WO2016027457A1 - コンテンツ認識装置およびコンテンツ認識方法 - Google Patents
コンテンツ認識装置およびコンテンツ認識方法 Download PDFInfo
- Publication number
- WO2016027457A1 WO2016027457A1 PCT/JP2015/004112 JP2015004112W WO2016027457A1 WO 2016027457 A1 WO2016027457 A1 WO 2016027457A1 JP 2015004112 W JP2015004112 W JP 2015004112W WO 2016027457 A1 WO2016027457 A1 WO 2016027457A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- fingerprint
- recognition
- video content
- unit
- recognition data
- Prior art date
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/80—Generation or processing of content or additional data by content creator independently of the distribution process; Content per se
- H04N21/83—Generation or processing of protective or descriptive data associated with content; Content structuring
- H04N21/835—Generation of protective data, e.g. certificates
- H04N21/8352—Generation of protective data, e.g. certificates involving content or source identification data, e.g. Unique Material Identifier [UMID]
-
- 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
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
-
- 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
- G06V20/48—Matching video sequences
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/80—Generation or processing of content or additional data by content creator independently of the distribution process; Content per se
- H04N21/83—Generation or processing of protective or descriptive data associated with content; Content structuring
- H04N21/835—Generation of protective data, e.g. certificates
- H04N21/8358—Generation of protective data, e.g. certificates involving watermark
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
- G06F16/68—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/683—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the 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
-
- 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
- G06F16/7837—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 objects detected or recognised in the video content
Definitions
- This disclosure relates to a content recognition apparatus and a content recognition method for recognizing video content.
- a communication service that uses technology to recognize content via the cloud has been proposed.
- a television receiver hereinafter referred to as “recognizing input video”, acquiring additional information related to this video via a communication network, and displaying the acquired additional information together with video content on a display screen
- TV Abbreviated as “TV”.
- a technique for recognizing an input video is called “ACR (Automatic Content Recognition)”.
- Patent Document 1 and Patent Document 2 disclose fingerprint technologies. In this technique, a contour of a face or the like reflected in an image frame in a video is detected, a fingerprint is generated based on the detected contour, and the generated fingerprint is collated with data stored in a database.
- This disclosure provides a content recognition apparatus and a content recognition method capable of reducing the processing related to the recognition of video content while increasing the recognition accuracy of the video content.
- the content recognition device includes a fingerprint generation unit, a selection unit, and a verification unit.
- the fingerprint generation unit generates a fingerprint for each of the plurality of acquired video content candidates.
- the sorting unit sorts video content candidates using attached information included in recognition data input from the outside.
- the collation unit collates the fingerprint of the video content candidate selected by the selection unit with the fingerprint included in the recognition data, and determines the video content corresponding to the fingerprint included in the recognition data as the video content candidate. Identify from.
- the content recognition apparatus can reduce processing related to recognition of video content while increasing recognition accuracy of video content.
- FIG. 1 is a block diagram illustrating a configuration example of the content recognition system according to the first embodiment.
- FIG. 2 is a block diagram illustrating a configuration example of the receiving apparatus according to the first embodiment.
- FIG. 3 is a diagram schematically illustrating an example of recognition data transmitted by the receiving device according to the first embodiment.
- FIG. 4 is a diagram schematically illustrating an example of recognition data stored in the fingerprint database according to the first embodiment.
- FIG. 5 is a diagram schematically illustrating an example of a relationship between an image frame of each frame rate extracted by the video extraction unit and the static area in the first embodiment.
- FIG. 6 is a diagram schematically illustrating an example of a relationship between an image frame of each frame rate extracted by the video extraction unit according to Embodiment 1 and a dynamic region.
- FIG. 1 is a block diagram illustrating a configuration example of the content recognition system according to the first embodiment.
- FIG. 2 is a block diagram illustrating a configuration example of the receiving apparatus according to the first embodiment.
- FIG. 7 is a block diagram illustrating a configuration example of the fingerprint generation unit according to the first embodiment.
- FIG. 8 is a flowchart showing an operation example of the content recognition apparatus provided in the content recognition system in the first exemplary embodiment.
- FIG. 9 is a flowchart illustrating an example of processing when generating recognition data in the first embodiment.
- FIG. 10 is a diagram schematically illustrating an example of a change of an image frame in the process of generating recognition data according to the first embodiment.
- FIG. 11 is a flowchart illustrating an example of a process for calculating the amount of change between image frames in the first embodiment.
- FIG. 12 is a diagram schematically illustrating an example of downscale conversion processing of an image frame in the first embodiment.
- FIG. 13 is a diagram schematically illustrating an example of a process for calculating a change amount between image frames in the first embodiment.
- FIG. 14 is a flowchart illustrating an example of a static fingerprint generation process according to the first embodiment.
- FIG. 15 is a diagram schematically illustrating an example of a static fingerprint generated based on the amount of change between image frames in the first embodiment.
- FIG. 16 is a flowchart illustrating an example of a dynamic fingerprint generation process in the first embodiment.
- FIG. 17 is a diagram schematically illustrating an example of an image frame in which a dynamic fingerprint is not generated in the first embodiment.
- FIG. 18 is a diagram schematically illustrating an example of a dynamic fingerprint generated based on the amount of change between image frames in the first embodiment.
- FIG. 14 is a flowchart illustrating an example of a static fingerprint generation process according to the first embodiment.
- FIG. 15 is a diagram schematically illustrating an example of a static fingerprint generated based on the amount of change between image frames in the first
- FIG. 19 is a flowchart illustrating an example of filtering processing executed by the fingerprint filtering unit according to the first embodiment.
- FIG. 20 is a flowchart illustrating an example of the property filtering process executed by the fingerprint filtering unit in the first embodiment.
- FIG. 21 is a diagram schematically illustrating a specific example of the property filtering process executed by the fingerprint filtering unit in the first embodiment.
- FIG. 22 is a flowchart illustrating an example of the property sequence filtering process executed by the fingerprint filtering unit in the first embodiment.
- FIG. 23 is a diagram schematically illustrating a specific example of the property sequence filtering process executed by the fingerprint filtering unit in the first embodiment.
- FIG. 24 is a flowchart illustrating an example of recognition data matching processing according to the first embodiment.
- FIG. 25 is a diagram schematically illustrating an example of a matching process for static fingerprints according to the first embodiment.
- FIG. 26 is a diagram schematically illustrating an example of a dynamic fingerprint matching process according to the first embodiment.
- FIG. 27 is a diagram illustrating an example of recognition conditions for video content in the first embodiment.
- FIG. 28 is a diagram schematically illustrating an example of video content matching processing according to the first embodiment.
- FIG. 29 is a diagram for explaining points that are problematic in recognition of video content.
- FIG. 1 is a block diagram illustrating a configuration example of the content recognition system 1 according to the first embodiment.
- the content recognition system 1 includes a broadcasting station 2, an STB (Set Top Box) 3, a reception device 10, a content recognition device 20, and an advertisement server device 30.
- STB Set Top Box
- Broadcast station 2 is a transmission device configured to convert video content into a video signal and broadcast it as a television broadcast signal (hereinafter also simply referred to as “broadcast signal”).
- the video content is, for example, broadcast content broadcasted by wireless or wired broadcasting or communication, such as program content such as a TV program, and video content for advertisement such as a commercial message (CM) (hereinafter referred to as “advertising content”).
- CM commercial message
- the program content and the advertisement content are switched to each other as time passes.
- Broadcast station 2 transmits the video content to STB 3 and content recognition device 20.
- the broadcasting station 2 transmits the video content to the STB 3 by broadcasting and transmits the video content to the content recognition device 20 by communication.
- the STB 3 is a tuner / decoder configured to receive a broadcast signal broadcast from the broadcast station 2 and output a video signal or the like based on the received broadcast signal.
- the STB 3 receives a broadcast channel selected based on a user instruction from broadcast signals broadcast from the broadcast station 2. Then, the received video content of the broadcast channel is decoded, and the decoded video content is output to the receiving device 10 via the communication path.
- the communication path is, for example, HDMI (registered trademark) (High-Definition Multimedia Interface).
- the receiving device 10 is a video receiving device such as a television.
- the receiving device 10 is connected to the content recognition device 20 and the advertisement server device 30 via the communication network 105.
- the receiving device 10 extracts a plurality of image frames from the received frame sequence of video content, and generates recognition data based on the extracted image frames.
- the receiving device 10 transmits the generated recognition data to the content recognition device 20 and receives the recognition result of the video content from the content recognition device 20.
- the receiving device 10 acquires additional information from the advertisement server device 30 based on the recognition result of the video content, and displays the acquired additional information on the display screen substantially in real time together with the video content.
- the recognition data is data representing video content, for example, and is used for video content recognition (for example, ACR).
- the recognition data includes a fingerprint (hash value) generated based on a change in image between image frames.
- An image frame is a picture that constitutes video content.
- the image frame includes a frame in the progressive method, a field in the interlace method, and the like.
- the content recognition device 20 is, for example, a server device.
- the content recognition device 20 generates respective recognition data from a plurality of input video contents, and collates the recognition data selected from these recognition data with the recognition data sent from the receiving device 10, The video content recognition process is executed.
- the video content recognition processing is also referred to as “image recognition processing” or “image recognition”. If the collation is successful, the content recognition device 20 selects one video content from among the plurality of video content as a result of the image recognition. That is, each of the plurality of video contents is a video content candidate that may be selected as the video content corresponding to the recognition data transmitted from the receiving device 10. Therefore, hereinafter, the video content received by the content recognition device 20 is also referred to as a “video content candidate”.
- the content recognition device 20 is an example of an image recognition device, and is a device configured to recognize video content (for example, ACR).
- the content recognition device 20 receives a plurality of video content candidates being broadcast from a plurality of broadcasting stations 2 and generates recognition data. Then, the content recognition device 20 receives the recognition data transmitted from the reception device 10, and recognizes the video content substantially in real time (image recognition processing) using the received recognition data and the recognition data of the video content candidate. I do.
- the advertisement server device 30 is, for example, a server device that distributes additional information related to a video content recognition result by the content recognition device 20.
- the advertisement server device 30 is an advertisement distribution server that holds and distributes advertisements for various products.
- the content recognition device 20 and the advertisement server device 30 are independent server devices, but the content recognition device 20 and the advertisement server device 30 are included in one Web server. It may be.
- reception apparatus 10 First, reception apparatus 10 according to the present embodiment will be described with reference to FIGS. 1 and 2.
- FIG. 2 is a block diagram illustrating a configuration example of the receiving device 10 according to the first embodiment.
- FIG. 2 shows a main hardware configuration of the receiving apparatus 10.
- the receiving device 10 includes a video reception unit 11, a video extraction unit 12, an additional information acquisition unit 13, a video output unit 14, and a recognition unit 100. More specifically, as illustrated in FIG. 2, the reception device 10 further includes a control unit 15, an operation signal reception unit 16, and an HTTP (Hyper Text Transfer Protocol) transmission / reception unit 17. As illustrated in FIG. 2, the additional information acquisition unit 13 includes an additional information storage unit 18 and an additional information display control unit 19.
- the control unit 15 is a processing unit configured to control each component included in the receiving device 10.
- the control unit 15 includes a nonvolatile memory, a CPU (Central Processing Unit), and a volatile memory.
- the nonvolatile memory is, for example, a ROM (Read Only Memory) or the like, and stores a program (an application program, etc.).
- the CPU is configured to execute the program.
- the volatile memory is, for example, a RAM (Random Access Memory), and is used as a temporary work area when the CPU is operating.
- the operation signal receiving unit 16 is a circuit configured to receive an operation signal output from an operation unit (not shown).
- the operation signal is a signal output from the operation unit when the user operates an operation unit (for example, a remote controller; hereinafter referred to as “remote controller”) in order to operate the receiving apparatus 10.
- the operation unit is a remote control having a gyro sensor
- the operation signal receiving unit 16 outputs information on the physical movement of the remote control itself output from the remote control (the user shakes, tilts, or changes the direction of the remote control). , Etc., may be configured to receive a signal indicating a motion when the same.
- the HTTP transmission / reception unit 17 is an interface configured to communicate with the content recognition device 20 and the advertisement server device 30 via the communication network 105.
- the HTTP transmission / reception unit 17 is a communication adapter for a wired LAN (Local Area Network) that conforms to the IEEE 802.3 standard, for example.
- the HTTP transmission / reception unit 17 transmits the recognition data to the content recognition apparatus 20 via the communication network 105. Further, the HTTP transmission / reception unit 17 receives the recognition result of the video content transmitted from the content recognition device 20 via the communication network 105.
- the HTTP transmission / reception part 17 acquires the additional information transmitted via the communication network 105 from the advertisement server apparatus 30, for example.
- the acquired additional information is stored in the additional information storage unit 18 via the control unit 15.
- the video receiver 11 includes a receiving circuit and a decoder (not shown) configured to receive video content. For example, the video receiver 11 selects a broadcast channel to be received, selects a signal input from the outside, or the like based on the operation signal received by the operation signal receiver 16.
- the video reception unit 11 includes a video input unit 11a, a first external input unit 11b, and a second external input unit 11c.
- the video input unit 11a is configured to input a video signal transmitted from the outside, such as a broadcast signal (indicated as “TV broadcast signal” in FIG. 2) received by an antenna (not shown), for example. Circuit.
- a broadcast signal indicated as “TV broadcast signal” in FIG. 2
- an antenna not shown
- the first external input unit 11b and the second external input unit 11c are a video signal ("external input signal" in FIG. 2) transmitted from an external device such as the STB 3 and a video signal recording / reproducing device (not shown). Interface) configured to input.
- the first external input unit 11b is, for example, an HDMI (registered trademark) terminal, and is connected to the STB 3 by a cable compatible with HDMI (registered trademark).
- the video extracting unit 12 extracts a plurality of image frames at a predetermined frame rate from the frame sequence constituting the video content received by the video receiving unit 11. For example, when the frame rate of the video content is 60 fps (Frames Per Second), the video extraction unit 12 extracts a plurality of image frames at a frame rate of 30 fps, 20 fps, or 15 fps. Note that if the recognition unit 100 at the subsequent stage has a processing capability capable of processing a video of 60 fps, the video extraction unit 12 may extract all the image frames constituting the frame sequence of the video content.
- the frame rate of the video content is 60 fps (Frames Per Second)
- the video extraction unit 12 extracts a plurality of image frames at a frame rate of 30 fps, 20 fps, or 15 fps. Note that if the recognition unit 100 at the subsequent stage has a processing capability capable of processing a video of 60 fps, the video extraction unit 12 may extract
- the additional information acquisition unit 13 operates as a circuit for acquiring information and a communication interface.
- the additional information acquisition unit 13 acquires additional information from the advertisement server device 30 based on the video content recognition result acquired by the recognition unit 100.
- the video output unit 14 is a display control circuit configured to output the video content received by the video receiving unit 11 to the display screen.
