WO2004051574A1 - 映像システム - Google Patents
映像システム Download PDFInfo
- Publication number
- WO2004051574A1 WO2004051574A1 PCT/JP2003/015370 JP0315370W WO2004051574A1 WO 2004051574 A1 WO2004051574 A1 WO 2004051574A1 JP 0315370 W JP0315370 W JP 0315370W WO 2004051574 A1 WO2004051574 A1 WO 2004051574A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- feature
- representative
- video sequence
- video
- features
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
- G06T7/49—Analysis of texture based on structural texture description, e.g. using primitives or placement rules
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- 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
- G06F16/78—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/783—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
- G06F16/7847—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using low-level visual features of the video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
Definitions
- the present invention relates to a video system and a video identification system.
- the video sequence self-system is similar to the similar video sequence.
- the first conventional video sequence search system using a film is from A. Mufit Fermanetal.
- FIG. 1 shows a video sequence verification system described in the first conventional example, and FIG. 1 shows a video sequence verification system.
- the search system includes a histogram generator 11, a representative histogram generator 12, a description file generator 13, and a representative histogram.
- the histogram generator 11 is a representative hiss that extracts a power histogram from each frame of the target video sequence recorded on the evening base.
- the histogram generator 1 2 is a histogram Description file generation unit 13 that calculates the average value of each bin of the power histogram for all frames extracted by the system generation unit 11 Describes the calculated average value based on a predetermined formula in pd, and stores it in the representative histogram storage unit 14.
- the histogram generation unit 15 queries the averaged image.
- the representative histogram generator 16 that outputs irh a histogram from each frame of the given video sequence is a histogram generator. 15 Calculate the average value of each bin of the force histogram for all frames extracted according to 5 as the representative his V program.
- the similarity calculating section 17 is composed of the representative histogram generated in the representative hiss V-gram generating section 16 and the storage of the representative hissgram and the representative hiss h-gram; Each representative hiss accumulated
- a similar note section 18 for calculating similarity and 0 similarity has a similar image display section 19 for rearranging images recorded in the order of higher similarity.
- Another typical image B that displays images in ascending order is the second system (second conventional example).] 2
- FIG. 2 disclosed in Japanese Patent Publication No. 0 0 0-25 9 8 3 2 shows a video sequence search system of the second conventional example.
- the video sequence identification system is characterized by an extracted frame selection unit 21, a force relay group generation unit 22, and a description file generation unit 2. 3 and a dynamic lag group storage unit 2.4 and a characteristic frame setting unit 25, a dynamic ladder group generating unit 26, and a video identification unit 27 are provided.
- the frame selection section 21 for extracting a feature is to specify a group of target frames for extracting a feature from a target video sequence recorded in the database.
- the force layer group generation unit 22 is a function of the force characteristic of the image from each of the selected frames.
- the description file generation unit 23 that extracts the characteristic villages that indicate the characteristics is described based on a predetermined formula in the generated special villages, and the description file generation unit 23 writes the group into the feature library group storage unit 24.
- the special extraction frame selection section 25 is a question ⁇ A function layer for selecting a target frame group for extracting a feature from a video sequence given as a video sequence
- the unit 26 extracts a feature indicating the characteristic of the image from each of the selected frames.
- the image identification unit 27 is a unit for generating a group of the input groups.
- the M whitening recognition device is provided for each V broadcast X relay.
- a CM extraction device that includes an M extraction device, a plurality of CM extraction devices, and a C-separation device communicably connected via a communication channel, m has a plurality of C-scanner and CM Each of the plurality of C ⁇ scanners provided with the extraction control device is provided with a C ⁇ identification target station in a predetermined TV broadcast X, and broadcasts the corresponding station.
- the CM extraction control device that receives and detects the C channel
- the CM extraction controller that detects the CM channel that was detected by multiple C channels
- the CM was broadcast Time and C
- the C end message including the number of broadcast seconds of M is transmitted to the CM device via the communication channel.
- a CM base including a CM recording file and a comparison unit
- the note is the sound section corresponding to each ⁇ V broadcast XU.
- the PX is set in the CM reference area.
- C The reference file is registered with each CM identification target station.
- the recording file is the one for each TV broadcaster.
- the comparison part which is spoken to the target station by CMi, is the C ⁇ of each YV broadcaster.
- a part similar to the previously registered reference signal is searched for from the input signal.
- the reference signal sequence is derived from the reference signal and the input characteristic is derived from the input signal.
- a quantity sequence is derived, and a note is set for the input feature quantity series.
