CN106791539B - A kind of storage and extracting method of film digital program - Google Patents

Abstract

The present invention relates to the storages and extracting method of a kind of film digital program, include the steps that: picture file acquisition；Picture file compression；Audio files acquisition；Sound picture is synchronous；Turn low bit- rate；Cataloguing；Storage publication.Retrieve film；Browse substance film；Submit downloading request；Transcoding synthesis；Output.The present invention increases the modes such as multilingual by picture compression and browsing, under the premise of utmostly guaranteeing film sound picture quality, reduces the memory space of film multi version, and optimization film stores manufacturing process, improves efficiency when film application.Simultaneously in such a way that speech and picture are stored separately, allows the dialogue of a variety of speeches and subtitle to generate, the miscellaneous editions of picture and subtitle, language are formed, to adapt to the needs of a variety of users.

Description

A kind of storage and extracting method of film digital program

Technical field

The present invention relates to the storages and extracting method of a kind of film digital program, are a kind of sides of computer data storage Method is a kind of by cinefilm or the movie program of magnetic tape recording, is changed to towards multilingual, the convenient standard utilized by all kinds of means Afterwards, the method for storing and extracting.

Background technique

Film prophase shoot mainly uses film photographic, or the digital filming gradually risen in recent years.The sound of shooting record, Picture is stored using separate mode, and picture is recorded in film, and SoundRec is in audiotape.During the editing of film later period, Different language voice, subtitle can be generated, mute is combined with multilingual voice, subtitle, a variety of mother matrixs of film are formed, as The basis that subsequent Moviemaking, distribution, value are excavated again, film mother matrix need to be stored using quality high as far as possible.

Existing way is stored using video-tape.Video-tape can stored picture and sound, producer be first simultaneously Movie pictures are recorded in video-tape, then find the synchronous point of picture sound, sound is recorded to the audio of video-tape one by one Track, sound can synchronous reduction with picture when to play back.Same portion's film, different languages are needed according to above method Production polydisc video-tape is saved respectively.If also needing to record more video-tapes it is required that movie pictures show spolen title.

Because there is multilingual multi version, existing video-tape storage mode, required video recording ribbon amount meeting in motion pictures It is very big, it is with high costs.Cause to have to the storage of discard portion mother matrix, instead remakes when utilizing future. Video-tape can only store the film of HD specification simultaneously, and the film specification new for 2K, 4K, 3D etc. can not then be supported.

Summary of the invention

To solve problem of the prior art, the present invention proposes a kind of storage of film digital program and extracting method.Described Under the premise of method is intended to utmostly guarantee film sound picture quality, the memory space of film multi version is reduced, optimization film is deposited Manufacturing process is stored up, efficiency when film application is improved.

The object of the present invention is achieved like this: a kind of storage method of film digital program, used in the method System includes: picture file compression subsystem, audio collection subsystem, sound picture synchronization subsystem, low bit- rate transcoding subsystem, compiles The step of mesh subsystem, matchmaker's money database, multilingual acquisition subsystem, spolen title production subsystem, the method, is as follows:

The step of picture file acquires: for acquiring movie pictures mother matrix, if film mother matrix picture source is that glue turns magnetic, Scanning generates image sequence file, if film mother matrix is digital cinema master, directly acquisition is image sequence file；

The step of picture file compresses: being used for picture file compression subsystem for image sequence file transcoding is lossless compression Video file；

The step of audio files acquires: digital collection is carried out for audiotape of the audio collection subsystem to film, is adopted Collection generates the audio files of the multichannel of PCM sample, this step or with the step of " picture file acquiring " and " picture file compression The step of " synchronize progress；

The synchronous step of sound picture: for the picture file and audio files of acquisition will to be separated, in sound picture synchronization subsystem It is synchronous to carry out sound picture, main method is to browse video and audio on the time line to pass through sound waves using image content as object of reference Shape, acoustic playback find the offset of audio & video in time, repeatedly the cheap amount of adjustment, so that the picture shape of the mouth as one speaks is reached with sound At matching, after confirming offset, the synchronized relation of sound and video is locked；

The step of turning low bit- rate: the lossless compression video and audio files of film, root are read for transcoding cluster subsystem Generate low bit- rate video-voice frequency flow media file according to synchronized relation, the time-code of the low bit- rate file and lossless compression video file when Code frame frame is corresponding；

The step of cataloguing: for Catalog can be carried out to film referring to low bit- rate video and audio file, comprising:

It catalogues the program layer of film, fills in the essential information of film；

Low bit- rate audio-visual content is browsed, access point is marked on a player, goes out point, is one by content record between out points Segment catalogues to segment, fills in the attribute information of segment；Or scene, camera lens are marked based on low bit- rate, it carries out respectively detailed Cataloguing；Inventory information provides database using matchmaker and stores；

The step of storage publication: after the completion of cataloguing, film is put in storage and externally issues, and browses for user search.

Further, the storage method further include:

The step of augmenting multilingual sound collection: it is by each newly-increased languages sound collection for audio collection subsystem Multi-channel sound file；

The synchronous step of multilingual sound picture: synchronous in sound picture for using the low bit-rate video of mother matrix in library for object of reference The time-code that newly-increased languages sound is found in system is cheaply measured, and is adjusted cheap amount, is matched so that the picture shape of the mouth as one speaks is reached with sound, and lock Surely the synchronized relation of newly-increased languages sound and picture；

The step of addendum file is put in storage: database purchase is provided for the audio files of each languages of supplement to be entered matchmaker；Enter matchmaker After providing database, the file of every film storage is respectively high-resolution lossless compression video file, a multilingual sound File, multi-channel sound file, the spolen title file of multiple languages, a low bit- rate video and audio file, between these files It is that frame frame is corresponding on time-code.