- the display screen is, for example, a display such as a liquid crystal display device or an organic EL (Electro Luminescence).
- the additional information storage unit 18 is a storage device configured to store additional information.
- the additional information storage unit 18 is a non-volatile storage element such as a flash memory, for example.
- the additional information storage unit 18 may hold program meta information such as EPG (Electronic Program Guide) in addition to the additional information acquired from the advertisement server device 30.
- EPG Electronic Program Guide
- the additional information display control unit 19 is configured to superimpose the additional information acquired from the advertisement server device 30 on the video content (for example, program content) received by the video receiving unit 11. For example, the additional information display control unit 19 generates a superimposed image by superimposing additional information on each image frame included in the program content, and outputs the generated superimposed image to the video output unit 14. When the video output unit 14 outputs the superimposed image to the display screen, the program content on which the additional information is superimposed is displayed on the display screen.
- the video output unit 14 outputs the superimposed image to the display screen, the program content on which the additional information is superimposed is displayed on the display screen.
- the recognition unit 100 is a processing unit configured to generate recognition data.
- the recognition unit 100 transmits the generated recognition data to the content recognition device 20 and receives a recognition result from the content recognition device 20.
- the recognition unit 100 includes a fingerprint generation unit 110, a fingerprint transmission unit 120, and a recognition result reception unit 130.
- the fingerprint generation unit 110 is an example of a recognition data generation circuit.
- the fingerprint generation unit 110 generates recognition data using a plurality of image frames extracted by the video extraction unit 12. Specifically, the fingerprint generation unit 110 generates a fingerprint for each frame based on an image change between frames. For example, each time the fingerprint generation unit 110 acquires the image frame extracted by the video extraction unit 12, the fingerprint generation unit 110 calculates the amount of change between the image frame acquired immediately before and the fingerprint based on the calculated amount of change. Is generated. The generated fingerprint is output to the fingerprint transmission unit 120.
- the detailed operation of the fingerprint generation unit 110 and a specific example of the generated fingerprint will be described later.
- the fingerprint transmission unit 120 transmits the recognition data generated by the fingerprint generation unit 110 to the content recognition device 20. Specifically, the fingerprint transmission unit 120 transmits the recognition data to the content recognition apparatus 20 via the HTTP transmission / reception unit 17 and the communication network 105 illustrated in FIG.
- FIG. 3 is a diagram schematically illustrating an example of the recognition data 40 transmitted by the receiving device 10 according to the first embodiment.
- the recognition data 40 includes a plurality of fingerprints 43 and timing information 41 and type information 42 associated with each fingerprint 43.
- the timing information 41 is information indicating the time when the fingerprint 43 is generated.
- the type information 42 is information indicating the type of the fingerprint 43.
- the type information 42 includes information indicating a static fingerprint (hereinafter also referred to as “A type”) in which an image change between frames is relatively small, and a dynamic in which an image change between frames is relatively large.
- a type a static fingerprint
- B type a type of information indicating fingerprints
- the fingerprint 43 is information (for example, a hash value) generated based on a change in image between frames of a plurality of image frames included in a frame sequence constituting the video content. As shown in FIG. 3, the fingerprint 43 includes a plurality of feature amounts such as “018” and “184”. Details of the fingerprint 43 will be described later.
- Fingerprint transmitter 120 transmits recognition data 40 to content recognition device 20. At this time, the fingerprint transmission unit 120 sequentially transmits the generated recognition data 40 every time the recognition data 40 is generated by the fingerprint generation unit 110.
- the receiving device 10 transmits the attached information to the content recognition device 20.
- the attached information will be described later.
- the receiving device 10 may transmit the auxiliary information by including it in the recognition data 40 or may transmit it independently of the recognition data 40. Alternatively, there may be both the accessory information included in the recognition data 40 and transmitted and the accessory information transmitted independently of the recognition data 40.
- the recognition result receiving unit 130 receives the recognition result of the video content from the content recognition device 20. Specifically, the recognition result receiving unit 130 receives recognition data from the content recognition apparatus 20 via the communication network 105 and the HTTP transmission / reception unit 17 illustrated in FIG.
- the video content recognition result includes information for specifying the video content. This information is, for example, information indicating a broadcasting station that broadcasts video content, information indicating the name of the video content, and the like.
- the recognition result receiving unit 130 outputs the recognition result of the video content to the additional information acquisition unit 13.
- the content recognition device 20 includes a content receiving unit 21, a fingerprint database (hereinafter referred to as “fingerprint DB”) 22, a fingerprint filtering unit 23, a fingerprint matching unit 24, A fingerprint history information DB 25 and a fingerprint generation unit 2110 are provided. Note that only the fingerprint DB 22 is shown in the content recognition apparatus 20 of FIG. 2 and other blocks are omitted.
- the content receiving unit 21 includes a receiving circuit and a decoder, and is configured to receive video content transmitted from the broadcasting station 2. When there are a plurality of broadcasting stations 2, the content reception unit 21 receives all the video contents generated and transmitted by each of the plurality of broadcasting stations 2. As described above, these received video contents are video content candidates. The content receiving unit 21 outputs the received video content candidates to the fingerprint generation unit 2110.
- Fingerprint generation unit 2110 generates recognition data 50 for each video content candidate. Specifically, the fingerprint generation unit 2110 generates a fingerprint 53 for each image frame based on a change between image frames of a frame sequence constituting the received video content candidate. Therefore, the fingerprint generation unit 2110 generates the fingerprint 53 at the received video content candidate frame rate. For example, if the frame rate of the video content candidate is 60 fps, the fingerprint generation unit 2110 generates 60 fingerprints 53 per second.
- the fingerprint generation unit 2110 included in the content recognition device 20 may have substantially the same configuration and operation as the fingerprint generation unit 110 included in the recognition unit 100 of the reception device 10, for example. Details of the fingerprint generation unit 2110 will be described later with reference to FIG.
- the fingerprint DB 22 is a database that stores recognition data 50 of a plurality of video content candidates.
- identification information for identifying a plurality of video contents for example, content ID (IDentifier)
- recognition data 50 are stored in association with each other.
- the content recognition apparatus 20 Each time new video content is received by the content receiver 21, the content recognition apparatus 20 generates a new fingerprint 53 by the fingerprint generator 2110 and updates the fingerprint DB 22.
- the fingerprint DB 22 is stored in a storage device (for example, an HDD (Hard Disk Drive) or the like) included in the content recognition device 20.
- the fingerprint DB 22 may be stored in a storage device installed outside the content recognition device 20.
- FIG. 4 is a diagram schematically illustrating an example of the recognition data 50 stored in the fingerprint DB 22 according to the first embodiment.
- recognition data 50 a, recognition data 50 b, and recognition data 50 c are stored as recognition data 50 in the fingerprint DB 22.
- the recognition data 50a, the recognition data 50b, and the recognition data 50c are examples of the recognition data 50.
- the recognition data 50a is the recognition data 50 corresponding to the video content ⁇ 1
- the recognition data 50b is the recognition data 50 corresponding to the video content ⁇ 1
- the recognition data 50c is a recognition data 50 corresponding to the video content ⁇ 1.
- the video content ⁇ 1 is a video content candidate broadcast from the broadcasting station ⁇
- the video content ⁇ 1 from the broadcasting station ⁇ the video content ⁇ 1 from the broadcasting station ⁇
- the video content ⁇ 1 from the broadcasting station ⁇ and the video content ⁇ 1 from the broadcasting station ⁇ .
- the recognition data 50a includes a plurality of fingerprints 53a
- the recognition data 50b includes a plurality of fingerprints 53b
- the recognition data 50c includes a plurality of fingerprints 53c.
- the fingerprint 53a, the fingerprint 53b, and the fingerprint 53c are examples of the fingerprint 53.
- Timing information 51a and type information 52a are associated with the fingerprint 53a
- timing information 51b and type information 52b are associated with the fingerprint 53b
- timing information 51c and type information 52c are associated with the fingerprint 53c. It is associated.
- Timing information 51a, timing information 51b, and timing information 51c are examples of timing information 51
- type information 52a, type information 52b, and type information 52c are examples of type information 52.
- the timing information 51 is the same data as the timing information 41 shown in FIG. 3 (that is, information indicating the time when the fingerprint 53 was generated), and the type information 52 is the same data as the type information 42 shown in FIG. (That is, information indicating the type of the fingerprint 53), and the fingerprint 53 is data similar to the fingerprint 43 shown in FIG.
- the fingerprint filtering unit 23 of the content recognition apparatus 20 is an example of a selection unit, and selects video content candidates (hereinafter referred to as “filtering”) using attached information included in the recognition data 40 input from the outside. .
- the fingerprint filtering unit 23 uses the attached information regarding the video content candidate received from the broadcasting station 2 and the attached information acquired from the receiving device 10 to perform matching in the image recognition process from the fingerprint DB 22.
- the recognition data 50 to be read out is filtered and narrowed down for use (that is, for the purpose of collation). In this way, the fingerprint filtering unit 23 selects video content candidates to be collated in the image recognition process by this filtering process.
- the attached information includes the type information 42 and 52 described above.
- the fingerprint filtering unit 23 performs property filtering and property sequence filtering using the type information 42 and 52. Details of property filtering and property sequence filtering will be described later with reference to FIGS.
- the attached information includes information indicating the position of the broadcasting station 2 that transmits the video content received by the receiving device 10 (hereinafter referred to as “geographic information”) or geographic information indicating the position of the receiving device 10. May be.
- Geographic information may be information indicating an area specified based on the IP (Internet Protocol) address of the receiving device 10, for example.
- the fingerprint filtering unit 23 When the geographic information is included in the attached information, the fingerprint filtering unit 23 performs a region filtering process using the geographic information.
- the region filtering process is a process for excluding video content candidates broadcast from the broadcasting station 2 that cannot be viewed in the area indicated by the geographic information, from matching targets in the image recognition process.
- the attached information may include information about the user associated with the receiving device 10 (hereinafter referred to as “user information”).
- the user information includes, for example, information indicating the user's hobbies, preferences, age, sex, or occupation.
- Information indicating a history of video content received by the user at the receiving device 10 may be included in the user information.
- the fingerprint filtering unit 23 When the user information is included in the attached information, the fingerprint filtering unit 23 performs a profile filtering process using the user information.
- the profile filtering process is a process of excluding video content candidates that do not match the user's characteristics or preferences indicated by the user information from the collation targets in the image recognition process.
- the fingerprint filtering unit 23 When the profile filtering process is performed by the fingerprint filtering unit 23, information indicating the characteristics of video content (hereinafter referred to as “content information”) is stored in the fingerprint DB 22 in association with the fingerprint 53. It is desirable.
- the content information includes, for example, information indicating the characteristics of the user who is expected to view the video content.
- the content information may include the genre of the video content, the age group and sex of the user who is expected to view the video content, and the like.
- the fingerprint collation unit 24 of the content recognition device 20 is an example of a collation unit.
- the fingerprint collation unit 24 includes a fingerprint 53 included in the recognition data 50 selected by the fingerprint filtering unit 23, a fingerprint 43 included in the recognition data 40 transmitted from the reception device 10 to the content recognition device 20, and And the video content corresponding to the fingerprint 43 included in the recognition data 40 is specified from among a plurality of video content candidates.
- specifying video content based on the fingerprint 43 is “video content recognition”.
- the fingerprint collation unit 24 selects each feature amount of the fingerprint 43 transmitted from the reception device 10 and received by the content recognition device 20, and the recognition data that is selected by the fingerprint filtering unit 23 and read from the fingerprint DB 22. 50, all the feature quantities of the fingerprint 53 included in 50 are collated. Thereby, the fingerprint collation unit 24 recognizes the video content corresponding to the recognition data 40 transmitted from the reception device 10 to the content recognition device 20.
- the fingerprint collation unit 24 collates the fingerprint 43 with the fingerprints 53a, 53b, and 53c. Both the fingerprint 43 and the fingerprint 53a include a plurality of common feature quantities such as “018” and “184”. Therefore, the fingerprint collation unit 24 returns information indicating the video content ⁇ 1 corresponding to the fingerprint 53a to the reception device 10 as a result of the image recognition.
- the fingerprint history information DB 25 is a database in which the recognition data 40 received by the content recognition device 20 from the reception device 10 is held in time series (for example, reception order).
- the fingerprint history information DB 25 is stored in a storage device (not shown) such as a memory provided in the content recognition device 20.
- the fingerprint history information DB 25 is updated by adding the recognition data 40.
- the recognition data 40 received by the content recognition apparatus 20 during a predetermined period may be held in the order of reception.
- the predetermined period may be, for example, a period from when the content recognition device 20 receives the recognition data 40 from the reception device 10 to when the image recognition process based on the recognition data 40 ends.
- the content recognition device 20 receives information indicating that image recognition could not be performed when the video content corresponding to the recognition data 40 transmitted from the reception device 10 could not be specified as a result of the image recognition processing. May be sent to or nothing may be sent.
- the content recognition device 20 includes a communication unit (not shown) and communicates with the reception device 10 via the communication unit and the communication network 105. For example, the content recognition device 20 receives the recognition data 40 transmitted from the reception device 10 via the communication unit, and receives the image recognition result based on the received recognition data 40 via the communication unit. 10 to send.
- the advertisement server device 30 is a Web server configured to distribute additional information related to video content transmitted from the broadcasting station 2. As shown in FIG. 1, the advertisement server device 30 includes an additional information DB 31.
- the additional information DB 31 is a database in which information representing video content and additional information are associated with each other for each video content.
- a content ID and additional information are associated with each other.
- the additional information DB 31 is stored in a storage device (for example, HDD) provided in the advertisement server device 30.
- the additional information DB 31 may be stored in a storage device installed outside the advertisement server device 30.
- the additional information is, for example, information indicating an attribute of an item (for example, a product to be advertised) displayed in the video content.
- the additional information is, for example, information about the product such as product specifications, a store (for example, a store address, a URL (Uniform Resource Locator), a telephone number, etc.), a manufacturer, a method of use, and efficacy.
- the fingerprint generation unit 110 is configured to generate a fingerprint based on at least one of a static area and a dynamic area in a frame sequence constituting the video content.
- the fingerprint generation unit 110 can be realized by, for example, an integrated circuit.
- the video extraction unit 12 is configured to extract a plurality of image frames at a predetermined frame rate from the frame sequence constituting the video content. This frame rate is set based on the processing capability and the like in the recognition unit 100.
- the frame rate of video content broadcast from the broadcasting station 2 is 60 fps
- the video extraction unit 12 extracts an example of an image frame at three frame rates of 30 fps, 20 fps, and 15 fps. explain.
- the video extraction unit 12 does not extract image frames at a plurality of frame rates. 5 and 6 only show an example of operation when the extracted frame rates are different. In the example shown in FIGS. 5 and 6, the video extraction unit 12 extracts an image frame at a frame rate of 30 fps, 20 fps, or 15 fps.