- the similarity between the reference feature village series and the input feature sequence in the note is calculated, and the similarity is calculated based on the calculated similarity.
- the skip and step width of the eye window are calculated, and the note is moved based on the skip V step width. Based on the similarity obtained at the moved note window position where the giant window is moved.
- a video structuring apparatus disclosed in Japanese Patent Application Laid-Open No. 2000-101-2005 is a conventional video structuring apparatus which extracts a video recording apparatus and image features.
- the recorded image is provided with an extraction device and a structure recording medium for recording the extracted features as a pinch image.
- the recorded image is a power-shot / short-change occupation recording.
- the information obtained by the environmental information recording device is added to the information obtained by the representative frame extraction device, the force information information recording device, and the environmental information recording device.
- a conventional CM fj detector is disclosed in Japanese Patent Application Laid-Open No. 2001-3599-125.
- the conventional CM fj detector is used at home and viewed at that home. Day about the majesty-evening Is sent to the CM Inspection Center as a CM Inspection Key.
- the CM survey device receives the CM output of each broadcasting station, and then receives the CM reduction time, which is the size of the surveyed video portion of the target video, reduced using a specific algorithm.
- Output section which has a recording section and the received CM reduction section-a reception section that records the evening.
- the memory that was actually viewed on the broadcast receiver is stored in the broadcast receiver.
- the output power is reduced by using a specific algorithm to reduce the output power and the output power is recorded. The output power is reduced and the output power is reduced.
- the decision is sent to the CM inspection center
- the image search apparatus disclosed in Japanese Patent Application Laid-Open No. 2002-117170 is a conventional image inspection apparatus.
- the frame feature extraction unit extracts the features of the few ⁇ and most frames included in the search target moving image.
- the frame feature to be output as the output frame feature is stored in the frame feature im output from the frame feature i output from the frame feature i output.
- the part extracts the feature ⁇ of the query image and outputs it as an image feature.
- the similarity calculation part is the frame feature and image feature stored in the frame feature part.
- the frame features that compare the image features output from the extraction unit and calculate the similarity between the two are stored in the frame features.
- the similarity of the selected frame features is less than 1 and the group whose frame similarity satisfies the predetermined condition is a group of similar images. ⁇ Within the group organized in the department By selecting at least one of the frame features with the highest similarity, the image with the frame feature selected by the similar image set is searched. Presented as a result
- the video content V is divided into the image thumbnail portions divided for each scene of the video content. Specify a representative still image for each scene. Then, the representative still images for each scene are stored in the video search information table (evening base) along with the U-development of the movie content.
- the digital The multimedia storage subsystem is configured to execute the multimedia storage and the multimedia storage according to the multimedia storage scheme.
- the computer that stores the information of the
- the computer processing device is operably connected to the first sub-system, which is the first sub-system. It is connected to a computer processing device in a state that allows it to record the sound.
- the reference signal which is a pre-recorded signal
- a feature village series is generated Entered standing
- Input signal for a fixed time with respect to the input signal that is the input signal Note The input signal that is specified and the input signal Note
- the generated feature sequence that generates a special sequence for the input signal in the input signal
- An input signal similarity value indicating the similarity between the generated signal sequence and the generated similarity sequence is calculated, and the movable region indicating the home position can be moved by moving the input signal note window based on the calculated input signal similarity value.
- the position of the input signal note window is determined based on the calculated movable area calculated, the input signal note is set at that position, and the above operation is repeated.
- An input signal similarity value is calculated for each position of the input signal Note S, and the input signal note window is present on the input signal based on a comparison result between the input signal similarity value and a predetermined threshold value. It is determined whether or not a reference signal exists at the position.
- the video sequence description is performed so that the video sequence requiring the reference in frame units can be selected.
- the purpose is to provide a video system and a video description system that can achieve high-speed separate processing when the degree of irregularity is reduced.
- the video description system is divided into a special feature extracting unit for extracting a relay feature from each frame of the video sequence and a special feature extracting unit.
- a representative feature calculation unit for calculating a feature value representing a video sequence from the extracted feature features group.
- the video description system extracts a feature from each frame of the video sequence.
- a representative feature for calculating a relay feature representative of a video sequence from the feature features extracted by the feature extractor and the feature extractor described above.
- A. A calculating section and a representative layer to record the representative rake features calculated by the representative feature section calculating section; a.
- the video description system extracts a characteristic feature from each frame of the video sequence, and a special smoke output unit.
- the representative feature quantity calculation unit obtains an average value of each element of the layout feature quantity extracted by the feature quantity extraction unit and sets the average value as a representative feature quantity. Is preferred.