Further, the storage method further include:

The step of making spolen title: it for using the low bit- rate video and audio of mother matrix in library for object of reference, is made in subtitle In system, video and audio is played back, and spolen title is made according to picture, sound sentence by sentence, forms the spolen title text with time-code information Part；

The step of addendum file is put in storage: data inventory is provided for the spolen title file of each languages of supplement to be entered matchmaker respectively Storage；Enter matchmaker provide database after, every film storage file be respectively a high-resolution lossless compression video file, it is multi-lingual Kind audio files, multi-channel sound file, the spolen title file of multiple languages, a low bit- rate video and audio file, these texts It is that frame frame is corresponding on time-code between part.

A kind of extracting method for the film digital program extracting above-mentioned storage file, system packet used in the method Include: retrieval subsystem, browsing subsystem, lower subsystems, transcoding synthesis subsystem, output end, the method includes walking as follows It is rapid:

The step of retrieving film: right in database for being provided by way of full text, keyword in matchmaker in retrieval subsystem Film is retrieved, and search result will list all motion pictures；

The step of browsing substance film: for entering film details, the detailed inventory information of film is checked, while playing browsing The low bit- rate audio-visual content of film；

The step of submitting downloading request: for the browsing film mother matrix chosen, application downloading；It is selected when downloading:

Film video and audio file, certain languages audio file, certain languages subtitle file, and set the lattice of downloading object file Formula；

Video format may be selected: 4K, 2K, HD, SD or other low bit- rate formats；

Entire chapter downloading or selection downloading: complete film is then downloaded in selection entire chapter downloading, selection downloading then choose segment or Scene or camera lens are downloaded；

Transcoding synthesize the step of: for transcoding subsystem according to downloading request, read video and audio, audio, subtitle file with And time-code relationship, transcoding synthesis is carried out, preset film format file is generated；

The step of output: the film format file for that will synthesize is exported by output end.

The beneficial effect comprise that: the present invention increases the modes such as multilingual by picture compression and browsing, Under the premise of utmostly guaranteeing film sound picture quality, the memory space of film multi version is reduced, optimization film storage made Journey improves efficiency when film application.Simultaneously in such a way that speech and picture are stored separately, allow a variety of speeches dialogue and Subtitle generates, and the miscellaneous editions of picture and subtitle, language is formed, to adapt to the needs of a variety of users.

Detailed description of the invention

Present invention will be further explained below with reference to the attached drawings and examples.

Fig. 1 is systematic schematic diagram used in one the method for the embodiment of the present invention；

Fig. 2 is the flow chart of one the method for the embodiment of the present invention；

Fig. 3 is the flow chart of the embodiment of the present invention two, three the methods；

Fig. 4 is systematic schematic diagram used in four the method for the embodiment of the present invention；

Fig. 5 is the flow chart of four the method for the embodiment of the present invention.

Specific embodiment

Embodiment one:

The present embodiment is a kind of storage method of film digital program.System packet used in the present embodiment the method It includes: picture file compression subsystem, audio collection subsystem, sound picture synchronization subsystem, low bit- rate transcoding subsystem, cataloguing subsystem System, matchmaker provide database, multilingual acquisition subsystem, spolen title and make subsystem, as shown in Figure 1, the process of the method As shown in Figure 2.

System described in the present embodiment takes double web frames and in such a way that SD material carries out preediting.

Double web frames: double web frames make video/audio exchange and information data exchange respectively use different links. Avoid waste investment or link obstructions, it is ensured that various data are unblocked.

Carry out preediting with SD material: SD equipment cheaper than high definition device reduce investment outlay more；SD source material flow Measure much smaller, bandwidth requirement of the reduction to network, memory；SD editing speed is fast, is first edited with SD, it is possible to reduce examines What piece may cause updates the time it takes, is conducive to improve efficiency.

Picture file compression subsystem: vision signal in film digital program platform and application management system (high definition part) Using lossless compression (lossless) algorithm, therefore, compressed program code rate is relatively high.To the section of 1920 × 1080 resolution ratio For mesh, code rate can be in 400Mbps(50MB/s) left and right.Quickly transmission is impossible to so high code rate in Ethernet It realizes, the storage area network technology that the present embodiment uses optical fiber to combine with Ethernet organizes web frame in pairs, realizes view The common transport of audio data and control information.

Simultaneously, it is contemplated that high-definition program code rate is very high, if each editor's website uses high definition material to be compiled Volume, not only make equipment investment high (high definition device is all very expensive), the quantity of website is also limited by on-line storage body bandwidth It can not extend, editor and processing speed are also greatly affected.From the angle of optimization system scheme, this embodiment introduces dicode rate volumes The technology of collecting.

Video material in film digital program platform and application management system saves two kinds of formats of high standard definition: a kind of It is the lossless compression high definition material of high code rate, for generating final program in blocks, is subdivided into 24psf and two versions of 50i；Separately One is the lower SD materials of code rate, broadcast for carrying out high definition preediting or Standard Definition Television, using mature MPEG2 lattice Formula, resolution ratio selection 720 × 576, code rate is limited in 50Mbps；

In high definition editing process, preediting first is carried out using SD material.Because there are frame frames to correspond to for high standard definition material Mirror according to the single information of EDL code, the synthesis of high definition material is carried out by backstage synthesis system, just so after preediting The available high-definition program made.