- the static region is a region where the amount of change in the image between two image frames is smaller than a predetermined threshold (hereinafter referred to as “first threshold”).
- first threshold a predetermined threshold
- the static area is, for example, a background in an image or an area occupied by a subject with little movement or change.
- the static area is determined by calculating an image change amount between image frames.
- FIG. 5 is a diagram schematically showing an example of the relationship between the image frame of each frame rate extracted by the video extraction unit 12 and the static area in the first embodiment.
- the video content of the broadcast video shown as an example in FIG. 5 is composed of 9 frames of the same scene with no significant change in the video. In the video, two subjects are moving, but the background is not moving.
- the static regions determined at each frame rate are similar to each other and broadcast. Similar to the static region determined in the 60 fps video content.
- the static area determined in the image frame extracted by the video extraction unit 12 and the video content to be broadcasted regardless of the frame rate at which the image frame is extracted is 30 fps, 20 fps, or 15 fps.
- the static region is a region occupied by a background, a subject whose movement or change is small in the image frame, and is a region that is highly likely to exist in the image frame for a predetermined period (for example, several seconds). Therefore, highly accurate recognition is possible by using a static region.
- the content recognition device 20 receives the video content broadcast from the broadcasting station 2, generates a static fingerprint based on the static area in the video content, and stores it in the fingerprint DB 22. Therefore, when the content recognition device 20 receives from the receiving device 10 a static fingerprint generated based on the video content being received by the receiving device 10, the content recognizing device 20 can recognize the video content being received by the receiving device 10. it can.
- the dynamic region is a region where the amount of change in image between two image frames is larger than a predetermined threshold (hereinafter referred to as “second threshold”).
- the dynamic region is, for example, a region where a large image change occurs when the scene is switched.
- FIG. 6 is a diagram schematically illustrating an example of a relationship between an image frame of each frame rate extracted by the video extraction unit 12 in the first embodiment and a dynamic region.
- the video content shown as an example in FIG. 6 includes scene switching.
- the video content shown in FIG. 6 includes first to third scenes that change over time.
- the first scene includes image frames A001 to A003
- the second scene includes image frames A004 to A006
- the third scene includes image frames A007 to A009.
- the dynamic area is determined by calculating the amount of change in the image between image frames.
- the plurality of image frames extracted by the video extraction unit 12 include the image frames of the three scenes. For this reason, when an image change amount is calculated between two temporally adjacent image frames, a large change amount is calculated in the image frames before and after the scene change.
- FIG. 6 shows, as an example, a dynamic region in switching from the first scene to the second scene.
- the switching between the first scene and the second scene is an image frame A003 and an image frame A005. Therefore, at 30 fps in FIG. 6, a dynamic region occurs between the image frame A003 and the image frame A005. Similarly, at 20 fps in FIG. 6, a dynamic region occurs between the image frame A001 and the image frame A004, and at 15 fps in FIG. 6, the dynamic region is between the image frame A001 and the image frame A005. appear.
- the switching between the first scene and the second scene is an image frame A003 and an image frame A004. Therefore, in the broadcast video content, the dynamic area is generated between the image frame A003 and the image frame A004.
- the dynamic area in the 60 fps video content to be broadcast and the dynamic area in each of 30 fps, 20 fps, and 15 fps extracted by the video extraction unit 12 are similar to each other as shown in FIG.
- the dynamic regions determined by each frame rate are similar to each other and broadcast. Similar to the dynamic region determined in 60 fps video content.
- the dynamic region determined based on the image frame extracted by the video extraction unit 12 and the video to be broadcast regardless of whether the frame rate at the time of extracting the image frame is 30 fps, 20 fps, or 15 fps.
- the dynamic region is a region where a large image change has occurred due to a scene change or the like, and is a region where a characteristic image change has occurred. Therefore, highly accurate recognition is possible by using a dynamic region.
- recognition is performed based on characteristic image changes, the number of frames required for recognition can be reduced as compared with the prior art, and recognition processing can be performed at high speed.
- the content recognition device 20 receives the video content broadcast from the broadcast station 2, generates a dynamic fingerprint based on the dynamic region in the video content, and stores it in the fingerprint DB 22. Therefore, when the content recognition device 20 receives a dynamic fingerprint generated based on the video content being received by the reception device 10 from the reception device 10, the content recognition device 20 can recognize the video content being received by the reception device 10. it can.
- FIG. 7 is a block diagram illustrating a configuration example of the fingerprint generation unit 110 according to the first embodiment.
- the fingerprint generation unit 2110 included in the content recognition device 20 has substantially the same configuration and operation as the fingerprint generation unit 110 included in the reception device 10, and thus a duplicate description is omitted.
- the fingerprint generation unit 110 includes an image acquisition unit 111 and a data generation unit 112.
- the image acquisition unit 111 acquires a plurality of image frames extracted by the video extraction unit 12.
- the data generation unit 112 generates a fingerprint as recognition data based on a change in image between frames of a plurality of image frames acquired by the image acquisition unit 111.
- the static fingerprint is generated based on an area (hereinafter referred to as “static area”) in which the amount of change in the image between frames is smaller than a preset threshold (hereinafter referred to as “first threshold”). It is a fingerprint.
- the dynamic fingerprint is generated based on an area (hereinafter referred to as “dynamic area”) in which an image change amount between frames is larger than a preset threshold (hereinafter referred to as “second threshold”). It is a fingerprint.
- the recognition data includes at least one of a static fingerprint and a dynamic fingerprint. Depending on the value of the first threshold and the value of the second threshold, both the static fingerprint and the dynamic fingerprint may not be generated. In this case, the recognition data does not include either a static fingerprint or a dynamic fingerprint.
- the data generation unit 112 includes a scale conversion unit 210, a difference calculation unit 220, a determination unit 230, and a generation unit 240.
- the scale conversion unit 210 performs scale conversion on each of the plurality of image frames acquired by the image acquisition unit 111. Specifically, the scale conversion unit 210 performs grayscale conversion and downscale conversion on each image frame.
- Grayscale conversion is to convert a color image into a grayscale image.
- the scale converter 210 converts the color information that each pixel of the image frame has into a luminance value, thereby converting the color image into a grayscale image.
- the present disclosure does not limit this conversion technique.
- the scale conversion unit 210 may extract one element of RGB from each pixel and convert it to the luminance value of the corresponding pixel.
- the luminance value is a numerical value indicating the brightness of the pixel and is an example of the pixel value.
- the scale conversion unit 210 may calculate the luminance value using an NTSC weighted average method or a simple average method.
- Down-scale conversion is conversion of the number of pixels constituting one image frame from the original number of pixels to a smaller number of pixels.
- the scale conversion unit 210 performs downscale conversion, and converts the image of the image frame into an image configured with a smaller number of pixels.
- the present disclosure does not limit this conversion technique.
- the scale conversion unit 210 may perform the downscale conversion by dividing each image into a plurality of blocks each including a plurality of pixels and calculating one numerical value for each block.
- the scale conversion unit 210 may calculate an average value, an intermediate value, or the like of the luminance value for each block, and use it as a numerical value representing the brightness of the block.
- the scale conversion unit 210 performs both grayscale conversion and downscale conversion, but the present disclosure is not limited to this configuration.
- the scale converter 210 may perform only one of these, or may not perform either. That is, the data generation unit 112 may not include the scale conversion unit 210.
- the difference calculation unit 220 generates an image change frame from each of the plurality of image frames acquired by the image acquisition unit 111.
- the image change frame is generated by calculating a difference in luminance value between two temporally adjacent image frames (for example, two temporally continuous image frames). Therefore, the image change frame indicates the amount of change in luminance value between two temporally adjacent image frames (hereinafter referred to as “luminance change value”).
- luminance change value is an example of the pixel change value, and is a value indicating the amount of change in the luminance value, which is an example of the pixel value.
- the difference calculation unit 220 generates an image change frame using the image frame that has been subjected to grayscale conversion and downscale conversion by the scale conversion unit 210.
- the determining unit 230 includes a static region determining unit 231 and a dynamic region determining unit 232.
- the determination unit 230 compares the absolute value of each luminance change value of the image change frame generated by the difference calculation unit 220 with the first threshold value and the second threshold value. Then, at least one of a static region where the absolute value of the luminance change value is smaller than the first threshold and a dynamic region where the absolute value of the luminance change value is larger than the second threshold is determined. Specifically, the determination unit 230 calculates the absolute value of each luminance change value of the image change frame, determines whether the absolute value is smaller than the first threshold value, and the absolute value is smaller than the second threshold value.
- the static area and the dynamic area are determined by executing the determination of whether the area is large or not.
- the absolute value of the luminance change value may be calculated by the difference calculation unit 220.
- the first threshold value and the second threshold value are set based on predetermined numerical values, and are determined based on a possible range of the luminance change value. For example, the first threshold value and the second threshold value are determined in the range of 0% to 20% of the maximum absolute value of the luminance change value. As a specific example, when the maximum value of the absolute value of the luminance change value is 255, the first threshold is “1” and the second threshold is “20”. These numerical values are merely examples. Each threshold is preferably set appropriately. The first threshold value and the second threshold value may be the same numerical value, or may be different numerical values. The second threshold value is preferably a numerical value larger than the first threshold value, but the second threshold value may be a numerical value smaller than the first threshold value.
- the static region determining unit 231 included in the determining unit 230 compares each absolute value of the luminance change value of the image change frame with the first threshold value, and determines whether the absolute value is smaller than the first threshold value. Determine the static area. For example, when the first threshold value is “1”, the static area determination unit 231 sets an area having a luminance change value of “0” as a static area.
- the region where the luminance change value is “0” is a region where the luminance value does not substantially change between two temporally adjacent image frames.
- the dynamic region determination unit 232 included in the determination unit 230 compares each absolute value of the luminance change value of the image change frame with the second threshold value, and determines whether the absolute value is greater than the second threshold value. Determine the dynamic region. For example, when the second threshold value is “20”, the dynamic region determination unit 232 sets a region where the absolute value of the luminance change value is “21” or more as the dynamic region. A region where the absolute value of the luminance change value is “21” or more is a region where the luminance value has changed by 21 or more between two temporally adjacent image frames.
- the static region determination unit 231 and the dynamic region determination unit 232 determine the absolute value of the luminance change value of the image change frame based on the image frame that has been subjected to grayscale conversion and downscale conversion by the scale conversion unit 210. Used for.
- the generation unit 240 includes a static fingerprint generation unit 241 and a dynamic fingerprint generation unit 242.
- the static fingerprint generation unit 241 determines whether or not the static region output from the static region determination unit 231 occupies a predetermined ratio (hereinafter referred to as “first ratio”) in the image change frame or more. Determine whether. And the static fingerprint production
- the static fingerprint generation unit 241 determines the static fingerprint when the range of the static area in the image change frame is large, in other words, when the image change between two temporally adjacent image frames is small. Is generated.
- the static fingerprint generation unit 241 generates a static frame by filtering one of the two image frames used for generating the image change frame with a static region. This filtering will be described later. Then, the static fingerprint generation unit 241 sets the generated static frame as a static fingerprint.
- the static frame includes the luminance value of one static region of the two image frames used for generating the image change frame, and the luminance value of the region other than the static region is a constant value (for example, “0”). Is a frame. Details of the static frame will be described later.
- the dynamic fingerprint generation unit 242 determines whether the dynamic region output from the dynamic region determination unit 232 occupies a predetermined ratio (hereinafter, “second ratio”) in the image change frame or more. Determine whether. And the dynamic fingerprint production
- the dynamic fingerprint generation unit 242 performs the dynamic fingerprint when the range of the dynamic region occupied in the image change frame is large, in other words, when there is a large image change between two temporally adjacent image frames. Is generated.
- the dynamic fingerprint generation unit 242 generates a dynamic frame by filtering the image change frame with a dynamic region. This filtering will be described later. Then, the dynamic fingerprint generation unit 242 sets the generated dynamic frame as a dynamic fingerprint.
- the dynamic frame is a frame that includes the luminance value of the dynamic region of the image change frame and has a constant luminance value (for example, “0”) in a region other than the dynamic region. Details of the dynamic frame will be described later.
- predetermined numerical values are set for the first ratio and the second ratio.
- the first ratio and the second ratio are determined in the range of 20% to 40%.
- the first ratio and the second ratio are each 30%. These numerical values are merely examples. It is desirable that the first ratio and the second ratio are set appropriately.
- the first ratio and the second ratio may be the same numerical value or may be different numerical values.
- the fingerprint generation unit 110 generates either a static fingerprint or a dynamic fingerprint for each image frame. Or neither. That is, when N image frames are acquired from the video content, the fingerprint generation unit 110 generates a fingerprint including a maximum of N ⁇ 1 in total including the static fingerprint and the dynamic fingerprint.
- each of the static fingerprints generated in the same continuous scene is likely to be similar to each other. Therefore, when a plurality of continuous image frames reflect the same scene, the static fingerprint generation unit 241 selects one static fingerprint from a plurality of static fingerprints generated from the same scene. May be selected and output.
- a relatively heavy processing such as contour detection is required for matching image frames.
- a fingerprint is generated based on an image change between image frames. Detection of an image change between image frames can be performed by a relatively light process such as calculation of a difference. That is, the fingerprint generation unit 110 according to the present embodiment can generate a fingerprint by a process with a relatively light load. The same applies to the fingerprint generation unit 2110.
- FIG. 8 is a flowchart showing an operation example of the content recognition apparatus 20 provided in the content recognition system 1 according to the first embodiment.
- the content receiving unit 21 receives video content from the broadcast station 2 (step S1).
- the content receiving unit 21 receives a plurality of video contents being broadcast from a plurality of broadcasting stations 2 before the receiving device 10 receives the contents.
- the content receiving unit 21 may receive in advance video content before being broadcast. As described above, the content receiving unit 21 receives each of a plurality of video contents as video content candidates.
- the fingerprint generation unit 2110 generates recognition data (step S2).
- the fingerprint generation unit 2110 generates a fingerprint corresponding to each of the plurality of video content candidates received by the content reception unit 21. Details of the fingerprint generation will be described later with reference to FIG.
- the fingerprint generation unit 2110 stores the recognition data generated in step S2 in the fingerprint DB 22 (step S3).
- the fingerprint generation unit 2110 associates the fingerprint 53 generated in step S ⁇ b> 2 with the timing information 51 and the type information 52, and the associated data Is stored in the fingerprint DB 22 as the recognition data 50.
- the fingerprint generation unit 2110 may store the fingerprint 53 for each broadcast station 2, for example. Further, content information used for profile filtering may be stored in association with the fingerprint 53.
- the content recognition device 20 determines whether or not the recognition data 40 has been received from the reception device 10 (step S4).
- the content recognition device 20 determines that the recognition data 40 has not been received from the reception device 10 (No in step S4), the content recognition device 20 returns to step S1 and executes the processing from step S1 onward. The content recognition device 20 repeats the processes in steps S1 to S3 until the recognition data 40 is received from the reception device 10.