- the representative feature value calculating unit may arrange the values of the layout feature values extracted by the feature value extracting unit in ascending or descending order. In addition, it is preferable to calculate the median and use it as the representative feature.
- the record can be either a layer or hedge information or a judge history program.
- one of the plurality of properties including Co 1 or Layo Ut and Edge Histogram as a representative feature can be selected from among a plurality of properties, which are representative features.
- AV e. tersecti As a method of calculating the levy, AV e. tersecti It is preferable that the description is based on a description scheme in which one can be selected from a method of nails including a small number of ⁇ .
- the computer-executable video description software port h is a relay feature from each frame of the video sequence.
- the feature of extracting the village The im feature and the feature feature
- the feature of feature extraction ⁇ Calculates the feature of the video sequence from the extracted feature group of features. Realizing the representative special calculation function
- the video recording method is a special extraction step for extracting a relay feature from each frame of the video sequence.
- Notation feature Ms is a special extraction step for extracting a relay feature from each frame of the video sequence.
- a relay feature representing a video sequence is stored as a representative relay feature.
- the video identification system describes a representative feature representing a video sequence as a representative feature.
- ' The sound part and the question ⁇ Layout characteristics from each frame of the video sequence From the feature features extracted by the feature extraction unit and HU flfl feature extraction unit, which represent the query video sequence.
- the representative layer calculated by the representative feature calculator and the representative feature calculator that calculates the feature value
- the feature value and the representative rate feature storage unit D'I By comparing the features with the sequence, a sequence similar to the superimposition sequence ⁇ a video sequence selection section to be selected is provided.
- a video identification sensor stores an attribute characteristic representing a sequence as a representative characteristic.
- Ray pair writing feature section and video sequence each frame: A lay-out feature that records the characteristics of the calculation.
- the feature dm extraction part which extracts the feature from each frame of the image sequence, and the feature extraction Inquiry from the group of featured features that are included in the query.
- a video sequence section for comparing a layer feature calculated by the section with the feature set in the layer feature amount storage section and selecting a sequence to be queried. The sequence selected by the video sequence selection section is described.
- the feature quantity extracted by the feature extraction section and the feature group described above are extracted.
- a collation unit that refers to the late features described in
- the item self-features output unit calculates the average value of each element of the digit feature extracted by the above-mentioned feature extraction unit, and sets it as a representative feature.
- the representative feature extraction part is a layer extracted by the feature extraction part, and each of the key features of the feature extraction part is sorted in ascending or descending order. Is calculated, and may be used as a surrogate feature quantity.
- the layout quantity may be color layout information or even history program information.
- a computer-executable video identification software X-app ⁇ -duck represents a relay feature representing a video sequence.
- the recording function to be described in the storage function of the recording and the question stored in the recording function.
- the representative layer of the video sequence The features and the recorded video. Compared with the sequence characteristics of the representative layer, it is possible to implement the m-set function to select the video sequence similar to the video sequence.
- the video distinction method is to write a layer feature representing a video sequence as a representative layer feature and a sound step.
- the question that was recorded by the storage step was represented by the following: ⁇ Representative of the video sequence y Features ⁇ Features and substitutes for the recorded video sequence
- a video sequence selection step is provided to select a video sequence similar to the query video sequence compared to the key features.
- the description scheme is a description scheme for describing the features of the video sequence, and is represented as Co 10 r Layo Ut. Abi E dge
- One of a plurality of features including at least H i st o g r a m can be selected. ⁇
- a V e r a g e M e d i a n In t e r S e C t i o n is reduced.
- the processing cost is reduced for the recorded video sequence determined to be unlikely to have the same content when compared with the representative layer feature. You can skip all the required frame-by-frame comparisons of the required special feature groups. Therefore, if the degree of el is reduced, the processing cost can be significantly reduced.
- FIG. 1 shows the system of the first conventional example.
- FIG. 2 is a diagram showing a system of a second conventional example.
- FIG. 3 is a block diagram showing a configuration according to the first embodiment of the present invention.
- FIG. 4 is a flow chart showing a procedure for generating a representative relay characteristic based on an average in the first embodiment.
- FIG. 5 is a flowchart showing a procedure for generating a representative relay characteristic by acquiring a main in the first embodiment.
- FIG. 6 is a diagram showing an example of a force-layer value obtained by averaging in one embodiment.
- FIG. 7 is a diagram showing an example of a force value for acquiring a main in the first embodiment.
- FIG. 8 is a diagram showing a description scheme in one embodiment.