In editor's application of film digital program platform, double nets, the dicode rate editing technique of movie program are preferred systems Make mode, it will bring numerous benefits such as low cost, high efficiency.

Low bit- rate transcoding subsystem: the conversion between the main video and audio file for realizing different-format provides a variety of code streams Conversion selection, and the work for extracting key frame automatically can be completed；Another vital task of the subsystem is to complete film Film material content exchange work between digital program storage and outer application system, is an interface with external system.

Audio collection subsystem: if the video/audio of all versions is all directly stored, due to lossless compression HD video has the code rate of 400Mbps or so, will bring powerful obstruction to storage multi version program in blocks.

For this purpose, the present embodiment realizes the optimization storage of multi version program, i.e. complete video file+multi-lingual using more rail technologies The mode of kind audio file++ EDL yards of multi-language captions file list multiple combinations, corresponding different editions solves this problem.

Since the film version of different language is dialogue sound and spolen title difference, a film is with 50i version Based on complete video file and international sound frequency file, various need are just constituted in conjunction with multilingual sound and subtitle file The film version wanted.

The film limited is needed for partial content, according to audit demand, after carrying out editing by software for editing, forms one A version, actually this editing are the editors to time-code information.Storage system does not need to store program in blocks, but stores packet The EDL code for having contained all clip informations of the version is single.By the information and film material of the EDL code list, at any time may be used Directly to generate corresponding program in blocks.When the program in blocks of some version needs called, original high definition is regarded sound by system Frequently, subtitle material is moved back to, and by corresponding EDL code dull time, by backstage synthesis system, automatically forms required program.

It, can also by the support of backstage algorithm due to having specific time-code corresponding relationship with 50i version for 24psf version With more accurate carry out frame alignment, guarantee the efficiency recycled.

For the multi version phenomenon of movie program, this method of the present embodiment can save a large amount of memory space, answer It uses also more flexible.The use of the technology has important practical significance for library of programmes.

Sound picture synchronization subsystem: subtitles for movies exists generally in the form of script holder's subtitle, the field of regular studio Note subtitle not only will record the content of dialogue, also will record the time of dialogue generation, this time will be with the picture of cinefilm Face lattice number measures.After movie program digitlization, spolen title will also be included into platform as a kind of independent material Management.Digital platform requires the time-code of spolen title and video-audio program corresponding, it is therefore desirable to film script holder subtitle and program Format conversion between the spolen title of library.

When there are when multiple language versions, do not need to store multiple video files for a film, it is only necessary to which storage is without word The original video of curtain, then it is equipped with corresponding subtitle version, the shadow of required language version can be obtained by mixing subtitle and video Piece.

Spolen title is stored separately with video, and carries time-code information, so that the work of production different language version subtitle Become fairly simple.

The application of multilingual spolen title processing technique, for reducing cost, abundant application and simple flow are all played Important role.

The multilingual dialogue processing technique feature of the present embodiment:

Spolen title is stored respectively with video and audio, is optionally combined, and is reduced amount of storage caused by due to spolen title difference and is increased Add.

Script holder's subtitle is automatically converted to movie program spolen title, avoids the work of manual edit spolen title, reduces The working strength of editor's spolen title, improves the accuracy of spolen title.

The spolen title of storage exists in the form of unique file, can according to need selection font, font size, and flexibility is big, So that changing subtitle effect, changing the job simplifications such as languages.

Enough information is contained in spolen title format, high-definition movie broadcasts or subtitle and picture are superimposed to cope with.

Matchmaker provides database: the value maximization in order to realize film, and matchmaker, which will provide the program material saved in database, to be Whole film material.And from the angle of application, for different publication direction, different user groups, a high-definition movie Program material may correspond to multiple versions in blocks.Due to taking the mode of individual languages file storage, multilingual, multi version section Mesh coexists simultaneously, but does not increase memory space.It, can be based on the editor basis of last time, than remaking when modifying program The efficiency of editor wants much higher, efficient and convenient.It searches, delete processing all than storing entire convenience in blocks, is convenient for management.

Matchmaker described in the present embodiment provides database and is intended for film applications, also facing in high definition television application, therefore, from inspection The angle of rope inquiry says that the catalogue standard that different application uses is different.In order to realize that the increment of library of programmes utilizes maximization, it should Sufficiently in conjunction with the cataloguing feature of film and TV catalogue standard.

In original TV and film catalogue standard, the volume to cinefilm and Standard Definition Television video and audio is essentially described Mesh.It is using digital high-definition movie program as the data bank of core in library of programmes, original cataloguing system does not build this domain variability Found corresponding catalogue standard.Therefore, it is necessary to be based on library of programmes, original catalogue standard is absorbed and extended, it should Meet the new feature of digital high-definition film, inherits original film and TV catalogue standard again.Specifically, the present embodiment Cataloguing design has following features:

1) tree-like cataloging structure: tree-like cataloging structure can support the multi version of movie program, advantageously reduce into Originally it and recycles.

2) correspondence of the film standard to television standard: film standard and television standard have very big difference on field description, In order to adapt to the retrieval user in two fields, movie program library will according to circumstances establish film standard word in cataloguing automatically Section arrives the corresponding relationship of TV catalogue standard.Whether journalist or film worker can practise according to profession It is used to easy retrieval to required resource.

3) expansible Catalogue content: as the development of catalogue standard can extend the cataloguing field of film in inventory.