- the fingerprint filtering unit 23 filters the recognition data 50 stored in the fingerprint DB 22 (Ste S5). Details of the filtering will be described later with reference to FIG.
- the fingerprint collation unit 24 collates the recognition data 50 received from the receiving device 10 in step S4 against the recognition data 50 filtered in step S5 (step S6). Details of the collation will be described later with reference to FIG.
- the fingerprint collation unit 24 determines whether or not collation in step S6 has been completed (step S7).
- Step S6 the content recognition apparatus 20 returns to Step S1 and executes the processes after Step S1.
- step S6 When the fingerprint collation unit 24 has collated in step S6 (Yes in step S7), the content recognition apparatus 20 transmits the collation result in step S6 (that is, the image recognition result) to the receiving apparatus 10 (step S8). ).
- the receiving device 10 can receive the image recognition result from the content recognition device 20 and execute processing such as superimposed display of additional information based on the received image recognition result.
- step S8 it is determined by the content recognition device 20 whether or not the video content recognition process is to be terminated (step S9).
- the present disclosure does not limit the determination method in step S9. For example, when the recognition data 40 is not transmitted from the receiving device 10 for a predetermined period, the content recognition device 20 may be set to make a Yes determination in step S9. Alternatively, when information indicating the end of processing is received from the receiving device 10, the content recognition device 20 may be set to make a Yes determination in step S9.
- step S 9 When new recognition data 40 is transmitted from the receiving device 10 to the content recognizing device 20, the content recognizing device 20 does not end the video content recognizing process (No in step S 9), returns to step S 1, and subsequent steps Repeat a series of processes.
- the video content recognition process ends (Yes in step S9), the content recognition apparatus 20 ends the process related to image recognition.
- FIG. 9 is a flowchart showing an example of processing when generating recognition data in the first embodiment.
- the flowchart of FIG. 9 shows an overview of the process executed in step S2 of FIG.
- the fingerprint generation unit 2110 calculates an image change amount between image frames for each image frame of a plurality of video content candidates received by the content reception unit 21 in step S1 (step S20). Details of the calculation of the image change amount will be described later with reference to FIGS.
- the fingerprint generation unit 110 included in the receiving device 10 calculates the amount of image change between image frames from the plurality of image frames extracted by the video extraction unit 12 (step S20). This is the difference between the fingerprint generation unit 110 and the fingerprint generation unit 2110. However, except for this point, the fingerprint generation unit 110 and the fingerprint generation unit 2110 perform substantially the same operation.
- the fingerprint generation unit 2110 generates a static fingerprint (step S21).
- Fingerprint generation unit 2110 determines a static region based on the image change frame, and generates a static fingerprint based on the determined static region. Details of generation of the static fingerprint will be described later with reference to FIGS. 14 and 15.
- the fingerprint generation unit 2110 generates a dynamic fingerprint (step S22).
- Fingerprint generation unit 2110 determines a dynamic region based on the image change frame, and generates a dynamic fingerprint based on the determined dynamic region. Details of generation of the dynamic fingerprint will be described later with reference to FIGS.
- step S21 may be executed first, or may be executed in parallel with each other.
- FIG. 10 is a diagram schematically illustrating an example of a change in an image frame in the process of generating recognition data according to the first embodiment.
- FIG. 10 a plurality of image frames (a) included in the video content candidates acquired by the content receiving unit 21 in step S1, an image frame (b) subjected to gray scale conversion in step S200 described later, The image frame (c) downscaled in step S201, the amount of change (d) calculated in step S202 described later, and the fingerprint (e) generated in steps S21 and S22 are schematically shown. Shown in
- each of the image frames A001 to A009 is included in any of the three scenes of the first scene to the third scene.
- Image frames A001 to A003 are included in the first scene
- image frames A004 to A006 are included in the second scene
- image frames A007 to A009 are included in the third scene.
- the image frames A001 to A009 are so-called color images and include color information.
- gray scale conversion is performed on the image frame (b) of FIG. 10 in step S200 of FIG. An example of when.
- the color information included in the image frames A001 to A009 is converted into luminance values for each pixel.
- step S201 of FIG. 10 shows an example when thereby, the number of pixels constituting the image frame is reduced.
- an image frame (c) in FIG. 10 shows an example when one image frame is divided into 25 blocks of 5 blocks ⁇ 5 blocks. In other words, the number of pixels constituting one image frame is downscaled to 25.
- the luminance value of each block shown in the image frame (c) of FIG. 10 is calculated from the luminance values of a plurality of pixels constituting each block.
- the luminance value of each block can be calculated by calculating an average value or an intermediate value of the luminance values of a plurality of pixels constituting the block for each block.
- the shade of each block corresponds to the magnitude of the luminance value. The larger the luminance value, the darker the block, and the smaller the luminance value, the thinner the block.
- step S202 one image change frame is generated by calculating the amount of change in luminance value (ie, luminance change value) between two temporally adjacent image frames.
- an image change frame B001 is generated from the downscaled image frame A001 and image frame A002.
- the shade of each block constituting the image change frame is the luminance change value of the image change frame, that is, the change amount of the luminance value between the two image frames subjected to downscale conversion. It corresponds to. The larger the amount of change in luminance value, the darker the block, and the smaller the amount of change in luminance value, the thinner the block.
- the image change frame B001 and the image change frame B002 are both generated from the image frames A001 to A003 included in the same scene. For this reason, the image change frame B001 and the image change frame B002 are similar to each other. Therefore, in step S21, one static fingerprint C002 can be generated from the image change frame B001 and the image change frame B002. The same applies to the image change frame B004 and the image change frame B005, and the image change frame B007 and the image change frame B008.
- the image change frame B003 is generated from the two image frames A003 and A004 that are switched scenes. Accordingly, in step S22, one dynamic fingerprint D003 can be generated from the image change frame B003. The same applies to the image change frame B006.
- the fingerprints of the video content generated from the image frames A001 to A009 in this way are three static fingerprints C002, C005, C008, and two dynamic fingerprints D003, D006. Have.
- the fingerprint of the generated video content includes at least two of one or more static fingerprints and one or more dynamic fingerprints.
- the fingerprint of the video content may consist of only two or more static fingerprints, may consist of only two or more dynamic fingerprints, or may include one or more static fingerprints and one or more It may be configured with a dynamic fingerprint.
- the shade of each block constituting the static fingerprint or the dynamic fingerprint corresponds to the magnitude of the luminance value of the block.
- FIG. 11 is a flowchart illustrating an example of a process for calculating the amount of change between image frames in the first embodiment.
- the flowchart in FIG. 11 shows an overview of the processing executed in step S20 in FIG.
- FIG. 12 is a diagram schematically illustrating an example of the image frame downscale conversion processing according to the first embodiment.
- FIG. 13 is a diagram schematically illustrating an example of a calculation process of a change amount between image frames in the first embodiment.
- the scale conversion unit 210 illustrated in FIG. 7 performs grayscale conversion on the plurality of extracted image frames (step S200).
- the scale conversion unit 210 converts one of the plurality of extracted image frames and an image frame temporally adjacent to the image frame to gray scale.
- one extracted image frame is referred to as “frame 91”
- an image frame temporally adjacent to frame 91 is referred to as “frame 92”.
- the scale conversion unit 210 converts the color information of the frames 91 and 92 into luminance values based on, for example, the NTSC weighted average method.
- an image frame immediately after the frame 91 is a frame 92.
- the frame 92 may be an image frame immediately before the frame 91.
- the frame 92 may be an image frame two or more frames before the frame 91, or may be an image frame two or more frames after the frame 91.
- the scale conversion unit 210 performs downscale conversion of the two image frames subjected to grayscale conversion (step S201).
- FIG. 12 shows an example in which the image frames A003 and A004 are downscaled.
- the image frame A003 corresponds to the frame 91
- the image frame A004 corresponds to the frame 92.
- the scale conversion unit 210 divides the image frame A003 into 25 blocks of 5 blocks ⁇ 5 blocks.
- each block includes 81 pixels of 9 pixels ⁇ 9 pixels.
- the upper left block of the image frame A003 includes 81 pixels having luminance values such as “77” and “95” as shown in FIG. Note that these numerical values are merely examples, and the present disclosure is not limited to these numerical values.
- the scale conversion unit 210 calculates a luminance value representing a block by, for example, calculating an average value of luminance values of a plurality of pixels included in each block for each block.
- a value of “103” is calculated by calculating an average value of the luminance values of 81 pixels constituting the upper left block of the image frame A003.
- the value (average value) calculated in this way is a luminance value representing the upper left block.
- the scale conversion unit 210 calculates the luminance value representing each block for each of all the blocks constituting the image frame A003.
- the number of pixels constituting the image frame can be converted (that is, downscaled) into the number of blocks.
- an image frame having the number of pixels of 45 pixels ⁇ 45 pixels is downscaled to an image frame composed of 25 blocks of 5 blocks ⁇ 5 blocks.
- the image frame having the number of pixels of 45 pixels ⁇ 45 pixels is downscaled to the image frame having the number of pixels of 5 pixels ⁇ 5 pixels.
- the image frame A003 after downscale conversion is composed of 25 blocks including average values such as “103” and “100”.
- the image frame A003 after downscale conversion is composed of 25 pixels having luminance values such as “103” and “100”.
- the image frame A004 is similarly downscaled.
- each block constituting the image frame after the downscale conversion is expressed as “pixel”, and the average value of the luminance calculated for each block is expressed as “the pixel of the image frame after the downscale conversion. It may be expressed as “luminance value”.
- the difference calculation unit 220 illustrated in FIG. 7 calculates a difference in luminance value between the frame 91 and the frame 92 after downscale conversion, and is configured with a difference in luminance value (that is, a luminance change value).
- the changed image change frame is generated (step S202).
- the difference calculation unit 220 calculates the luminance value of each pixel that configures the frame 91 after downscale conversion and the luminance value of each pixel that configures the frame 92 after downscale conversion. Each difference is calculated. At this time, the difference calculation unit 220 calculates a difference in luminance value between pixels at the same position. For example, the difference calculation unit 220 subtracts the upper left luminance value “89” of the image frame A 004 from the upper left luminance value “103” of the image frame A 003 to obtain the upper left luminance change value “14” of the image change frame B 003. Is calculated.
- the difference calculation unit 220 calculates a difference in luminance value for all pixels (that is, all blocks) between two image frames after downscale conversion, and generates an image change frame.
- an image change frame B003 is generated from the image frames A003 and A004 after downscale conversion.
- FIG. 14 is a flowchart illustrating an example of a static fingerprint generation process according to the first embodiment.
- the flowchart in FIG. 14 shows an overview of the processing executed in step S21 in FIG.
- FIG. 15 is a diagram schematically illustrating an example of a static fingerprint generated based on the amount of change between image frames in the first embodiment.
- the static area determination unit 231 shown in FIG. 7 determines a static area (step S210).
- the static area determination unit 231 calculates the absolute value of the luminance change value of the image change frame and compares the absolute value with the first threshold value. Then, it is determined whether or not the absolute value of the luminance change value is smaller than the first threshold value, and an area where the absolute value of the luminance change value is smaller than the first threshold value is set as a static area. In this way, the static area is determined.
- the absolute value of the luminance change value is a change amount of the luminance value between two temporally adjacent image frames.
- the static area determination unit 231 determines that the area where the luminance change value of the image change frame is “0”, that is, two temporally adjacent image frames. A region in which the luminance value does not substantially change between them is defined as a static region. In the case of this setting, in the example shown in FIG. 15, 13 blocks in which “0” is entered as the luminance change value in the image change frame B002 are static areas.
- the static fingerprint generation unit 241 illustrated in FIG. 7 filters the frame 91 in the static area determined in step S210 to generate a static frame (step S211).
- This filtering is to perform the following processing on the luminance value of each block constituting the frame 91.
- the luminance value of the block of the frame 91 corresponding to the static area is used as it is, and the luminance value is set to a constant value (for example, “0”) for blocks other than the static area.
- the static frame generated by filtering the frame 91 is the static frame C002 in the example shown in FIG.
- the luminance value of the frame 91 is used as it is for the block (static area) whose luminance change value is “0” in the image change frame B002, and the luminance value of the block other than the static area is the same. It becomes “0”.
- the static fingerprint generation unit 241 illustrated in FIG. 7 calculates the ratio of the static area determined in step S210 and compares it with the first ratio, and the ratio of the static area is equal to or greater than the first ratio. It is determined whether or not there is (step S212).
- the static fingerprint generation unit 241 calculates the ratio of the static area based on the number of blocks determined as the static area in step S210 with respect to the total number of blocks constituting the image change frame.
- the ratio of the static area is 52%. Therefore, if the first ratio is, for example, 30%, it is determined as Yes in step S212 in the example illustrated in FIG.
- step S212 When it is determined in step S212 that the ratio of the static area is equal to or higher than the first ratio (Yes in step S212), the static fingerprint generation unit 241 uses the static frame generated in step S211 as the static fingerprint. (Step S213).
- the static frame C002 is stored in the fingerprint DB 22 of the content recognition device 20 as the static fingerprint C002.
- the static fingerprint is stored in a storage device (for example, an internal memory of the recognition unit 100 or the like not shown) included in the receiving device 10.
- step S212 When it is determined in step S212 that the ratio of the static area is less than the first ratio (No in step S212), the static fingerprint generation unit 241 does not store the static frame generated in step S211. Discard (step S214). Therefore, when it is determined No in step S212, a static fingerprint is not generated.
- the value (for example, “103”, “100”, etc.) of each block of the static fingerprint C002 is the fingerprint feature amount.
- the fingerprint generation unit 2110 generates a static fingerprint based on whether or not the static region is larger than the first ratio in the image frame. That is, the fingerprint generation unit 2110 can generate a static fingerprint by appropriately extracting a background, a region with little movement / change, and the like from an image frame.
- step S212 is executed.
- step S211 is executed to generate a static frame
- step S213 the static frame is generated.
- the order of each process may be set so that the frame is saved as a static fingerprint.
- FIG. 16 is a flowchart illustrating an example of a dynamic fingerprint generation process according to the first embodiment.
- the flowchart in FIG. 16 shows an overview of the processing executed in step S22 in FIG.
- FIG. 17 is a diagram schematically illustrating an example of an image frame in which a dynamic fingerprint is not generated in the first embodiment.
- FIG. 18 is a diagram schematically illustrating an example of a dynamic fingerprint generated based on the amount of change between image frames in the first embodiment.
- the dynamic area determination unit 232 shown in FIG. 7 determines a dynamic area (step S220).
- the dynamic region determination unit 232 calculates the absolute value of the luminance change value of the image change frame, and compares the absolute value with the second threshold value. Then, it is determined whether or not the absolute value of the luminance change value is larger than the second threshold, and a region where the absolute value of the luminance change value is larger than the second threshold is set as a dynamic region. In this way, the dynamic area is determined.
- a block whose absolute value of the luminance change value is “21” or more in the image change frame is a dynamic region.