- FIG. 9 is a flowchart showing the video identification procedure of the first embodiment.
- FIG. 10 is a block diagram showing the configuration of the second embodiment of the present invention.
- FIG. 11 is a block diagram showing the configuration of the third embodiment of the present invention.
- FIG. 12 is a block diagram showing the configuration of the fourth embodiment of the present invention.
- FIG. 13 is a block diagram showing the configuration of the fifth embodiment of the present invention.
- FIG. 14 is a diagram showing an aggregation method that can be used for each feature value in the first embodiment.
- FIG. 15 is a diagram showing the binary description syntax in the first embodiment.
- FIG. 16 shows the aggregation method in the first embodiment.
- C, ina U is a diagram showing an example of the current description.
- FIG. 17 is a diagram showing another description scheme in the first embodiment. BEST MODE FOR CARRYING OUT THE INVENTION
- FIG. 3 is a block diagram showing a configuration example of the video identification system according to the first embodiment of the present invention.
- the video identification system includes a description section 31, a pL shirt section 32, and a description section 31 provided with an identification section 33.
- a feature description that describes the signal features of a video by generating and generating a feature from the video sequence.
- the feature description section 3 2 describes the feature quantity of the video to be registered.
- '1 Identifier 3 3 identifies the query video sequence
- the description part 31 is composed of a layer feature extraction part 101 and a representative layer feature day part.
- the feature extraction unit 101 is a dynamic RAM extractor that extracts the rate features from each frame of the recorded video sequence. Is different from the query image in color, but it is difficult to identify images with the same content. Since it is unlikely that the image will change, it is effective to use a relay feature that expresses a rough composition of the image to separate the images. Therefore, the late features are used. As a special feature A variety of things can be considered, but a
- the representative layer feature calculating unit 102 stores a video sequence to be registered from the group of layer features generated by the rate feature extracting unit 101. Calculate the representative representative characteristics. There are various methods for calculating the representative characteristic features, but a method using the average of the characteristic features is used. Figure 4 and Figure 5 show the features of these methods.
- Figure 4 shows the procedure for generating a representative relay characteristic amount by averaging.
- the characteristic is a relay characteristic, which is a U-channel, with 6 luminosity coefficients and 3 chrominance coefficients.
- Fig. 6 shows the procedure using RLAY 0 UT as an example.
- Fig. 6 shows the value of each element of Co L 0 RLAY ⁇ U ⁇ of each frame.
- YDC is the luminance DC component C b DC
- C r DC is the chrominance DC component
- YAC [n] is the ⁇ -th hundredth luminance AC component
- C b AC [n] is the ⁇ -th largest chrominance AC component
- the ⁇ -measurement value of each element of the rate feature is set to 0 (step A1).
- the rate feature is extracted by the rate feature extraction unit 101.
- the frame characteristic amount of the frame is inserted 57C per frame ⁇ 3
- the frame characteristic amount All values of each element of) are calculated [] for the frame
- E RA G E is the representative lake feature.
- FIG. 5 is a flowchart showing a procedure for generating a representative layout feature by a main engineer.
- C 0 L has 6 luminance coefficients and 3 color difference coefficients as layout features.
- ⁇ YDC is the luminance DC component
- CbDC and CrDC are the color difference DC components
- YAC [n] is the nth largest luminance AC component
- AC [n] C r AC [n] indicates the n-th largest color difference AC component.
- the total value of each element of the layer ⁇ h feature is reset to 0 (step B 1).
- the layout feature power of all the frames extracted by the layout feature extraction unit S The elements of all the frame features that are read in and read The values are sorted in ascending order as needed (steps B2 to B3). For example, when focusing on the key YDC (luminance DC component) in the table in Fig. 7,
- the results are sorted in ascending order for each element of the rate feature of all frames, which is 3 2, 3 1 2 1, 1 5 1 1 -
- the median is obtained from the character string and the main is determined (Step B4).
- the element YDC luminance DC component
- the calculated group of mains that is 1 is regarded as the representative layer feature (Step B5).
- the value indicated in the MEDIAN The representative layer is a special village
- the extracted representative features are defined by a description scheme that is described in advance based on a predetermined description scheme. It can be exchanged with other terminals and used between the terminals.
- Fig. 8 shows an example of the description scheme. Either Lay 0 ut or Edge Histogram is selected and the method of generating the representative layer characteristic (aggreg 'a. Tic * n) is equalized (: A ⁇ Ver ⁇ age; Defined so that any one of the median (Median) minimum values (Intersection) is selected.
- Fig. 15 shows the scene description of the feature description in the pina U expression.