4) configurable multilayer cataloguing process: the multi-level cataloguing of program layer, video layer (camera lens, scene, segment) is provided.

Above-mentioned catalogue standard and searching system feature: for the requirement of library of programmes, incorporating things of diverse nature, while having development to create again Newly, meet the cross-application of film and TV.Ease for use, scalability are fully considered in cataloguing design, using and rising for future Grade slot milling.Cataloging structure design is detailed, provides technical foundation for multi-level comprehensive utilization film source material.

The detailed step of the present embodiment the method is as follows, and the process of the method is as shown in Figure 2:

1. the step of picture file acquires: for acquiring movie pictures mother matrix, if film mother matrix picture source is that glue turns magnetic, Generation image sequence file is then scanned, the lossless compression algorithm used during this, it is ensured that video compress file can be by bit It is reduced to original DPX picture, if film mother matrix is digital cinema master, directly acquisition is image sequence file.

Acquisition includes the acquisition and browsing version program acquisition of high definition and SD version.

Current high-definition program source, resolution ratio are all 1920x1080, and breadth ratio is that 16:9(retains the practical breadth of film Than).There are two high definition material is applied, one is digital movie application, the other is high definition television application, therefore frequency is 24psf and two kinds of 50i.The material of high definition version is obtained, the process being related to is coding.

High gel resin turns magnetic acquisition: the HD-SDI vision signal that collection of server is exported by Da Vinci is downloaded in high definition, it is raw At the AVI video file of Zaxel nondestructive compression type.Film is righted for more since glue turns magnetic system program source, is acquired It needs to carry out repeatedly, to generate multiple lossless compression video files.

SD version is obtained from the material of high definition version, treatment process is related to two aspects, first is that picture reduces, i.e., Down conversion, second is that coding.The method of down conversion and the subsequent use of SD version material are related, need to combine follow-up business process To design, it is proposed that obtained by the way of hardware down conversion.The resolution ratio of SD material is 720x576, picture and selecting 16:9 Breadth ratio (retains the practical breadth ratio of film).SD material is applied to Standard Definition Television channel, and frequency should be the 50i of standard.

For browsing material for retrieving, it should reflect the overall picture of high definition material.And from application requirement, point Resolution can be arbitrary (but to meet the requirement of MPEG compression, i.e. the ratio of width to height is 16 multiple).Browsing material should be by It is designed to the breadth ratio of 16:9.It is proposed with 352x208 resolution ratio, breadth ratio is approximately 16:9.Browsing version material can be with It is obtained from SD version material.2. the step of picture file compresses: being used for picture file compression subsystem for image sequence file Transcoding is lossless compression video file.On the one hand the selection of video code model depends on wanting image quality and application It asks, on the other hand depends on the limitation of memory capacity, comprehensively consider the superiority and inferiority and feasibility of existing coding techniques, this system 2K/4K material and HD video material use lossless compression (lossless) coded format, SD video material to use MPEG-2 Coded format；It browses material and is applied to network retrieval, using the Streaming Media WMV format of mainstream.

3. it is rapid that audio files acquires side step: digital collection is carried out for audiotape of the audio collection subsystem to film, Acquisition generates the audio files of the multichannel of PCM sample, this step or with the step of " picture file acquiring " and " picture file pressure The step of contracting " synchronous progress.

The audio signal in high-definition program source includes uncompressed signal (most 8 channels), 5.1 sound channels, stereo, monophone Road；Main carriers are 8 rail audiotapes or D5 video-tape.The wav file preservation at PCM is all acquired regardless of signal.

A. host language audio obtains:

If host language audio is acquired simultaneously with video, the file of generation is the mixed file that video and audio interweaves, and is Separation storage and management are carried out, need to carry out video and audio separation.Independent video file (Zaxel file) and sound are formed after separation Frequency file (multiple wav files), separate storage.

B. multilingual audio obtains:

Multilingual audio may for international sound (left and right acoustic channels), be also likely to be complete 5.1 sound channel；Acquisition is usually independent It carries out, generates wav file so directly acquiring by audio workstation.

C. stereo audio obtains:

Stereo audio needs to be obtained by mixing (Down Mix) under host language or multilingual lossless compression audio, from simplification Process considers that Down Mix is preferred software algorithm (1. referring to " ATSC AC3 standard ") and directly carries out, and audio quality is complete It can guarantee.

D. Streaming Media obtains:

SD video and stereo audio carry out streaming media transcoding and obtain WMV file.

4. the synchronous step of sound picture: for the picture file and audio files of acquisition will to be separated, in sound picture synchronization subsystem Middle progress sound picture is synchronous, and main method is to browse video and audio on the time line to pass through sound using image content as object of reference Waveform, acoustic playback find the offset of audio & video in time, repeatedly the cheap amount of adjustment, so that the picture shape of the mouth as one speaks and sound Reach matching, after confirming offset, locks the synchronized relation of sound and video.

5. the step of turn low bit- rate: the lossless compression video and audio files of film are read for transcoding cluster subsystem, Low bit- rate video-voice frequency flow media file, time-code and the lossless compression video file of the low bit- rate file are generated according to synchronized relation Time-code frame frame is corresponding.

To make editor's more aspect and saving Internet resources, the present embodiment takes the files in stream media of establishment low bit- rate.? High definition broadcasting has been used, can effectively have been saved in this way after the completion of editor using the files in stream media of low bit- rate in editing process Hardware resource about in editing process.

6. the step of cataloguing: for Catalog can be carried out to film, comprising: right referring to low bit- rate video and audio file The program layer of film is catalogued, and the essential information of film is filled in.