- two blocks in which a numerical value of “21” or more or “ ⁇ 21” or less is entered as the luminance change value in the image change frame B002 are dynamic regions.
- eleven blocks in which numerical values of “21” or more or “ ⁇ 21” or less are entered as the luminance change values in the image change frame B003 are dynamic regions.
- the dynamic fingerprint generation unit 242 illustrated in FIG. 7 filters the image change frame in the dynamic region determined in step S220 to generate a dynamic frame (step S221).
- This filtering is to perform the following processing on the luminance change value of each block constituting the image change frame.
- the luminance change value of the block corresponding to the dynamic area is used as it is, and the luminance change value is set to a constant value (for example, “0”) for blocks other than the dynamic area. To do.
- the dynamic frame generated by filtering the image change frame is the dynamic frame D002 in the example shown in FIG. 17, and the dynamic frame D003 in the example shown in FIG.
- the luminance change values of the image change frames B002 and B003 for the blocks (dynamic areas) in which the luminance change value is “21” or more or “ ⁇ 21” or less in the image change frames B002 and B003. Are used as they are, and the luminance change value is “0” for blocks other than the dynamic region.
- step S220 and step S221 for the image change frame is executed as a batch process, for example, replacing the luminance change value with “0” for a block whose absolute value of the luminance change value is equal to or smaller than the second threshold. can do.
- the dynamic fingerprint generation unit 242 calculates the ratio of the dynamic area determined in step S220 and compares it with the second ratio, and determines whether the ratio of the dynamic area is equal to or greater than the second ratio. Determination is made (step S222).
- the dynamic fingerprint generation unit 242 calculates the ratio of the dynamic area based on the number of blocks determined as the dynamic area in step S220 with respect to the total number of blocks constituting the image change frame.
- the ratio of the dynamic area is 8%.
- the total number of blocks constituting the image change frame is 25, and the number of blocks in the dynamic area is 11, so the ratio of the dynamic area is 44%. Therefore, if the second ratio is 30%, for example, it is determined No in step S222 in the example illustrated in FIG. 17, and Yes is determined in step S222 in the example illustrated in FIG.
- step S222 When it is determined in step S222 that the proportion of the dynamic area is equal to or greater than the second proportion (Yes in step S222), the dynamic fingerprint generation unit 242 uses the dynamic frame generated in step S221 as the dynamic fingerprint. (Step S223).
- the dynamic fingerprint generation unit 242 discards the dynamic frame generated in step S221 without saving it. (Step S224). Therefore, when it is determined No in step S222, a dynamic fingerprint is not generated.
- the dynamic frame D003 determined as Yes in step S222 is stored in the fingerprint DB 22 included in the content recognition device 20 as the dynamic fingerprint D003.
- the dynamic fingerprint is stored in a storage device (for example, an internal memory of the recognition unit 100 or the like not shown) included in the receiving device 10.
- the dynamic frame D002 determined No in step S222 is discarded without being saved.
- the value (for example, “0”, “24”, etc.) of each block of the dynamic fingerprint D003 is the fingerprint feature amount.
- the fingerprint generation unit 2110 generates a dynamic fingerprint based on whether or not the dynamic region is larger than the second ratio in the image frame. That is, the fingerprint generation unit 2110 can generate a dynamic fingerprint by appropriately using an image frame for a region where a large image change has occurred due to a scene change or the like.
- step S222 is executed.
- a dynamic frame is generated by executing step S221, and the dynamic frame is generated in subsequent step S223.
- the order of each process may be set so that the frame is saved as a dynamic fingerprint.
- FIG. 19 is a flowchart illustrating an example of filtering processing executed by the fingerprint filtering unit 23 according to the first embodiment.
- the flowchart in FIG. 19 shows an overview of the processing executed in step S5 in FIG.
- the fingerprint filtering unit 23 of the content recognition device 20 sequentially performs region filtering, profile filtering, property filtering, and property sequence filtering, and selects a matching target in the image recognition processing. Narrow down.
- the fingerprint filtering unit 23 performs region filtering processing on the video content candidates stored in the fingerprint DB 22 using the geographic information included in the attached information (step S50).
- the region filtering process is a process of excluding video content candidates broadcast from the broadcasting station 2 that cannot be viewed in the area indicated by the geographic information from the verification targets in the image recognition process.
- the fingerprint filtering unit 23 executes a region filtering process based on the geographical information. Therefore, since the content recognition apparatus 20 can narrow down the fingerprint 53 read from fingerprint DB22 as a collation target in an image recognition process, it can reduce the process concerning image recognition.
- the fingerprint filtering unit 23 reads the fingerprint 53 to be read from the fingerprint DB 22 as a collation target in the image recognition process. Narrow down to fingerprints 53 of video content candidates transmitted from the broadcasting station 2 that can be received in Tokyo.
- the fingerprint filtering unit 23 executes a profile filtering process using the user information included in the attached information (step S51).
- the profile filtering process is a process of excluding video content candidates that do not match the user characteristics indicated by the user information from the collation targets in the image recognition process.
- the fingerprint filtering unit 23 executes the profile filtering process based on the user information. Thereby, since the content recognition apparatus 20 can narrow down the fingerprints 53 to be collated in the image recognition process, the process for image recognition can be reduced.
- the fingerprint filtering unit 23 selects video content candidates that do not target the age group (for example, infants as the viewing target). Infant programs, etc.) are excluded from comparison targets in image recognition.
- the fingerprint filtering unit 23 executes property filtering processing (step S52).
- the property filtering process is a process for comparing the type information 52 of the fingerprint 53 included in the recognition data 50 of the video content candidate and the type information 42 included in the recognition data 40 transmitted from the receiving device 10. .
- the fingerprint filtering unit 23 executes the property filtering process to obtain the recognition data 50 that does not include the fingerprint 53 of the type information 52 of the same type as the type information 42 included in the recognition data 40 transmitted from the receiving device 10. Then, the recognition data 50 including the fingerprint 53 of the type information 52 of the same type as the type information 42 is excluded from the verification target in the image recognition process, and is selected as a verification target candidate in the image recognition process. Thereby, since the content recognition apparatus 20 can narrow down the fingerprints 53 to be collated in the image recognition process, the process for image recognition can be reduced.
- the type information 42 included in the recognition data 40 transmitted from the receiving device 10 includes “A type” and “B type”. Therefore, the fingerprint filtering unit 23 excludes the recognition data 50b having the type information 52b of “A type” from the collation target in the image recognition process.
- the fingerprint filtering unit 23 executes a property sequence filtering process (step S53).
- the property sequence filtering process is a process of comparing the arrangement order of the type information 52 included in the recognition data 50 of the video content candidate with the arrangement order of the type information 42 included in the recognition data 40 transmitted from the receiving device 10. That is.
- the arrangement order of the type information 42 and 52 may be set based on the time indicated by the timing information 41 and 51. In this case, the arrangement order is determined based on the generation order of the fingerprints 43 and 53.
- the fingerprint filtering unit 23 executes the property sequence filtering process to recognize that the type information 52 arranged in the same order as the type information 42 included in the recognition data 40 transmitted from the receiving device 10 is not included.
- the data 50 is excluded from the verification target in the image recognition process, and the recognition data 50 including the type information 52 arranged in the same order as the type information 42 is selected as the verification target in the image recognition process.
- the fingerprint filtering unit 23 excludes the recognition data 50c including the type information 52c arranged as “A type, A type, B type” from the collation target in the image recognition process, and “A type, B type,
- the recognition data 50a including the type information 52a arranged as “A type” is set as a collation target in the image recognition processing.
- the fingerprint filtering unit 23 outputs the recognition data 50 of the video content candidates selected by the filtering processes in steps S50 to S53 to the fingerprint collating unit 24 (step S54).
- the fingerprint collation unit 24 of the content recognition device 20 collates the recognition data 50 selected from the fingerprint filtering unit 23 with the recognition data 40 transmitted from the reception device 10 and executes an image recognition process. Then, the content recognition device 20 transmits the result of the image recognition processing (that is, information representing video content candidates including the recognition data 50 corresponding to the recognition data 40) to the reception device 10.
- the content recognition device 20 can narrow down the recognition data 50 to be collated in the image recognition process by the filtering performed by the fingerprint filtering unit 23. As a result, processing related to video content recognition (image recognition) can be reduced.
- the fingerprint filtering unit 23 executes the profile filtering process of step S51 on the recognition data 50 selected by the region filtering process of step S50, and the recognition data 50 selected by the process is displayed.
- An example of an operation in which the property filtering process in step S52 is executed and the property sequence filtering process in step S53 is executed on the recognition data 50 selected by the process has been shown. It is not limited. The order of the filtering processes may be changed. Alternatively, each filtering process may be executed independently.
- FIG. 20 is a flowchart illustrating an example of the property filtering process executed by the fingerprint filtering unit 23 according to the first embodiment.
- FIG. 21 is a diagram schematically illustrating a specific example of the property filtering process executed by the fingerprint filtering unit 23 according to the first embodiment.
- the fingerprint filtering unit 23 acquires the type information 42 and the timing information 41 from the recognition data 40 transmitted from the receiving device 10 (step S520).
- the recognition data 40 transmitted from the receiving device 10 includes timing information 41, type information 42, and a fingerprint 43.
- the recognition data 40 transmitted from the receiving device 10 includes “A type” as the type information 42 and “07/07/2014 03: 32: 36.125” as the timing information 41.
- the fingerprint filtering unit 23 acquires these pieces of information in step S520.
- the fingerprint filtering unit 23 reads and acquires a plurality of recognition data 50 from the fingerprint DB 22 (step S521).
- the fingerprint filtering unit 23 acquires the recognition data 50 selected by the region filtering process in step S50 and the profile filtering process in step S51.
- the recognition data 50 includes timing information 51, type information 52, and a fingerprint 53.
- the fingerprint filtering unit 23 applies to each of the video contents ⁇ 1, ⁇ 1, ⁇ 1, ⁇ 1, and ⁇ 1 transmitted from each broadcasting station ⁇ , ⁇ , ⁇ , ⁇ , and ⁇ and accumulated in the fingerprint DB 22.
- the operation example which acquires the corresponding recognition data 50 from fingerprint DB22 is shown.
- video contents ⁇ 1, ⁇ 1, ⁇ 1, ⁇ 1, and ⁇ 1 are video content candidates.
- the fingerprint filtering unit 23 is generated at the recognition data 50 having the timing information 51 near the time indicated by the timing information 41 received from the receiving device 10 (that is, generated at a time close to the generation time of the fingerprint 43).
- the recognition data 50) having the fingerprint 53 is acquired from the fingerprint DB 22.
- the fingerprint filtering unit 23 determines that the time indicated by the timing information 51 is the time indicated by the timing information 41 received from the receiving device 10 and the time preceding that time by a predetermined time (for example, 3 seconds).
- the recognition data 50 included between the two may be acquired from the fingerprint DB 22.
- FIG. 21 shows an operation example in which the predetermined time is set to 2.5 seconds.
- the time indicated by the timing information 41 is “07/07/2014 03: 32: 36.125”.
- the fingerprint filtering unit 23 calculates 33.625 seconds by subtracting 2.5 seconds from 36.125 seconds.
- the recognition data 50 included in the time range indicated by the timing information 51 in the range from “07/07/2014 03: 32: 33.625” to “07/07/2014 03: 32: 36.125” is fingered. Obtained from the print DB 22.
- the fingerprint filtering unit 23 uses the type information 42 acquired from the receiving device 10 and the type information 52 included in the recognition data 50 to select recognition data 50 candidates to be collated in the image recognition process. (Step S522).
- the fingerprint filtering unit 23 excludes the recognition data 50 that does not include the type information 52 of the same type as the type information 42 acquired from the receiving device 10 from the collation target in the image recognition processing, and the type information 42
- the recognition data 50 including the same type information 52 is set as a candidate for a collation target in the image recognition process.
- the fingerprint filtering unit 23 acquires the type information 42 of “A type” from the receiving device 10. Accordingly, the fingerprint filtering unit 23 selects the recognition data 50 including the type information 52 of “A type” as a candidate for collation in the image recognition process, and does not include the type information 52 of “A type”. The recognition data 50 including only the type information 52 is excluded from the comparison target. In the example illustrated in FIG. 21, the recognition data 50 of the video content ⁇ 1, the video content ⁇ 1, and the video content ⁇ 1 are selected as the candidates for collation in the image recognition process, and the recognition data of the video content ⁇ 1 and the video content ⁇ 1. 50 is excluded from the comparison target.
- FIG. 22 is a flowchart illustrating an example of the property sequence filtering process executed by the fingerprint filtering unit 23 according to the first embodiment.
- FIG. 23 is a diagram schematically illustrating a specific example of the property sequence filtering process executed by the fingerprint filtering unit 23 in the first embodiment.
- the fingerprint filtering unit 23 acquires the past recognition data 40 received by the content recognition device 20 from the reception device 10 from the fingerprint history information DB 25 (step S530).
- the fingerprint filtering unit 23 receives new recognition data 40 (denoted as “recognition data 40n”) from the receiving apparatus 10, the past recognition including the recognition data 40 received immediately before the recognition data 40n.
- Data 40 is read from the fingerprint history information DB 25 and acquired.
- the recognition data 40 includes a fingerprint 43 and type information 42 and timing information 41 corresponding to the fingerprint 43.
- the fingerprint filtering unit 23 receives new recognition data 40n from the receiving device 10
- the latest three recognition data 40 accumulated in the fingerprint history information DB 25 (in FIG. (Data 40a ",” Recognition data 40b “, and” Recognition data 40c ") are acquired from the fingerprint history information DB 25.
- the number of pieces of recognition data 40 read from the fingerprint history information DB 25 is preferably set appropriately according to the specifications of the content recognition apparatus 20 and the like.
- the fingerprint filtering unit 23 performs the sequence 60 of the type information 42 based on the respective timing information 41 of the recognition data 40n received from the receiving device 10 and the recognition data 40 acquired from the fingerprint history information DB 25. Is generated (step S531).
- the sequence 60 is information generated by arranging the type information 42 in the order of the time indicated by the timing information 41.
- the fingerprint filtering unit 23 uses the recognition data 40n and the recognition data 40a, 40b, and 40c acquired from the fingerprint history information DB 25 in the order of the times indicated by the respective timing information 41 (the time is new).
- An example of generating the sequence 60 by arranging various types of information 42 in order) is shown.
- the recognition data 40n, 40a, 40b, and 40c are in the order of the time indicated by the timing information 41, and all of the type information 42 is “A type”, so the sequence 60 is “A type, A type”. , A type, A type ".
- the fingerprint filtering unit 23 reads and acquires a plurality of recognition data 50 from the fingerprint DB 22 (step S532).
- the fingerprint filtering unit 23 acquires, from the fingerprint DB 22, the recognition data 50 that is selected as a candidate for matching in the image recognition process in the property filtering process in step S52.
