, is drawn
Video watermark.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710937101.5A CN107908647A (en) | 2017-10-10 | 2017-10-10 | A kind of scalable video search method based on digital watermarking |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710937101.5A CN107908647A (en) | 2017-10-10 | 2017-10-10 | A kind of scalable video search method based on digital watermarking |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107908647A true CN107908647A (en) | 2018-04-13 |
Family
ID=61840327
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710937101.5A Pending CN107908647A (en) | 2017-10-10 | 2017-10-10 | A kind of scalable video search method based on digital watermarking |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107908647A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110545299A (en) * | 2018-05-29 | 2019-12-06 | 腾讯科技（深圳）有限公司 | content list information acquisition method, content list information providing method, content list information acquisition device, content list information providing device and content list information equipment |
| CN109101791B (en) * | 2018-09-21 | 2021-03-02 | 北京仁信证科技有限公司 | Watermark parameter automatic acquisition method and system based on data characteristics |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663375A (en) * | 2012-05-08 | 2012-09-12 | 合肥工业大学 | Active target identification method based on digital watermark technology in H.264 |
| CN103617233A (en) * | 2013-11-26 | 2014-03-05 | 烟台中科网络技术研究所 | Method and device for detecting repeated video based on semantic content multilayer expression |
| US20160071229A1 (en) * | 2014-09-05 | 2016-03-10 | Brady Jennings Beaubien | Watermarking system |
| CN106557545A (en) * | 2016-10-19 | 2017-04-05 | 北京小度互娱科技有限公司 | Video retrieval method and device |
-
2017
- 2017-10-10 CN CN201710937101.5A patent/CN107908647A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663375A (en) * | 2012-05-08 | 2012-09-12 | 合肥工业大学 | Active target identification method based on digital watermark technology in H.264 |
| CN103617233A (en) * | 2013-11-26 | 2014-03-05 | 烟台中科网络技术研究所 | Method and device for detecting repeated video based on semantic content multilayer expression |
| US20160071229A1 (en) * | 2014-09-05 | 2016-03-10 | Brady Jennings Beaubien | Watermarking system |
| CN106557545A (en) * | 2016-10-19 | 2017-04-05 | 北京小度互娱科技有限公司 | Video retrieval method and device |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110545299A (en) * | 2018-05-29 | 2019-12-06 | 腾讯科技（深圳）有限公司 | content list information acquisition method, content list information providing method, content list information acquisition device, content list information providing device and content list information equipment |
| CN110545299B (en) * | 2018-05-29 | 2022-04-05 | 腾讯科技（深圳）有限公司 | Content list information acquisition method, content list information providing method, content list information acquisition device, content list information providing device and content list information equipment |
| CN109101791B (en) * | 2018-09-21 | 2021-03-02 | 北京仁信证科技有限公司 | Watermark parameter automatic acquisition method and system based on data characteristics |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108228915B (en) | Video retrieval method based on deep learning | |
| CN104504109B (en) | Image searching method and device | |
| WO2020259510A1 (en) | Method and apparatus for detecting information embedding region, electronic device, and storage medium | |
| CN103384883B (en) | Top-K process is utilized to make semantic abundant | |
| CN103608826A (en) | In-video product annotation with web information mining | |
| CN109272440B (en) | Thumbnail generation method and system combining text and image content | |
| JP2011198364A (en) | Method of adding label to medium document and system using the same | |
| CN111191022A (en) | Method and device for generating short titles of commodities | |
| CN101673266A (en) | Method for searching audio and video contents | |
| CN110910175B (en) | Image generation method for travel ticket product | |
| CN107515934A (en) | A kind of film semanteme personalized labels optimization method based on big data | |
| CN102236714A (en) | Extensible markup language (XML)-based interactive application multimedia information retrieval method | |
| CN108268875A (en) | A kind of image meaning automatic marking method and device based on data smoothing | |
| CN104317946A (en) | Multi-key image-based image content retrieval method | |
| CN106529492A (en) | Video topic classification and description method based on multi-image fusion in view of network query | |
| CN111353055B (en) | Cataloging method and system based on intelligent tag extension metadata | |
| CN107451120B (en) | Content conflict detection method and system for open text information | |
| CN113656700A (en) | Hash retrieval method based on multi-similarity consistent matrix decomposition | |
| CN107908647A (en) | A kind of scalable video search method based on digital watermarking | |
| CN113435438B (en) | Image and subtitle fused video screen plate extraction and video segmentation method | |
| CN113992944A (en) | Video cataloging method, device, equipment, system and medium | |
| CN117112814A (en) | False media content mining and identification system and identification method thereof | |
| CN107656760A (en) | Data processing method and device, electronic equipment | |
| CN101714158A (en) | Image searching method based on MPEG7 standards | |
| CN110674678A (en) | Method and device for identifying sensitive mark in video |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180413 |
|
| RJ01 | Rejection of invention patent application after publication |