Cataloguing is to carry out tissue by establishing Metadata object to film digital resource, arrange and adequately disclose, build The connection of object coupled mode between the vertical various format knowledges of film, to make the management more specification and side of film assets Just.

Low bit- rate audio-visual content is browsed, access point is marked on a player, goes out point, is one by content record between out points Segment catalogues to segment, fills in the attribute information of segment；Or scene, camera lens are marked based on low bit- rate, it carries out respectively detailed Cataloguing；Inventory information provides database using matchmaker and stores.

In general, one section of video is made of the scene of some independent story units of description；One scene is by some semantemes Relevant camera lens composition；And each camera lens is made of some continuous frames, it can be indicated by one or more key frames.

Text index uses keyword to close equally in Video stream information as the pointer of mark sentence, paragraph, document Key frame plays the effect similar with keyword.People often identify scene, story contour level semantic primitive with key frame.Compare vertical frame dimension The some video basic units of grade are camera lenses, and the camera lens in usual video flowing is made of continuous video frame in time, its generation Temporally and spatially continuous movement, corresponds to the primary record start-stop operation of video camera in one scene of table.Camera lens it Between a plurality of types of transient modes, most commonly shear (Cut) may be present, show as the mutation between consecutive frame Camera lens conversion.In addition, there is also some more complex transient modes, such as gradual change.

Key frame is extracted in automation:

Typical Key Frame Extraction algorithm can mainly divide following a few classes:

Method based on camera lens: a key frame is chosen for each camera lens, such as the first frame of each camera lens is elected to be pass The operand of key frame, this method is small, is very suitable for the camera lens that content activity is small or remains unchanged, but not for video camera The camera lens of disconnected movement, the key frame that this method extracts can not effectively express its main contents.

Method based on content analysis: this method is by analyzing video content (color histogram, motion information) at any time Between situation of change to choose the number of required key frame, and extract key frame according to certain rules for camera lens.

Method based on motion analysis: this method finds the part of camera motion most by the calculating to every frame light stream Dot, corresponding frame will be elected to be key frame, and this selection is generally meant that important based on following observation when video camera stops The presence of personage or event.

Certainly there are also other methods, such as with the method based on lens movable, and Unsupervised clustering technology selects to close Key frame etc..

Shot segmentation:

Camera lens in usual video flowing, is made of the video frame of Time Continuous.It corresponds to video camera and once records Start-stop operation, represent the continuous movement of a scene temporally and spatially.There are many transition of type between camera lens Mode, most commonly " shear " show as the camera lens conversion of the mutability occurred in adjacent two interframe.In addition, there is also one A little more complex transient modes, such as fade in, fade out.

Shot segmentation method is divided into two class of uncompressed domain and compression domain.

Uncompressed domain method have the point-to-point based on frame poor (frame difference) compare with 2 kinds of histogram, due to The frame difference comparison algorithm of point-to-point is too sensitive for noise, and uncompressed domain algorithm is all based on histogram mostly at present.Pressure On compression of the contracting domain method based on video frame images is basic, the foundation of cutting is the compressed coefficient of video frame images before and after comparison (generally DCT coefficient), when meeting certain condition i.e. their cuttings be 2 groups of camera lenses.

Currently based on the pumping frame software of uncompressed domain, shear frames or gradual change frame can be extracted simultaneously in video upload, it can also To realize that, to the MPEG2 I frame for having uploaded completion, IBP frame, frame is taken out in the automation of the video formats file such as MPEG4 and DV.Simultaneously We are seeking extensive technological cooperation, and such as VIRAGE and KONAN can also be integrated in the cataloguing and upload system of oneself The technology of equal companies.

By using the subject classification table of comparisons and rear complaint table, the Automatic subject indexing of system may be implemented or divide automatically Class index.If system uses above-mentioned keyword extraction (CKM) technology based on content, automatic classified Indexing skill Art can classify automatically to catalog record while keyword extraction, and Catalogue Work efficiency is greatly improved.

When object of cataloguing has multi-level structure, we can be retouched according to the characteristics of each level using different Item (or attribute) is stated to carry out clear, complete and accurate description to the content characteristic and form feature of cataloguing object and disclose, Here it is layerings to catalogue.Another layer of meaning of layering cataloguing is: cataloguing object gradually being analyzed and indexed, for difference The object of type and recycle value is analyzed and is indexed according to different level depth.

Itself with regard to video, following several levels: complete video, segment, scene, camera lens, key frame can be divided into.Needle To the level of hierarchy structure of this complexity, if only using the index of simple not level, it will cause description information It is disorderly and unsystematic.Because will be with the technical detail to camera lens to the description information of the summary content of entire complete video file Description information is mixed in together.It is all very unfavorable either from the perspective of cataloguing and retrieval recycling.Such side Face can cause the reduction of precision ratio in retrieval, and on the other hand can also be buried in largely very valuable scene, camera lens In complete video file.

Since layering cataloguing mode is more flexible, the cataloguing demand of different levels can satisfy.It on the one hand can be according to each The professional ability of unit determines cataloguing indexing depth.Unit very limited for human and material resources, they can only make The cataloguing of first level can also substantially meet the needs of retrieval.If there is more abundant strength, it is very careful to carry out Cataloguing indexing work, improve the efficiency of resource reutilization, reduce cost of manufacture.It on the other hand, can also be according to different Program category determines different cataloguing indexing depths.For example, may include more with recycle value in news Camera lens can be deep into camera lens layer when indexing to news controlling.And the scene of the program of variety class or game class Camera lens may there is no too big values, and the index of complete video layer can be only carried out in index.This flexible processing side Formula can greatly improve the efficiency of cataloguing, avoid unnecessary labour.