- a method of generating a feature layer by first representing a feature layer in a feature of a V-style representation of a V-type example. on) is specified. 1 is described when writing an aggregation, otherwise 0 is allocated.
- an average (A) that describes the method for generating the ⁇ representative layered features when describing the aggregation is described.
- an aggregation method that specifies the method of augmentation to generate a surrogate feature from the frame / image group A scene can apply one of the following three types, which are performed on the entire element included in the frame or image group.
- An averaged U-gage is a feature area of each frame image, which is obtained by successively adding each element, and is normalized by the number of frames / images to form a feature area. Taste as each one is served
- A-Z-A-G-A-A-N is a list generated by rearranging the values of each element in each frame Z image in ascending order. To get the price and taste
- Each element that constitutes the representative characteristic is defined by the elements in the feature that are included in all frames / images. Taking the lowest value of and getting it better
- Figure 17 shows another example of a skim.
- a representative feature it is not limited to Co 1 or Layo Ut or Edge Histogram, but it is specified by ⁇ MPEG-7. If one of the image features ⁇ , multiple instances can be created.
- PturectureTyPe is a recording scheme that unites still image features and that the aggregating method was always established. It is not necessary to select ⁇ , undefined (uns Specified), or use ⁇ , to define a system-specific aggregation method based on ⁇ . You can do it
- the record section 32 that can describe itself with a significantly smaller amount of information is represented by a representative feedback feature storage section 103 and a relay feature quantity group record '1 1 sound section 10. 4 and are provided
- the representative layer feature record '1 sound section 103 is used to record the representative layer feature amount calculated by the representative layer feature calculation section 102.
- the feature group storage unit stores the rake feature of each frame extracted by the layer feature extraction unit 101.
- the discriminating section 33 is composed of a layer h special author extraction section 105, a representative layer feature area calculating section 106, and a video sequence ⁇ setting section.
- the feature extraction unit 105 extracts a feature from each frame of the input query video sequence.
- the Vb output method is a feature. Same as calculation unit 101
- the representative lay-out feature calculation unit 106 represents a video sequence inquired from the lay-out feature set extracted by the lay-out feature extraction unit 105.
- the calculation method for calculating the representative relay characteristic amount is the same as that of the representative relay characteristic calculation unit 102.
- the video sequence selection unit 107 is responsible for the questions generated by the representative rate feature value calculation unit 106.
- ⁇ The representatives of the video sequence and the representative relays.
- ⁇ Compare the features and evaluate the similarity to ask questions.
- the content is the same as the video sequence. Select only registered video sequences that have the potential to be registered Movie Sequence Separate Part 1 08 is Movie Sequence m fixed part
- the relay feature calculation unit 105 For the recorded video sequence selected in 107, the relay feature calculation unit 105 The query generated from the 105 ⁇ Layout feature of the video sequence The group and the layer feature group of the registered video sequence stored in the relay feature group section 104 are compared. Evaluate the similarity by comparison, determine whether the content is the same as the query video sequence, and output the registered video sequence identifier of the content when there is a recorded video with the same content
- FIG. 9 is a flowchart illustrating the video sequence identification procedure in this embodiment.
- the representative layer features are stored in advance in the representative layer feature storage section 103, and the identifiers representing the images are stored for the representative layers for all the registered video sequences.
- the rake feature villages in all of the recorded video sequences are identified with the identifiers representing the video. Note that ⁇ (Step C 1)
- the video sequence selection unit 107 asks the representative video sequence that contains one representative layer feature of the registered video sequence. Compare each of the representative relays in the frame with each other (Step
- Step C 4 C 5 Calculate the distance to the layer (feature) of the transcribed phenomena of the captured phenomena, and judge whether the calculated distance is less than or equal to a predetermined threshold.
- Step C 6 If the calculated distance is less than or equal to the specified threshold value Q: Judge that the video sequence and the registered video sequence may be the contents And the corresponding registered video sequence identifier is output (YES in step C6)
- the video sequence identification unit 108 incorporates a relay feature group corresponding to the identifier of the video sequence determined to possibly have the same content (step Step C 7)
- the video sequence selection section 107 and the video sequence separate section 108 are provided.
- the registered video sequence determined to have no possibility of having the same content in the occupation is a layer that requires processing in the registered video sequence. Since skipping all the comparison processing in the early stage of the special group can be difficult, reducing the discrimination m degree can greatly reduce the processing time.
- the video relay is performed by the representative relay
- the video identification system has a layer h characteristic area iptl output section 105 and a representative rake characteristic calculation section 10 0.