- FIG. 23 shows an operation example in which the video content ⁇ 1 transmitted from the broadcast station ⁇ , the video content ⁇ 1 transmitted from the broadcast station ⁇ , and the video content ⁇ 1 transmitted from the broadcast station ⁇ are acquired.
- the video contents ⁇ 1 and ⁇ 1 have already been excluded by the property filtering process.
- each of the video contents ⁇ 1, ⁇ 1, and ⁇ 1 is a video content candidate.
- the fingerprint filtering unit 23 generates a sequence candidate 61 for each of the recognition data 50 acquired from the fingerprint DB 22 in Step S532 (Step S533).
- the sequence candidate 61 is information generated by substantially the same method as the sequence 60 generated in step S531, and the type information 52 of the recognition data 50 is displayed in the order of the time indicated by the timing information 51 of the recognition data 50 (time). Is the information generated by arranging them in a new order).
- the fingerprint filtering unit 23 has a time closest to the time indicated by the timing information 41 of each recognition data 40 included in the sequence 60 from the plurality of recognition data 50 acquired in step S532. Recognition data 50 having timing information 51 is selected. Based on the selected recognition data 50, sequence candidates 61 are generated.
- the fingerprint filtering unit 23 selects, for each of the video contents ⁇ 1, ⁇ 1, and ⁇ 1, the recognition data 50 having the timing information 51 that indicates the time closest to that time from the recognition data 50.
- the recognition data having the timing information 51 indicating the time “07/07/2014 03: 32: 36.000” from the respective recognition data 50 of the video contents ⁇ 1, ⁇ 1, and ⁇ 1. 50 is selected.
- the timing information 41 indicates the recognition data 40a, 40b, and 40c from the respective recognition data 50 of the video contents ⁇ 1, ⁇ 1, and ⁇ 1, similarly to the recognition data 40n.
- the recognition data 50 having the timing information 51 indicating the time closest to the time is selected.
- the fingerprint filtering unit 23 sorts the type information 52 included in the selected recognition data 50 for each of the video contents ⁇ 1, ⁇ 1, and ⁇ 1 in the order of the times indicated by the timing information 51 included in the recognition data 50.
- the sequence candidates 61 are generated in order (in order from the newest time).
- the sequence candidate 61 of the video content ⁇ 1 is “A type, A type, A type, A type”
- the sequence candidate 61 of the video content ⁇ 1 is “B type, B type, A type, “B type”
- the sequence candidate 61 of the video content ⁇ 1 is “B type, A type, B type, A type”.
- the sequence candidate 61 is generated for each video content candidate.
- the fingerprint filtering unit 23 uses the sequence 60 generated based on the recognition data 40 in step S531 to determine the recognition data 50 to be collated for image recognition (step S534).
- step S534 the fingerprint filtering unit 23 compares the sequence 60 generated in step S531 with the sequence candidate 61 generated in step S533. Then, the fingerprint filtering unit 23 selects a sequence candidate 61 having the type information 52 arranged in the same order as the type information 42 in the sequence 60.
- the recognition data 50 of the sequence candidate 61 selected in this way is a target for image recognition verification.
- the fingerprint filtering unit 23 selects the sequence candidates 61 of the video content ⁇ 1 in which the type information 52 is arranged in the order of “A type, A type, A type, A type”, and the type information 52 arranged in that order.
- the sequence candidates 61 of the video content ⁇ 1 and the video content ⁇ 1 that do not have the above are excluded.
- the fingerprint filtering unit 23 sets the video content ⁇ 1 as the final video content candidate as a result of the property sequence filtering process, and sets the recognition data 50 of the video content ⁇ 1 as the image recognition collation target. To do.
- FIG. 24 is a flowchart showing an example of recognition data matching processing in the first embodiment.
- the flowchart in FIG. 24 shows an overview of the processing executed in step S6 in FIG.
- FIG. 25 is a diagram schematically illustrating an example of a matching process for static fingerprints according to the first embodiment.
- FIG. 26 is a diagram schematically illustrating an example of the matching process of the dynamic fingerprint in the first embodiment.
- FIG. 27 is a diagram showing an example of video content recognition conditions in the first embodiment.
- FIG. 27 shows five recognition conditions (a) to (e) as an example.
- FIG. 28 is a diagram schematically illustrating an example of video content matching processing according to the first embodiment.
- the fingerprint collation unit 24 recognizes the static fingerprint included in the recognition data 40 transmitted from the receiving apparatus 10 as the recognition data 50 filtered by the fingerprint filtering unit 23 (that is, as a collation target for image recognition in step S534). Check against the static fingerprint included in the selected recognition data 50). Then, the fingerprint collation unit 24 includes a static fingerprint included in the recognition data 40 transmitted from the receiving device 10 and a static fingerprint included in the recognition data 50 filtered by the fingerprint filtering unit 23. Calculate similarity.
- the fingerprint collation unit 24 calculates the degree of coincidence of the static areas as the similarity. Specifically, the fingerprint collation unit 24 adds the static region position of the static fingerprint included in the recognition data 40 transmitted from the receiving device 10 and the recognition data 50 filtered by the fingerprint filtering unit 23. The positions of the static regions of the included static fingerprints are compared with each other. And the fingerprint collation part 24 counts the number of the area
- whether or not they match each other is determined only by whether or not they are static regions, and the luminance value of each block is not considered. If the blocks at the same position are both static regions, the fingerprint collation unit 24 determines that both match even if the luminance values of the blocks are different from each other.
- a static fingerprint C002 shown in FIG. 25 is a static fingerprint included in the recognition data 40 transmitted from the receiving device 10.
- a static fingerprint C00X shown in FIG. 25 is a static fingerprint included in the recognition data 50 filtered by the fingerprint filtering unit 23.
- the number of static area blocks included in the static fingerprint C002 and the number of blocks in the static area included in the static fingerprint C00X are both 13, which is the same number. However, there are some differences in position. Between the static fingerprint C002 and the static fingerprint C00X, the positions of the blocks in the static area coincide with each other in the 5th block from the top of the 25 blocks in the static fingerprint. In other words, there are 11 blocks in total, one in the second row from the top (the block whose luminance value is “128”) and five in the fifth row from the top.
- the fingerprint collation unit 24 compares the calculated similarity with a predetermined static threshold value, and performs similarity determination based on the comparison result.
- the fingerprint collation unit 24 determines “similar” if the calculated similarity is greater than or equal to the static threshold, and determines “not similar” if the calculated similarity is less than the static threshold.
- the static threshold is set to 40%, for example, the fingerprint collation unit 24 determines that the static fingerprint C002 and the static fingerprint C00X are similar.
- the numerical value of the static threshold value is merely an example, and is desirably set appropriately.
- the luminance value of each block constituting the static fingerprint is not considered when calculating the similarity of the static fingerprint. It is not limited.
- the fingerprint collation unit 24 may use the brightness value of each block constituting the static fingerprint when calculating the similarity of the static fingerprint. For example, when collating two static fingerprints, the fingerprint collation unit 24 may calculate the similarity of static fingerprints by counting the number of blocks whose luminance values match each other as well as the position. Good. Or the fingerprint collation part 24 may calculate the similarity of a static fingerprint using normalized cross-correlation matching (Normalized Cross Correlation).
- the fingerprint collation unit 24 calculates the similarity of the dynamic fingerprint (step S61).
- the fingerprint collation unit 24 collates the dynamic fingerprint included in the recognition data 40 transmitted from the receiving device 10 with the dynamic fingerprint included in the recognition data 50 filtered by the fingerprint filtering unit 23. Then, the fingerprint collation unit 24 includes a dynamic fingerprint included in the recognition data 40 transmitted from the receiving device 10 and a dynamic fingerprint included in the recognition data 50 filtered by the fingerprint filtering unit 23. Calculate similarity.
- the fingerprint collation unit 24 calculates the degree of matching of the dynamic areas as the similarity. Specifically, the fingerprint collation unit 24 filters the position of the dynamic region of the dynamic fingerprint and the sign of the luminance change value included in the recognition data 40 transmitted from the receiving device 10 and the fingerprint filtering unit 23. The position of the dynamic region of the dynamic fingerprint included in the recognized recognition data 50 and the sign of the luminance change value are compared with each other. And the fingerprint collation part 24 counts the number of the area
- whether or not they match each other is determined based on whether or not they are dynamic regions and the sign of the luminance change value, and the numerical value of the luminance change value of each block is not considered. . If the blocks at the same position are both dynamic regions and the signs of the luminance change values are the same, the fingerprint collation unit 24 may have different values of the luminance change values of the respective blocks. It is determined that both match.
- the dynamic fingerprint D003 shown in FIG. 26 is a dynamic fingerprint included in the recognition data 40 transmitted from the receiving device 10.
- a dynamic fingerprint D00X illustrated in FIG. 26 is a dynamic fingerprint included in the recognition data 50 filtered by the fingerprint filtering unit 23.
- the number of blocks in the dynamic area that the dynamic fingerprint D003 has is 11, and the number of blocks in the dynamic area that the dynamic fingerprint D00X has is 8. And, between the dynamic fingerprint D003 and the dynamic fingerprint D00X, the position of the block in the dynamic area and the sign of the luminance change value are the same among the 25 blocks in the dynamic fingerprint. , Two blocks in the first row from the top, two in the second row from the top, and one in the fifth row from the top.
- the fingerprint collation unit 24 compares the calculated similarity with a predetermined dynamic threshold and performs similarity determination based on the comparison result.
- the fingerprint collation unit 24 determines “similar” if the calculated similarity is greater than or equal to the dynamic threshold, and determines “not similar” if the calculated similarity is less than the dynamic threshold. In the above example, if the dynamic threshold is set to 30%, for example, the fingerprint collation unit 24 determines that the dynamic fingerprint D003 is not similar to the dynamic fingerprint D00X.
- the numerical value of the dynamic threshold is merely an example, and it is desirable to set it appropriately.
- the static threshold value and the dynamic threshold value described above may be set to the same numerical value, or may be set to different numerical values.
- the fingerprint collation unit 24 performs the similarity determination regarding the static fingerprint based on the similarity calculated in step S60 and the similarity determination regarding the dynamic fingerprint based on the similarity calculated in step S61. To do.
- the fingerprint collation unit 24 may use the absolute value of the luminance change value of each block constituting the dynamic fingerprint when calculating the similarity of the dynamic fingerprint. For example, when collating two dynamic fingerprints, the fingerprint collation unit 24 counts the number of blocks in which the absolute value of the luminance change value in addition to the position and the sign matches each other, and the similarity of the dynamic fingerprints May be calculated. Or the fingerprint collation part 24 may calculate the similarity of a dynamic fingerprint only using the position of the block of a dynamic area similarly to the time of calculating the similarity of a static fingerprint. Or the fingerprint collation part 24 may calculate the similarity degree of a dynamic fingerprint using normalized cross correlation matching.
- step S60 may be executed first, or both may be executed in parallel. May be.
- the fingerprint collation unit 24 recognizes video content (image recognition) based on the result of fingerprint similarity determination (step S62).
- the fingerprint collation unit 24 recognizes the video content based on the result of the static fingerprint similarity determination, the result of the dynamic fingerprint similarity determination, and a predetermined recognition condition. As described above, the fingerprint collation unit 24 includes each of the static fingerprint and the dynamic fingerprint included in the recognition data 40 transmitted from the receiving device 10, and the recognition data 50 filtered by the fingerprint filtering unit 23. Are collated with a plurality of fingerprints 53 included in. The fingerprint collation unit 24 selects one recognition data 50 from the recognition data 50 filtered by the fingerprint filtering unit 23 based on the result of the collation and a predetermined recognition condition, and selects the recognition data 50 as the selected recognition data 50. Information indicating the corresponding video content is output as a result of image recognition.
- the information indicating the video content includes, for example, a file name of the video content, a channel name of a broadcasting station that broadcasts the video content, an EPG ID, and the like.
- the recognition condition is a condition determined based on at least one of a static fingerprint and a dynamic fingerprint.
- An example of recognition conditions is shown in FIG. Note that the recognition conditions shown in FIG. 27 are conditions used in a predetermined period. This predetermined period is a period of a predetermined number of frames. The predetermined period is, for example, a period of 10 frames or less.
- the fingerprint collation unit 24 converts the static fingerprint and the dynamic fingerprint included in the recognition data 40 transmitted from the receiving device 10 during a predetermined period into the recognition data 50 filtered by the fingerprint filtering unit 23. Check with the included fingerprint 53.
- the number of frames here is the number of image change frames. Therefore, the actual period is a coefficient based on the extraction frame rate set in the video extraction unit 12 and the content frame rate (for example, in the examples shown in FIGS. 30 fps, “2”, 20 fps, “3”, 15 fps, “4”, etc.). This number of frames may be the number of image change frames or the number of fingerprints.
- the recognition conditions (a) to (e) shown as an example in FIG. 28 are as follows.
- A At least one of a static fingerprint and a dynamic fingerprint is similar.
- B At least two of the static fingerprint and the dynamic fingerprint are similar.
- C At least one of the static fingerprints is similar and at least one of the dynamic fingerprints is similar.
- D The static fingerprint or the dynamic fingerprint is similar twice in succession.
- E The static fingerprint or the dynamic fingerprint is similar three times in succession.
- Fingerprint collation part 24 judges as follows, for example, when performing collation processing based on recognition conditions (a).
- the fingerprint collation unit 24 can recognize the video content when it is determined that the similarity is “similar” to at least one of the static fingerprint and the dynamic fingerprint in the similarity determination described above. Judgment is made (Yes in step S63). Otherwise, the fingerprint collation unit 24 determines that the video content has not been recognized (No in step S63).
- the fingerprint collation unit 24 executes the following processing during the period of 3 frames of the image change frame.
- the fingerprint collation unit 24 performs the above-described similarity determination with respect to the static fingerprint and the dynamic fingerprint included in the recognition data 40 transmitted from the receiving device 10. If there is a fingerprint 53 that is determined to be “similar” to at least one of them, the fingerprint collation unit 24 determines that the video content has been recognized. And the fingerprint collation part 24 outputs the information which shows the video content corresponding to the recognition data 50 which has the fingerprint 53 as a result of image recognition.
- the fingerprint collation unit 24 determines as follows when performing collation processing based on the recognition condition (b), for example.
- the fingerprint collation unit 24 is able to recognize the video content when it is determined that the similarity is “similar” to at least two of the static fingerprint and the dynamic fingerprint in the similarity determination described above. Judgment is made (Yes in step S63). Otherwise, the fingerprint collation unit 24 determines that the video content has not been recognized (No in step S63).
- the recognition condition (b) includes a case where two or more static fingerprints are determined to be “similar”, a case where two or more dynamic fingerprints are determined to be “similar”, and a static condition. And the case where one or more of the target fingerprints are determined to be “similar” and one or more of the dynamic fingerprints are determined to be “similar”.
- the fingerprint collation unit 24 performs the following processing in the period of 5 frames of the image change frame.