In addition, the professional skill difference that the content of different level needs in cataloguing is very big, some contents for example: topic Name, languages, responsibility item etc. is relatively simple, some contents are then needed by certain induction and conclusion such as: brief introduction, descriptor, Classification etc., there are also some contents then it is very professional such as: shooting gimmick, for these different levels content cataloguing to cataloguing The requirement difference of personnel is far-reaching, can make full use of by using pipeline system to program material progress procedure cataloguing existing Resource improves cataloging efficiency and the quality of data, is a good solution.Using procedure, the cataloguing of stratification has Standby following characteristics:

To cataloguing object gradually analyzed and indexed, for different type and recycle value object according to different Level depth is analyzed and is indexed.

Cataloguing mode is more flexible, can satisfy the cataloguing demand of different levels.

Cataloging structure and workflow are clear.

Resource utilization is higher, and professional Category-makers ' can only be engaged in high-level Catalogue Work, avoids the wasting of resources.

Data processing is checked on layer by layer, and Catalogue quality is relatively easy to control.

Responsibility of cataloguing is clear, and work management is more convenient.

7. the step of storage publication: after the completion of cataloguing, film is put in storage and externally issues, and browses for user search.

The step of augmenting multilingual sound collection: the audiotape for other languages may be after first mother matrix be put in storage It obtains, each newly-increased languages sound collection is multi-channel sound file by audio collection subsystem.

Embodiment two:

The present embodiment is the improvement of embodiment one, is the refinement of embodiment one.Storage method described in the present embodiment is also wrapped It includes:

The step of augmenting multilingual sound collection: it is by each newly-increased languages sound collection for audio collection subsystem Multi-channel sound file.One film often translates into multilingual, appreciates for the spectators of different mother tongues, therefore, a shadow There are many language versions for piece, acquire different language version, are made as the audio files storage of different language.

The synchronous step of multilingual sound picture: synchronous in sound picture for using the low bit-rate video of mother matrix in library for object of reference The time-code that newly-increased languages sound is found in system is cheaply measured, and is adjusted cheap amount, is matched so that the picture shape of the mouth as one speaks is reached with sound, and lock Surely the synchronized relation of newly-increased languages sound and picture.

As obtained the relevant information from film production management system, filling film production task information etc.；Complete every portion Film 50i version HD video and SD video material and SD video are synchronous with stereo audio material, complete every shadow Piece 24psf version HD video is synchronous with audio material, while generating synchronizing information.

The step of addendum file is put in storage: database purchase is provided for the audio files of each languages of supplement to be entered matchmaker.Enter matchmaker After providing database, the file of every film storage is respectively high-resolution lossless compression video file, a multilingual sound File, multi-channel sound file, the spolen title file of multiple languages, a low bit- rate video and audio file, between these files It is that frame frame is corresponding on time-code.

Embodiment three:

The present embodiment is the improvement of above-described embodiment, is the refinement of above-described embodiment.Storage method described in the present embodiment Further include:

The step of making spolen title: it for using the low bit- rate video and audio of mother matrix in library for object of reference, is made in subtitle In system, video and audio is played back, and spolen title is made according to picture, sound sentence by sentence, forms the spolen title text with time-code information Part.

To save memory capacity and adapting to the demand of DVB, film digital program platform and application management system will be to films Spolen title stored.

Caption content will be obtained from studio, be converted by format, be allowed to become the high definition that non-linear equipment can identify Standard spolen title file, the subtitle file will be saved with Unicode format, so that font is modified and is replaced.

Script holder's subtitle of film be usually using film size as standard, and nonlinear subtitle be using timing code as Standard, it to format between the two.Therefore not only contain dialogue content in subtitle file, it also can be containing converted next High definition time-code information, can be corresponding with video and audio realization of material frame frame.

The step of addendum file is put in storage: data inventory is provided for the spolen title file of each languages of supplement to be entered matchmaker respectively Storage；Enter matchmaker provide database after, every film storage file be respectively a high-resolution lossless compression video file, it is multi-lingual Kind audio files, multi-channel sound file, the spolen title file of multiple languages, a low bit- rate video and audio file, these texts It is that frame frame is corresponding on time-code between part.

The data object stored in film digital program storage further includes in addition to the media class data object such as video and audio, subtitle For illustrating and describing the index information (metadata) of these data.Index class data object is mainly the index of metadata of material Information, part of information media class data object generate simultaneously establish, remaining most information needs by Category-makers ' into Row supplement.

Example IV:

The present embodiment is a kind of extracting method of film digital program for extracting above-mentioned storage file, and the method is used System include: retrieval subsystem, browsing subsystem, lower subsystems, transcoding synthesis subsystem, output end, as shown in Figure 3.

Retrieval subsystem: the purpose of retrieval subsystem is by providing retrieval method and retrieval mode as much as possible as use Family provides retrieval service, such as parameter retrieval, full-text search, systematic searching, image retrieval etc..Parameter retrieval is exactly that we are usual The keyword retrieval of described specified range, such as: Title Searching, date retrieval, column retrieval etc..Inspection is then indifferent in full-text search Rope range carries out keyword retrieval generally be directed to all text informations, and this technology is used very simply, interconnected What is used on the net is the most universal.Systematic searching is actually one kind of parameter retrieval, by using the classification system structure of standard, It can be convenient and accurately carry out data-searching.Image retrieval belongs to the technology in more forward position, and more common mode is root at present According to the texture of image, the features such as tone carry out image retrieval, and content-based image retrieval technology is not yet mature at present.