- the representative layer features of the video sequence and the layer features are pre-extracted, and the representative layer features are stored. It is stored in the feature record section 104.
- the video sequence selection unit 107 is a representative record of the representative record feature.
- the question recorded in the storage unit 103 is the representative record of the video sequence. H Features and recorded video sequence. Only the registered video sequences that were judged to be likely to be the query video sequence and the
- Video sequence ⁇ 5 ⁇ Separate part 1 08 is the video sequence selection part
- the registered video sequence specified in 107 is defined in the following table. A comparison is made between the relay feature group and the registered feature sequence of the registered video sequence to evaluate the similarity, determine whether the content is the same as the query video sequence, and determine whether the content is the same as the query video sequence. If there is a registered video ⁇ A differentiator of the registered video sequence is output at the time of feature registration. It becomes possible.
- a video sequence selection section 107 and a video sequence separate section 108 were set up, and when comparing the representative layer and the feature area, they were the same. Skip the comparison of the relay feature group that requires a processing slot for the recorded video sequence judged to be unlikely to be the content. Therefore, if the discrimination phase is omitted, the collation processing h can be greatly reduced.
- FIG. 11 shows a video identification system according to a third embodiment of the present invention.
- the video identification system according to the third embodiment is a description unit 31 that extracts a feature amount from a video sequence and describes a signal feature of the video. It has a storage unit 35 for recording special features of the video and a selection unit 36 for the question ⁇ video sequence.
- the description part 31 is the same as in the first embodiment
- the storage unit 35 has a representative layer h record section 103. o
- a representative record characteristic record '1' section 103 is a description section.
- the m constant part 36 is a layer h characteristic area i idms output part 105, a representative relay characteristic calculation part 106 and a video sequence constant part.
- the functions of the layer ⁇ h feature extraction unit 105, representative relay unit calculation unit 106, and video sequence selection unit 107, which are provided with 107, are the first. Same as the embodiment o
- the video sequence selection unit 107 is marked X and a registered video sequence that is unlikely to have the same content is specified.
- Output of characteristic features According to the part 102, the amount of the relay sequence features of the video sequence is generated and the representative layer ⁇ h features are described in Part 1 104. Can select a registered video sequence with the same content
- FIG. 12 shows an image-specific system according to a fourth embodiment of the present invention.
- Extraction unit 105 and representative layer ⁇ ⁇ Characteristic calculation
- ⁇ The representative layer h characteristic of the video sequence is calculated in advance and the representative layer.
- the sound is recorded in the special memory section 103.
- the video sequence selection unit 107 is deselected so that a registered video sequence that is unlikely to have the same content can be selected.
- the calculation unit 102 generates a representative layer feature amount of the video sequence, which is stored in the representative layer feature amount storage unit 104, and the same contents can be obtained with a small amount of cost. It is possible to define a registered video sequence with
- FIG. 13 shows a video-specific system according to the fifth embodiment of the present invention. As shown in FIG. 13, the video description system according to the present embodiment is more specific than the video sequence. Ms
- It has a description part 31 that describes the signal characteristics of the video and a part 35 that describes the characteristics of the video to be registered.
- the operation of the description part 3 1 and '[' sound part 3 5 is the same as in the three examples.
- the representative layer feature calculation unit 102 Therefore, a representative layer of the video sequence is generated, and a feature layer is generated, and the representative layer feature section 104 is used. Information will be provided to select a case
- This system shows a mechanical identification system using commercials (CMs) as images.
- CMs commercials
- a power-layer descriptor is generated from received broadcast images. It is compared with each CM registered in the evening. ⁇ It is identified what the CM on the air is going to be.The identification result is notified to the user as broadcast control report h.
- a 15-second C-image is registered.
- a power layer is output from each frame of the CM image, and each CM image is output from the dm-output power-layer group.
- the calculated force layer is calculated.
- the calculated force layer and the extracted force layer group are pre-recorded in c.
- the CM that could have the same content as the video that is being broadcast, which is a video sequence is determined. Then, the selected C ⁇ is compared with the force ladder of all the frames in the current image sequence.
- the comparison processing in the frame unit of the feature feature group that requires processing steps is skipped, so that the degree of difference is reduced. ⁇ Processing ⁇ Costs are greatly reduced
- the description of the oxtsu is based on the schema, and the representative relay features are described on a base.