- the fingerprint collation unit 24 performs the above-described similarity determination with respect to the static fingerprint and the dynamic fingerprint included in the recognition data 40 transmitted from the receiving device 10. Then, if there is a fingerprint 53 that is determined to be “similar” to at least two of them, the fingerprint collation unit 24 determines that the video content has been recognized. And the fingerprint collation part 24 outputs the information which shows the video content corresponding to the recognition data 50 which has the fingerprint 53 as a result of image recognition.
- the fingerprint collation unit 24 makes the following determination when performing collation processing based on the recognition condition (c), for example.
- the fingerprint collation unit 24 recognizes video content when it is determined that the similarity is “similar” to at least one of the static fingerprint and at least one of the dynamic fingerprints in the above-described similarity determination. It is determined that it has been completed (Yes in step S63). Otherwise, the fingerprint collation unit 24 determines that the video content has not been recognized (No in step S63).
- the fingerprint collation unit 24 performs the following processing in the period of 5 frames of the image change frame.
- the fingerprint collation unit 24 performs the above-described similarity determination with respect to the static fingerprint and the dynamic fingerprint included in the recognition data 40 transmitted from the receiving device 10. Then, if there is a fingerprint 53 that is determined to be “similar” to at least one of the static fingerprint and at least one of the dynamic fingerprint, the fingerprint collation unit 24 can recognize the video content. Judge. And the fingerprint collation part 24 outputs the information which shows the video content corresponding to the recognition data 50 which has the fingerprint 53 as a result of image recognition.
- a condition regarding the order of the static fingerprint and the dynamic fingerprint may be added to the recognition condition.
- the fingerprint collation unit 24 determines as follows when performing collation processing based on the recognition condition (d), for example.
- the fingerprint collation unit 24 determines that the video content can be recognized when it is determined that the static fingerprint or the dynamic fingerprint is “similar” twice in succession in the similarity determination described above ( Yes in step S63). Otherwise, the fingerprint collation unit 24 determines that the video content has not been recognized (No in step S63).
- This recognition condition (d) has the following meaning.
- the temporally continuous fingerprint 43 included in the recognition data 40 transmitted from the receiving device 10 is determined to be “similar” continuously two or more times. This includes two cases where a static fingerprint generated two or more times consecutively is determined to be “similar” two or more times consecutively, and two dynamic fingerprints generated two or more times consecutively When it is determined to be “similar” continuously more than once and when it is determined to be “similar” continuously two times or more after static fingerprints and dynamic fingerprints generated continuously while switching to each other And are included.
- the fingerprint collation unit 24 performs the following processing in the period of 5 frames of the image change frame.
- the fingerprint collation unit 24 performs the above-described similarity determination with respect to the static fingerprint and the dynamic fingerprint included in the recognition data 40 transmitted from the receiving device 10. Then, if there is a fingerprint 53 for which the static fingerprint or the dynamic fingerprint is determined to be “similar” twice in succession, the fingerprint collation unit 24 determines that the video content has been recognized. And the fingerprint collation part 24 outputs the information which shows the video content corresponding to the recognition data 50 which has the fingerprint 53 as a result of image recognition.
- the fingerprint collation unit 24 determines as follows when performing collation processing based on the recognition condition (e), for example.
- the fingerprint collation unit 24 determines that the video content has been recognized when it is determined that the static fingerprint or the dynamic fingerprint is “similar” three consecutive times in the similarity determination described above ( Yes in step S63). Otherwise, the fingerprint collation unit 24 determines that the video content has not been recognized (No in step S63).
- This recognition condition (e) has the following meaning. It is determined that the temporally continuous fingerprint 43 included in the recognition data 40 transmitted from the receiving device 10 is “similar” continuously three or more times. This includes three cases where a static fingerprint generated three or more times consecutively is determined to be “similar” three or more times consecutively and three dynamic fingerprints generated three or more times consecutively. When it is determined to be “similar” continuously for more than once and when it is determined to be “similar” for three times or more consecutively generated static fingerprints and dynamic fingerprints while switching to each other And are included.
- the fingerprint collation unit 24 performs the following processing in the period of 8 frames of the image change frame.
- the fingerprint collation unit 24 performs the above-described similarity determination with respect to the static fingerprint and the dynamic fingerprint included in the recognition data 40 transmitted from the receiving device 10. Then, if there is a fingerprint 53 for which the static fingerprint or the dynamic fingerprint is determined to be “similar” three times in succession, the fingerprint collation unit 24 determines that the video content has been recognized. And the fingerprint filtering part 23 outputs the information which shows the video content corresponding to the recognition data 50 which has the fingerprint 53 as a result of image recognition.
- collation image recognition
- collation is performed by increasing the number of fingerprints determined to be “similar” or the number of fingerprints determined to be “similar” successively.
- Accuracy of processing can be improved.
- FIG. 28 schematically shows an example of the operation when the fingerprint collation unit 24 performs collation processing based on the recognition condition (e).
- the fingerprint collation unit 24 sets the recognition condition that the static fingerprint or the dynamic fingerprint is similar three times in succession.
- the fingerprint 53 of the content 00X filtered by the fingerprint filtering unit 23 is a static fingerprint A, a dynamic fingerprint B, a static fingerprint C, a dynamic fingerprint D, or a static fingerprint E. Suppose they are in order. In FIG. 28, “static A”, “dynamic B”, “static C”, “dynamic D”, and “static E” are described.
- the fingerprints 43 transmitted from the receiving apparatus 10 are arranged in the order of the static fingerprint A, the dynamic fingerprint B, and the static fingerprint C.
- FIG. 28 “Static A”, “Dynamic B”, and “Static C” are indicated respectively.
- the fingerprint collation unit 24 outputs a “similar” determination result for each of the static fingerprint A, the dynamic fingerprint B, and the static fingerprint C in the above-described similarity determination. . That is, the fingerprint collation unit 24 determines “similar” three times in succession.
- the fingerprint collation unit 24 determines that the fingerprint 43 included in the recognition data 40 transmitted from the receiving device 10 is similar to the fingerprint 53 included in the recognition data 50 of the content 00X. That is, the fingerprint collation unit 24 recognizes that the video content received by the receiving device 10 is the content 00X. Then, the fingerprint collation unit 24 outputs information indicating the content 00X as a collation result.
- the fingerprint collation unit 24 When the fingerprint collation unit 24 recognizes the video content (image recognition) (Yes in step S63), the fingerprint collation unit 24 outputs the result of the image recognition to the communication unit (not shown) (step S64). ).
- the communication unit (not shown) of the content recognition device 20 transmits information representing the image recognition result received from the fingerprint collation unit 24 to the reception device 10 (step S8 in FIG. 8).
- Information transmitted from the content recognition device 20 to the reception device 10 is information representing the video content selected by the fingerprint collation unit 24 by performing image recognition based on the recognition data 40 transmitted from the reception device 10. This is information representing the video content being received by the receiving device 10.
- the information is, for example, a content ID, but may be any information as long as the information can identify the video content.
- the receiving device 10 can acquire additional information regarding the video content being received from, for example, the advertisement server device 30.
- step S64 the processing of the content recognition device 20 returns to step S1 in FIG. 8, and a series of processing after step S1 is repeated.
- the content recognition device 20 receives information indicating that image recognition could not be performed when the video content corresponding to the recognition data 40 transmitted from the reception device 10 could not be specified as a result of the image recognition processing. May be sent to. Alternatively, nothing may be transmitted.
- the content recognition apparatus includes a fingerprint generation unit, a selection unit, and a collation unit.
- the fingerprint generation unit generates a fingerprint for each of the plurality of acquired video content candidates.
- the sorting unit sorts video content candidates using attached information included in recognition data input from the outside.
- the collation unit collates the fingerprint of the video content candidate selected by the selection unit with the fingerprint included in the recognition data, and determines the video content corresponding to the fingerprint included in the recognition data as the video content candidate. Identify from.
- the content recognition device 20 is an example of a content recognition device.
- the fingerprint generation unit 2110 is an example of a fingerprint generation unit.
- the fingerprint filtering unit 23 is an example of a selection unit.
- the fingerprint collation unit 24 is an example of a collation unit.
- Each of the video contents ⁇ 1, ⁇ 1, ⁇ 1, ⁇ 1, and ⁇ 1 is an example of a video content candidate.
- the fingerprint 53 is an example of a fingerprint.
- the recognition data 40 is an example of recognition data input from the outside.
- the type information 42 is an example of attached information.
- the fingerprint 43 is an example of a fingerprint included in the recognition data.
- FIG. 29 is a diagram for explaining points that are problematic in recognition of video content.
- the content recognition apparatus 20 that recognizes video content substantially in real time receives a plurality of video contents broadcast from a plurality of broadcasting stations 2, and based on the received video contents.
- a fingerprint 53 is generated, and the generated fingerprint 53 is stored in the fingerprint DB 22.
- fingerprints 53 corresponding to the number of received video contents are accumulated in the fingerprint DB 22 with the passage of time. For this reason, the fingerprint 53 accumulate
- the content recognition device 20 does not include the fingerprint filtering unit 23
- the content recognition device 20 stores a huge number of fingerprints 43 included in the recognition data 40 received from the reception device 10 stored in the fingerprint DB 22. Therefore, it takes a long time to obtain the recognition result of the video content.
- the recognition data 50 selected by the fingerprint filtering unit 23 using the attached information is used as a verification target of the recognition data 40. Therefore, according to the present disclosure, since the number of data used for video content recognition can be reduced, it is possible to reduce the processing related to video content recognition while increasing the recognition accuracy of the video content.
- the recognition data input from the outside may include type information indicating the type of fingerprint as attached information.
- the selection unit may select the video content candidate by comparing the type information included in the recognition data with the type of fingerprint of the video content candidate.
- the static fingerprint and the dynamic fingerprint are examples of fingerprint types, and the type information 42 is an example of type information.
- the type information can be represented by a 1-bit information amount, and therefore the processing required for selection in the fingerprint filtering unit 23 can be reduced.
- the selection unit selects the video content candidate by comparing the arrangement order of the type information included in the recognition data input from the outside with the arrangement order related to the fingerprint type of the video content candidate. May be performed.
- the sequence 60 is an example of the arrangement order of the type information included in the recognition data input from the outside, and the sequence candidate 61 is an example of the arrangement order regarding the fingerprint type of the video content candidate.
- the fingerprint filtering unit 23 can further narrow down the number of recognition data 50 used for collation in the fingerprint collation unit 24. Therefore, it is possible to further reduce the process for collation in the fingerprint collation unit 24.
- the recognition data input from the outside may include information indicating the generation time of the fingerprint included in the recognition data.
- the selection unit may select the fingerprint of the video content candidate based on the information indicating the generation time and the generation time of the fingerprint of the video content candidate.
- the timing information 41 is an example of information indicating a fingerprint generation time included in recognition data input from the outside.
- the timing information 41 is an example of information indicating the generation time of the fingerprint of the video content candidate.
- the accuracy of sorting performed by the fingerprint filtering unit 23 can be increased.
- the selection unit selects a fingerprint generated at a time closest to the information indicating the generation time included in the recognition data input from the outside from the fingerprints of the video content candidates, Video content candidates may be selected based on a comparison between the selected fingerprint type and the type information included in the recognition data.
- the accuracy of sorting performed by the fingerprint filtering unit 23 can be further increased.
- the fingerprint generation unit generates a static fingerprint based on a static region in which an image change amount between a plurality of image frames constituting the video content candidate is smaller than a first threshold.
- the dynamic fingerprint may be generated based on a dynamic region in which the image change amount between the frames is larger than the second threshold.
- the static area is an area occupied by a subject whose background and movement or change are small in the image frame. That is, in a continuous image frame, the subject in the static area has relatively little movement or change. Therefore, it is possible to improve the accuracy of image recognition by specifying a static region and performing image recognition.
- the dynamic area is an area where a relatively large image change occurs due to a scene change or the like. That is, since the dynamic region is a region where characteristic image changes occur, it is possible to increase the accuracy of image recognition by identifying the dynamic region and performing image recognition. In addition, since the number of frames in which a dynamic area is generated is relatively small, the number of frames necessary for image recognition can be reduced.
- the attached information may include geographical information indicating the position of the device that transmits the video content that is the generation source of the recognition data input from the outside, or the position of the device that transmits the recognition data.
- the broadcast station 2 is an example of a device that transmits video content.
- the receiving device 10 is an example of a device that transmits recognition data.
- video content broadcast from the broadcasting station 2 that cannot be received by the receiving device 10 can be excluded by the fingerprint filtering unit 23, so that the fingerprint verification unit 24 can reduce the processing for verification.
- the attached information may include user information stored in a device that transmits recognition data input from the outside.
- video content that does not apply to the user information can be excluded by the fingerprint filtering unit 23, so that the fingerprint verification unit 24 can reduce the processing for verification.
- a recording medium such as a system, an apparatus, an integrated circuit, a computer program, or a computer-readable CD-ROM.
- the system, the apparatus, the integrated circuit, and the computer program And any combination of recording media.
- the first embodiment has been described as an example of the technique disclosed in the present application.
- the technology in the present disclosure is not limited to this, and can also be applied to embodiments in which changes, replacements, additions, omissions, and the like are performed.
- the operation example in which the content recognition apparatus 20 recognizes video content substantially in real time is shown, but the present disclosure is not limited to this operation example.
- the receiving device 10 reads and displays video content (for example, recorded program content) stored in a recording medium
- the content recognition device 20 operates in the same manner as in the first embodiment described above. Video content can be recognized.
- the content recognition device 20 receives the recognition data 40 from the reception device 10, performs a filtering process and a collation process for recognizing video content, and transmits the result to the reception device 10 will be described. did.
- This operation is called “online matching”.
- performing the collation process for recognizing the video content by the receiving device 10 is referred to as “local matching”.
- the receiving device 10 can perform local matching by acquiring the recognition data 50 stored in the fingerprint DB 22 from the content recognition device 20. At this time, the receiving device 10 does not have to acquire all the recognition data 50 stored in the fingerprint DB 22. For example, the receiving device 10 may acquire the recognition data 50 after performing each filtering (for example, region filtering and profile filtering) by the fingerprint filtering unit 23.
- each filtering for example, region filtering and profile filtering
- the fingerprint filtering unit 23 As an operation example of the fingerprint filtering unit 23, an operation example in which filtering is performed in the order of region filtering, profile filtering, property filtering, and property sequence filtering has been described. It is not limited to.
- the fingerprint filtering unit 23 may perform each filtering in an order different from that in the first embodiment, or may perform filtering by selecting one or more and three or less from these filterings.
- the fingerprint filtering unit 23 reads out and acquires the recognition data 50 from the fingerprint DB 22 every time each filtering process is performed.
- the present disclosure is not limited to this operation example.
- the fingerprint filtering unit 23 temporarily stores a plurality of recognition data 50 read from the fingerprint DB 22 immediately before the first filtering process in a storage device such as a memory, and each time the recognition data 50 is excluded in each filtering process.