For full-text search mode, if the new Chinese aptitude search enging is added, it can be achieved that including that Chinese presses glossarial index, word rope The intelligent retrieval based on linguistic knowledges such as semantic dictionaries drawn, it is little in view of system initial stage business, first use database itself Own full-text search function.

Browsing subsystem: for browsing material for retrieving, it should reflect the overall picture of high definition material.And from application requirement From, resolution ratio can be arbitrary (but to meet the requirement of MPEG compression, i.e. the ratio of width to height is 16 multiple).Browsing Material should be designed to the breadth ratio of 16:9.It is proposed with 352x208 resolution ratio, breadth ratio is approximately 16:9.Browse version This material can be obtained from SD version material.

Lower subsystems: the content of downloading includes: file download and the downloading of high definition film.

File download: the materials such as high standard definition video, audio, subtitle are output to a variety of storages with document form and are situated between in library of programmes In matter.Current period completes the file download of digital movie material, high definition material, SD material, and a 2K/4K file download lower phase is complete At.

High definition film downloading: in high definition download server export HD-SDI video/audio signal be recorded to HD-D5 or HDCAM-SR.

Film downloading needs video and audio, subtitle superposition output, since high definition film material uses lossless compression (Lossless) it encodes, carries out this superposition and not only need powerful computing capability, also occupy many bandwidth resources.This reality It applies example and solves the problems, such as that subtitle version film is downloaded using two methods:

The subtitle distribution synthetic method on backstage, this method uses background processing mode, transparent for user, obtains It is the file that final downloading needs.The disadvantage is that needing certain waiting time before downloading.

High definition caption machine synchronizes stacking method, and this method advantage is exactly to directly download without waiting, but need to downloading System is simply provided, and guarantees caption machine linkage output.

Transcoding synthesizes subsystem: in film digital program platform and application management system, video and audio media need difference Format supports, needs the support of the video encoding format of high quality such as material is saved and recycled, editing saving then needs respectively Kind SD video encoding is supported, for the ease of retrieval and Internet video and audio browsing, then needs Streaming Media to support.For not The platform value-added service of disconnected extension, it is also desirable to obtain the material of corresponding format, from material database to meet different application environment It needs.

The demand for coping with continually changing a variety of video and audio formats, will be relatively more suitable using pure software transcoding and synthesis system It closes.High definition, the conversion of SD material format for handling entire library of programmes, not only need outstanding encoding and decoding algorithm, hardware environment Seem critically important, the CPU operand of the bigger needs of code rate is bigger, if relying only on single device carries out high definition transcoding and synthesis, effect Rate will be very low.Therefore, the present embodiment makes high definition transcoding by the way of more transcodings and synthesis server distributed arithmetic And the concept feasible of synthesis.

Transcoding and the task of synthesis system carry out United Dispatching by Manager, distribute to idle transcoding and synthesis Server, numerous transcodings and synthesis server form a cluster and can be completed jointly same by distributed computing Transcoding or synthesis task；Manager carrys out the task of dynamic adaptation segmentation by the state of monitoring transcoding and synthesis server, with System operations resource is set to maximize the use.

Originally huge high resolution video and audio data volume can be made to be divided processing in this way, by cooperating, Accelerate the process of transcoding and synthesis.Transcoding and synthesis cluster are formed with cheap PC Server in this way, so that it may complete valuableness The work that UNIX minicomputer could be handled.

In addition, transcoding and synthesis subsystem use exploitation formula architecture design, system supports the standardization exchange lattice such as XML, MXF Formula makes it that can not only be applied to film digital program platform interior, is even more that entire film digital program platform extends industry to outside The important interface of business, what various outer application systems can be convenient by transcoding and synthesis system carries out program friendship with library of programmes It changes.

Transcoding synthon system features: establishing modular system interface, realizes multisystem interconnection；Server background work, collection Group's distributed treatment, automatically dynamic adaptation task, inexpensive, high efficiency；Pure software processes, upgrading prospects are good, at any time can be with More new algorithm, and without increasing additional equipment when increase function.

Output end:

The present embodiment the method (process is as shown in Figure 4), comprises the following specific steps that:

1. the step of retrieving film: it is right in database to be provided by way of full text, keyword in matchmaker for retrieval subsystem Film is retrieved, and search result will list all motion pictures.

Retrieval is to realize the recycling of film digital resource in film digital program storage by retrieval and inquisition, applying On be not independent, but close fusion provides convenient resource retrieval and calls in subsystems.

2. the step of browsing substance film: for entering film details, checking the detailed inventory information of film, while playing clear Look at the low bit- rate audio-visual content of film.

3. the step of submitting downloading request: for the browsing film mother matrix chosen, application downloading；It is selected when downloading:

Film video and audio file, certain languages audio file, certain languages subtitle file, and set the lattice of downloading object file Formula.

Video format may be selected: 4K, 2K, HD, SD or other low bit- rate formats.

Entire chapter downloading or selection downloading: complete film is then downloaded in selection entire chapter downloading, selection downloading then choose segment or Scene or camera lens are downloaded.When film downloading application, some segment, scene or camera lens can be also chosen, application downloading, backstage turns Code group system will synthesize the audio-visual content of the part according to out points information, generate target film file.