- the representative relay features are described by a plurality of systems. You can call the rake feature
- ⁇ is a CPU (not shown) in a well-known computer system. Implements the functions of each part by executing a software viewer stored in a memo U (not shown)
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Library & Information Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Description
Claims
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2003801005074A CN1692373B (zh) | 2002-12-02 | 2003-12-02 | 图像识别系统图像识别方法 |
EP03776000A EP1569169A4 (en) | 2002-12-02 | 2003-12-02 | VIDEO SYSTEM |
US10/516,086 US20060104514A1 (en) | 2002-12-02 | 2003-12-02 | Image system |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2002350021 | 2002-12-02 | ||
JP2002-350021 | 2002-12-02 | ||
JP2003-066423 | 2003-03-12 | ||
JP2003066423A JP2004234613A (ja) | 2002-12-02 | 2003-03-12 | 映像記述システムおよび方法、映像識別システムおよび方法 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2004051574A1 true WO2004051574A1 (ja) | 2004-06-17 |
Family
ID=32473672
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2003/015370 WO2004051574A1 (ja) | 2002-12-02 | 2003-12-02 | 映像システム |
Country Status (5)
Country | Link |
---|---|
US (1) | US20060104514A1 (ja) |
EP (1) | EP1569169A4 (ja) |
JP (1) | JP2004234613A (ja) |
CN (1) | CN1692373B (ja) |
WO (1) | WO2004051574A1 (ja) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE102004042360A1 (de) * | 2004-09-01 | 2006-03-09 | Sonavis Gmbh | Verfahren zum Erkennen von Videosequenzen in einem aufgenommenen Videofilm |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4336813B2 (ja) * | 2002-12-06 | 2009-09-30 | 日本電気株式会社 | 画像記述システムおよび方法 |
JP4420085B2 (ja) | 2007-08-20 | 2010-02-24 | ソニー株式会社 | データ処理装置、データ処理方法、プログラムおよび記録媒体 |
WO2010084736A1 (ja) * | 2009-01-23 | 2010-07-29 | 日本電気株式会社 | 映像記述子生成装置 |
EP2383990B1 (en) * | 2009-01-29 | 2017-09-20 | Nec Corporation | Time segment representative feature vector generation device |
KR101599465B1 (ko) * | 2009-03-03 | 2016-03-04 | 삼성전자주식회사 | 동기화 정보 제공 서버 및 방법과, 클라이언트 기기 및 방송 프로그램과 부가 정보의 동기화 방법 |
JP5510012B2 (ja) * | 2010-04-09 | 2014-06-04 | ソニー株式会社 | 画像処理装置および方法、並びにプログラム |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0668163A (ja) * | 1992-08-17 | 1994-03-11 | Nippon Telegr & Teleph Corp <Ntt> | 映像地図作成方法 |
US5805746A (en) * | 1993-10-20 | 1998-09-08 | Hitachi, Ltd. | Video retrieval method and apparatus |
EP1195692A1 (en) * | 1999-06-30 | 2002-04-10 | Sharp Corporation | Dynamic image search information recording apparatus and dynamic image searching device |
US20020169761A1 (en) * | 2001-05-14 | 2002-11-14 | Fujitsu Limited | Data search system and data search method |
Family Cites Families (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5085746A (en) * | 1990-09-10 | 1992-02-04 | North Carolina State University | Method of fabricating scanning tunneling microscope tips |
JP3780623B2 (ja) * | 1997-05-16 | 2006-05-31 | 株式会社日立製作所 | 動画像の記述方法 |
US6052492A (en) * | 1997-12-09 | 2000-04-18 | Sun Microsystems, Inc. | System and method for automatically generating an image to represent a video sequence |
JP3615657B2 (ja) * | 1998-05-27 | 2005-02-02 | 株式会社日立製作所 | 映像検索方法及び装置ならびに記録媒体 |
KR100305591B1 (ko) * | 1998-07-22 | 2001-11-30 | 오길록 | 관절점기반동작정보를사용한동영상검색방법 |
US6389181B2 (en) * | 1998-11-25 | 2002-05-14 | Eastman Kodak Company | Photocollage generation and modification using image recognition |
US6396963B2 (en) * | 1998-12-29 | 2002-05-28 | Eastman Kodak Company | Photocollage generation and modification |
CN100409234C (zh) * | 2000-03-23 | 2008-08-06 | 三菱电机株式会社 | 图像检索配送系统和图像检索配送方法 |
JP4078085B2 (ja) * | 2001-03-26 | 2008-04-23 | キヤノン株式会社 | 変倍画像生成装置、方法、及びそのコンピュータプログラムとコンピュータ読み取り可能な記憶媒体 |
-
2003
- 2003-03-12 JP JP2003066423A patent/JP2004234613A/ja active Pending
- 2003-12-02 CN CN2003801005074A patent/CN1692373B/zh not_active Expired - Fee Related
- 2003-12-02 US US10/516,086 patent/US20060104514A1/en not_active Abandoned
- 2003-12-02 EP EP03776000A patent/EP1569169A4/en not_active Withdrawn