4. transcoding synthesis step: for transcoding subsystem according to downloading request, read video and audio, audio, subtitle file with And time-code relationship, transcoding synthesis is carried out, generates preset film format file, and export by output end.

5. the step of output: the film format file for that will synthesize is exported by output end.

Finally it should be noted that being only used to illustrate the technical scheme of the present invention and not to limit it above, although referring to preferable cloth The scheme of setting describes the invention in detail, those skilled in the art should understand that, it can be to technology of the invention Scheme (such as composition of system, the sequencing etc. of step) is modified or replaced equivalently, without departing from the technology of the present invention The spirit and scope of scheme.

Claims (3)

1. a kind of storage method of film digital program, system used in the method include: picture file compression subsystem, Audio collection subsystem, sound picture synchronization subsystem, low bit- rate transcoding subsystem, Catalogue, matchmaker provide database, multilingual The step of acquisition subsystem, spolen title production subsystem, the method, is as follows:

The step of picture file acquires: it for acquiring movie pictures mother matrix, if film mother matrix picture source is that glue turns magnetic, scans Image sequence file is generated, if film mother matrix is digital cinema master, directly acquisition is image sequence file；

The step of picture file compresses: being used for picture file compression subsystem for image sequence file transcoding is lossless compression video File；

The step of audio files acquires: digital collection, acquisition life are carried out for audiotape of the audio collection subsystem to film At the audio files of the multichannel of PCM sample, this step or with the step of " picture file acquiring " and " step that picture file compresses Suddenly " synchronous to carry out；

The synchronous step of sound picture: it for the picture file and audio files of acquisition will to be separated, is carried out in sound picture synchronization subsystem Sound picture is synchronous, and main method is to browse video and audio on the time line, using image content as object of reference, by sound waveform, Acoustic playback finds the offset of audio & video in time, offset is adjusted repeatedly, so that the picture shape of the mouth as one speaks is reached with sound Matching after confirming offset, locks the synchronized relation of sound and video；

The step of turning low bit- rate: the lossless compression video and audio files of film are read for transcoding cluster subsystem, according to same Step relationship generates low bit- rate video-voice frequency flow media file, the time-code of the low bit- rate file and the time-code frame of lossless compression video file Frame is corresponding；

The step of cataloguing: for Catalog can be carried out to film referring to low bit- rate video and audio file, comprising:

It catalogues the program layer of film, fills in the essential information of film；

Low bit- rate audio-visual content is browsed, access point is marked on a player, goes out point, is a piece by content record between out points Section, catalogues to segment, fills in the attribute information of segment；Or scene, camera lens are marked based on low bit- rate, it is compiled in detail respectively Mesh；Inventory information provides database using matchmaker and stores；

The step of storage publication: after the completion of cataloguing, film is to provide database into matchmaker and externally issue, clear for user search It lookes at；

It is characterized in that, the storage method further include:

The step of augmenting multilingual sound collection: for audio collection subsystem by each newly-increased languages sound collection be more sound Road audio files；

The synchronous step of multilingual sound picture: for using the low bit-rate video of mother matrix in library for object of reference, in sound picture synchronization system The middle time-code offset for finding newly-increased languages sound, adjusts offset, matches so that the picture shape of the mouth as one speaks is reached with sound, and locks new Increase the synchronized relation of languages sound and picture；

The step of addendum file is put in storage: database purchase is provided for the audio files of each languages of supplement to be entered matchmaker；Enter matchmaker and provides number Behind library, every film storage file be respectively a high-resolution lossless compression video file, multilingual audio files, Multi-channel sound file, the spolen title file of multiple languages, a low bit- rate video and audio file, between these files on time-code It is that frame frame is corresponding.

2. storage method according to claim 1, which is characterized in that the storage method further include:

The step of making spolen title: for using the low bit- rate video and audio of mother matrix in library for object of reference, in subtitle making system In, video and audio is played back, and spolen title is made according to picture, sound sentence by sentence, forms the spolen title file with time-code information；

The step of addendum file is put in storage: database purchase is provided for the spolen title file of each languages of supplement to be entered matchmaker respectively； Enter matchmaker provide database after, every film storage file be respectively a high-resolution lossless compression video file, it is multilingual Audio files, multi-channel sound file, the spolen title file of multiple languages, a low bit- rate video and audio file, these files Between on time-code be that frame frame is corresponding.

3. the extracting method for the film digital program that the storage method of film digital program described in a kind of claim 1 is stored, System used in the method includes: retrieval subsystem, browsing subsystem, lower subsystems, transcoding synthesis subsystem, output Choose subsystem in end, which is characterized in that described method includes following steps:

The step of retrieving film: for being provided in database in matchmaker to film by way of full text, keyword in retrieval subsystem It is retrieved, search result will list all motion pictures；

The step of browsing substance film: for entering film details, the detailed inventory information of film is checked, while playing browsing film Low bit- rate audio-visual content；

The step of submitting downloading request: for the browsing film mother matrix chosen, application downloading；It is selected when downloading:

Film video and audio file, certain languages audio file, certain languages subtitle file, and set the format of downloading object file；

Video format may be selected: 4K, 2K, HD, SD or other low bit- rate formats；

Entire chapter downloading or selection downloading: complete film is then downloaded in selection entire chapter downloading, and segment or scene are then chosen in selection downloading Or camera lens is downloaded；

The step of transcoding synthesizes: it is requested for transcoding subsystem according to downloading, reads video and audio, audio, subtitle file with timely Code relationship, carries out transcoding synthesis, generates preset film format file；

The step of output: the film format file for that will synthesize is exported by output end.