- 2003-12-02 WO PCT/JP2003/015370 patent/WO2004051574A1/ja active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0668163A (ja) * | 1992-08-17 | 1994-03-11 | Nippon Telegr & Teleph Corp <Ntt> | 映像地図作成方法 |
US5805746A (en) * | 1993-10-20 | 1998-09-08 | Hitachi, Ltd. | Video retrieval method and apparatus |
EP1195692A1 (en) * | 1999-06-30 | 2002-04-10 | Sharp Corporation | Dynamic image search information recording apparatus and dynamic image searching device |
US20020169761A1 (en) * | 2001-05-14 | 2002-11-14 | Fujitsu Limited | Data search system and data search method |
Non-Patent Citations (1)
Title |
---|
See also references of EP1569169A4 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE102004042360A1 (de) * | 2004-09-01 | 2006-03-09 | Sonavis Gmbh | Verfahren zum Erkennen von Videosequenzen in einem aufgenommenen Videofilm |
DE102004042360B4 (de) * | 2004-09-01 | 2006-06-08 | Sonavis Gmbh | Verfahren zum Erkennen von Videosequenzen in einem aufgenommenen Videofilm |
Also Published As
Publication number | Publication date |
---|---|
JP2004234613A (ja) | 2004-08-19 |
CN1692373A (zh) | 2005-11-02 |
EP1569169A1 (en) | 2005-08-31 |
EP1569169A4 (en) | 2009-07-15 |
US20060104514A1 (en) | 2006-05-18 |
CN1692373B (zh) | 2010-04-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9769545B2 (en) | System and method for automatically authoring interactive television content | |
US9098585B2 (en) | Clustering multimedia search | |
US6892193B2 (en) | Method and apparatus for inducing classifiers for multimedia based on unified representation of features reflecting disparate modalities | |
US6404925B1 (en) | Methods and apparatuses for segmenting an audio-visual recording using image similarity searching and audio speaker recognition | |
US7616840B2 (en) | Techniques for using an image for the retrieval of television program information | |
US6774917B1 (en) | Methods and apparatuses for interactive similarity searching, retrieval, and browsing of video | |
US6751354B2 (en) | Methods and apparatuses for video segmentation, classification, and retrieval using image class statistical models | |
US8170392B2 (en) | Method and apparatus for generation, distribution and display of interactive video content | |
JP3568117B2 (ja) | ビデオ画像の分割、分類、および要約のための方法およびシステム | |
US20030123850A1 (en) | Intelligent news video browsing system and method thereof | |
CN1524237A (zh) | 自动内容分析及多媒体演示的表示 | |
KR20060027377A (ko) | 멀티미디어 스트림들의 멀티미디어 요약을 생성하기 위한시스템 및 방법 | |
JP2008533580A (ja) | オーディオ及び/又はビジュアルデータの要約 | |
Bost et al. | Remembering winter was coming: Character-oriented video summaries of TV series | |
WO2004051574A1 (ja) | 映像システム | |
WO2019236661A1 (en) | Manufacture of image inventories | |
CN1672210A (zh) | 用于生成基于内容的内容表格的方法、系统和程序产品 | |
KR100436828B1 (ko) | 주문형 동영상 요약 서비스 시스템 | |
US20080162451A1 (en) | Method, System and Computer Readable Medium for Identifying and Managing Content | |
O'Connor et al. | Fischlar: an on-line system for indexing and browsing broadcast television content | |
US8166024B2 (en) | Information processing apparatus and method, program, and recording medium | |
JP5213747B2 (ja) | 映像コンテンツ保管視聴システムおよび方法 | |
Taskiran | Automatic methods for content-based access and summarization of video sequences | |
WO2005098670A1 (en) | Method and electronic device for classifying a video content item | |
JPH11175536A (ja) | 映像情報構造化処理方法および装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): CN US |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IT LU MC NL PT RO SE SI SK TR |
|
DFPE | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101) | ||
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
ENP | Entry into the national phase |
Ref document number: 2006104514 Country of ref document: US Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 10516086 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2003776000 Country of ref document: EP Ref document number: 20038A05074 Country of ref document: CN |
|
WWP | Wipo information: published in national office |
Ref document number: 2003776000 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 10516086 Country of ref document: US |