- the recognition data 50 may be deleted from the storage device.
- Embodiment 1 a configuration example using both static fingerprints and dynamic fingerprints for video content recognition has been shown, but the present disclosure is not limited to this configuration.
- Video content may be recognized using only one of a static fingerprint and a dynamic fingerprint.
- the fingerprint generation units 2110 and 110 may be configured to include only one of the static region determination unit 231 and the dynamic region determination unit 232, for example.
- the fingerprint generation units 2110 and 110 may be configured to include only one of the static fingerprint generation unit 241 and the dynamic fingerprint generation unit 242, for example. Also, there may be three or more fingerprints.
- the content recognition apparatus 20 shown in the first embodiment can be used for recognition of advertising content, for example. Alternatively, it can be used for recognizing program content such as dramas and variety.
- the receiving device 10 relates to, for example, the performer's own profile, the clothes worn by the performer, the place where the performer visited, etc. as additional information based on the image recognition result from the advertisement server device 30. Information may be acquired, and the acquired information may be displayed superimposed on the video being displayed.
- the content recognition apparatus 20 may receive not only advertising content but also video content such as program content, and generate a fingerprint corresponding to the video content.
- the fingerprint DB 22 may hold not only the advertising content but also a fingerprint corresponding to the program content in association with the content ID.
- each component may be configured by dedicated hardware, or may be realized by executing a software program suitable for each component.
- Each component may be realized by a program execution unit such as a CPU or a processor reading and executing a software program recorded on a recording medium such as a hard disk or a semiconductor memory.
- the software that realizes the content recognition apparatus 20 of the first embodiment is the following program.
- the program is a program for causing a computer to execute a content recognition method, and includes generating a fingerprint for each of a plurality of acquired video content candidates and an attached data included in recognition data input from outside The video content corresponding to the fingerprint included in the recognition data by collating the fingerprint of the selected video content candidate with the fingerprint included in the recognition data by selecting the video content candidate using information Identifying from among video content candidates.
- the above program may be recorded on a recording medium and distributed or distributed.
- the distributed program by installing the distributed program in the devices and causing the processors of the devices to execute the programs, it is possible to cause the devices to perform various processes.
- the system LSI is an ultra-multifunctional LSI manufactured by integrating a plurality of components on a single chip.
- the system LSI is a computer system including a microprocessor, a ROM, a RAM, and the like.
- a computer program is stored in the ROM.
- the system LSI achieves its functions by the microprocessor loading a computer program from the ROM to the RAM and performing operations such as operations in accordance with the loaded computer program.
- each of the above-described devices may be configured from an IC card that can be attached to and detached from each device or a single module.
- the IC card or module is a computer system that includes a microprocessor, ROM, RAM, and the like.
- the IC card or the module may include the super multifunctional LSI described above.
- the IC card or the module achieves its functions by the microprocessor operating according to the computer program. This IC card or this module may have tamper resistance.
- the present disclosure also relates to a computer-readable recording medium such as a flexible disk, a hard disk, a CD-ROM, an MO, a DVD, a DVD-ROM, a DVD-RAM, a BD (Blu-Ray Disc ( (Registered trademark)), or recorded in a semiconductor memory or the like. Further, it may be realized by a digital signal recorded on these recording media.
- a computer-readable recording medium such as a flexible disk, a hard disk, a CD-ROM, an MO, a DVD, a DVD-ROM, a DVD-RAM, a BD (Blu-Ray Disc (Registered trademark)), or recorded in a semiconductor memory or the like. Further, it may be realized by a digital signal recorded on these recording media.
- the computer program or digital signal in the present disclosure may be transmitted via an electric communication line, a wireless or wired communication line, a network such as the Internet, data broadcasting, or the like.
- the present disclosure can be implemented by another independent computer system by recording and transferring a program or digital signal on a recording medium, or by transferring a program or digital signal via a network or the like. Also good.
- each processing may be realized by centralized processing by a single device (system), or may be realized by distributed processing by a plurality of devices. Also good.
- the present disclosure is applicable to a content recognition apparatus and a content recognition method that recognize video content using a communication network. Specifically, the present disclosure can be applied to a video receiving device such as a television or a server device.
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Security & Cryptography (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
Description
[1-1.コンテンツ認識システム]
まず、本実施の形態におけるコンテンツ認識システムについて、図1を用いて説明する。
まず、本実施の形態における受信装置10について、図1および図2を用いて説明する。
次に、本実施の形態におけるコンテンツ認識装置20について、図1を用いて説明する。
次に、広告サーバ装置30について説明する。
次に、本実施の形態におけるフィンガープリント生成部110について説明する。
静的領域とは、2つの画像フレーム間で画像の変化量が、あらかじめ定められた閾値(以下、「第1閾値」と記す)よりも少ない領域のことである。静的領域は、例えば、画像中の背景または動きや変化の少ない被写体が占める領域、等である。静的領域は、画像フレーム間で画像の変化量を算出することにより決定される。
動的領域とは、2つの画像フレーム間で画像の変化量が、あらかじめ定められた閾値(以下、「第2閾値」と記す)より大きい領域のことである。動的領域は、例えば、シーンの切り替わり時に大きな画像の変化が発生する領域、等である。
次に、本実施の形態におけるフィンガープリント生成部110について、図7を用いて説明する。
次に、本実施の形態におけるコンテンツ認識システム1の動作を、図8~図28を用いて説明する。
まず、本実施の形態におけるコンテンツ認識システム1の全体の動作について、図8を用いて説明する。
次に、本実施の形態で認識データが生成されるときの処理(図8のステップS2の処理)の詳細を、図9~図18を用いて説明する。
次に、本実施の形態で画像フレーム間の変化量を算出するときの処理の詳細を、図11~図13を用いて説明する。
次に、本実施の形態で静的フィンガープリントを生成するときの処理の詳細を、図14、図15を用いて説明する。
次に、本実施の形態で動的フィンガープリントを生成するときの処理の詳細を、図16~図18を用いて説明する。
次に、コンテンツ認識装置20のフィンガープリントフィルタリング部23で実行されるフィルタリング処理について、図19~図23を用いて説明する。まず、図19を用いて、フィルタリング処理の概要を説明する。
次に、コンテンツ認識装置20のフィンガープリントフィルタリング部23で実行されるプロパティフィルタリング処理について、図20、図21を用いて説明する。
次に、コンテンツ認識装置20のフィンガープリントフィルタリング部23で実行されるプロパティシーケンスフィルタリング処理について、図22、図23を用いて説明する。
次に、本実施の形態において認識データの照合を実行するときの処理の詳細を、図24~図28を用いて説明する。
図24のフローチャートを説明する。図1に示したフィンガープリント照合部24は、静的フィンガープリントの類似度を算出する(ステップS60)。
次に、フィンガープリント照合部24は、動的フィンガープリントの類似度を算出する(ステップS61)。
次に、フィンガープリント照合部24は、フィンガープリントの類似判定の結果に基づき、映像コンテンツの認識(画像認識)を行う(ステップS62)。
(a)静的フィンガープリントおよび動的フィンガープリントの少なくとも1つが類似する。
(b)静的フィンガープリントおよび動的フィンガープリントの少なくとも2つが類似する。
(c)静的フィンガープリントの少なくとも1つが類似し、動的フィンガープリントの少なくとも1つが類似する。
(d)静的フィンガープリントまたは動的フィンガープリントが2回連続して類似する。
(e)静的フィンガープリントまたは動的フィンガープリントが3回連続して類似する。
以上のように、本実施の形態において、コンテンツ認識装置は、フィンガープリント生成部と、選別部と、照合部と、を備える。フィンガープリント生成部は、取得した複数の映像コンテンツ候補のそれぞれについてフィンガープリントを生成する。選別部は、外部から入力される認識データに含まれる付属情報を用いて映像コンテンツ候補を選別する。照合部は、選別部によって選別された映像コンテンツ候補のフィンガープリントと、認識データに含まれるフィンガープリントと、を照合して、認識データに含まれるフィンガープリントに対応する映像コンテンツを映像コンテンツ候補の中から特定する。
以上のように、本出願において開示する技術の例示として、実施の形態1を説明した。しかしながら、本開示における技術は、これに限定されず、変更、置き換え、付加、省略等を行った実施の形態にも適用できる。また、上記実施の形態1で説明した各構成要素を組み合わせて、新たな実施の形態とすることも可能である。
2 放送局
3 STB
10 受信装置
11 映像受信部
11a 映像入力部
11b 第1外部入力部
11c 第2外部入力部
12 映像抽出部
13 付加情報取得部
14 映像出力部
15 制御部
16 操作信号受信部
17 HTTP送受信部
18 付加情報格納部
19 付加情報表示制御部
20 コンテンツ認識装置
21 コンテンツ受信部
22 フィンガープリントDB
23 フィンガープリントフィルタリング部
24 フィンガープリント照合部
25 フィンガープリント履歴情報DB
30 広告サーバ装置
31 付加情報DB
40,40a,40b,40c,40n,50,50a,50b,50c 認識データ
41,51,51a,51b,51c タイミング情報
42,52,52a,52b,52c 種別情報
43,53,53a,53b,53c フィンガープリント
60 シーケンス
61 シーケンス候補
91,92 フレーム
100 認識部
105 通信ネットワーク
110,2110 フィンガープリント生成部
111 画像取得部
112 データ生成部
120 フィンガープリント送信部
130 認識結果受信部
210 スケール変換部
220 差分算出部
230 決定部
231 静的領域決定部
232 動的領域決定部
240 生成部
241 静的フィンガープリント生成部
242 動的フィンガープリント生成部
Claims (10)
- 取得した複数の映像コンテンツ候補のそれぞれについてフィンガープリントを生成するフィンガープリント生成部と、
外部から入力される認識データに含まれる付属情報を用いて前記映像コンテンツ候補を選別する選別部と、
前記選別部によって選別された前記映像コンテンツ候補のフィンガープリントと、前記認識データに含まれるフィンガープリントと、を照合して、前記認識データに含まれるフィンガープリントに対応する映像コンテンツを前記映像コンテンツ候補の中から特定する照合部と、を備える、
コンテンツ認識装置。 - 前記フィンガープリントは、静的フィンガープリントと動的フィンガープリントとの少なくとも2種類に分類され、
前記認識データには、前記フィンガープリントの種類を示す種別情報が前記付属情報として含まれ、
前記選別部は、前記認識データに含まれる前記種別情報と、前記映像コンテンツ候補のフィンガープリントの種類と、を比較して前記映像コンテンツ候補の選別を行う、
請求項1に記載のコンテンツ認識装置。 - 前記選別部は、前記認識データに含まれる前記種別情報の配列順序と、前記映像コンテンツ候補のフィンガープリントの種類に関する配列順序と、を比較して前記映像コンテンツ候補の選別を行う、
請求項2に記載のコンテンツ認識装置。 - 前記認識データには、前記認識データに含まれるフィンガープリントの生成時刻を示す情報が含まれ、
前記選別部は、前記生成時刻を示す情報と、前記映像コンテンツ候補のフィンガープリントの生成時刻とに基づき、前記映像コンテンツ候補のフィンガープリントの選択を行う、
請求項2に記載のコンテンツ認識装置。 - 前記選別部は、前記映像コンテンツ候補のフィンガープリントの中から、前記認識データに含まれる前記生成時刻を示す情報に最も近い時刻に生成されたフィンガープリントを選択し、選択したフィンガープリントの種類と、前記認識データに含まれる種別情報との比較に基づき、前記映像コンテンツ候補の選別を行う、
請求項4に記載のコンテンツ認識装置。 - 前記フィンガープリント生成部は、
前記映像コンテンツ候補を構成する複数の画像フレームのフレーム間の画像の変化量が第1閾値より小さい静的領域に基づいて前記静的フィンガープリントを生成し、前記フレーム間の画像の変化量が第2閾値より大きい動的領域に基づいて前記動的フィンガープリントを生成する、
請求項2に記載のコンテンツ認識装置。 - 前記付属情報には、前記認識データの生成元となる映像コンテンツを送信する装置の位置または前記認識データを送信する装置の位置、を示す地理情報が含まれる、
請求項1に記載のコンテンツ認識装置。 - 前記付属情報には、前記認識データを送信する装置に保存されたユーザ情報が含まれる、
請求項1に記載のコンテンツ認識装置。 - 取得した複数の映像コンテンツ候補のそれぞれについてフィンガープリントを生成し、
外部から入力される認識データに含まれる付属情報を用いて前記映像コンテンツ候補を選別し、
選別された前記映像コンテンツ候補のフィンガープリントと、前記認識データに含まれるフィンガープリントと、を照合して、前記認識データに含まれるフィンガープリントに対応する映像コンテンツを前記映像コンテンツ候補の中から特定する、
コンテンツ認識方法。 - 請求項9に記載のコンテンツ認識方法をコンピュータに実行させるためのプログラム。
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP15832982.1A EP3185577B1 (en) | 2014-08-21 | 2015-08-19 | Content identification apparatus and content identification method |
US15/302,460 US10200765B2 (en) | 2014-08-21 | 2015-08-19 | Content identification apparatus and content identification method |
JP2016543816A JP6432047B2 (ja) | 2014-08-21 | 2015-08-19 | コンテンツ認識装置およびコンテンツ認識方法 |
CN201580020600.7A CN106233746B (zh) | 2014-08-21 | 2015-08-19 | 内容辨识装置、内容辨识方法以及记录介质 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2014168709 | 2014-08-21 | ||
JP2014-168709 | 2014-08-21 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2016027457A1 true WO2016027457A1 (ja) | 2016-02-25 |
Family
ID=55350420
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2015/004112 WO2016027457A1 (ja) | 2014-08-21 | 2015-08-19 | コンテンツ認識装置およびコンテンツ認識方法 |
Country Status (5)
Country | Link |
---|---|
US (1) | US10200765B2 (ja) |
EP (1) | EP3185577B1 (ja) |
JP (1) | JP6432047B2 (ja) |
CN (1) | CN106233746B (ja) |
WO (1) | WO2016027457A1 (ja) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2019511138A (ja) * | 2016-02-29 | 2019-04-18 | グレースノート インコーポレイテッド | 複数一致検出とロケーションに基づく曖昧性排除とを用いたメディアチャネル識別 |
CN111133765A (zh) * | 2017-09-13 | 2020-05-08 | 尼尔森(美国)有限公司 | 通过自动内容识别进行冷匹配 |
KR20200103117A (ko) * | 2017-09-13 | 2020-09-01 | 더 닐슨 컴퍼니 (유에스) 엘엘씨 | 자동 콘텐츠 인식에 대한 광고 프레임들의 플래깅 |
Families Citing this family (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9607223B2 (en) * | 2015-04-09 | 2017-03-28 | Facebook, Inc. | Systems and methods for defining and analyzing video clusters based on video image frames |
US9723347B2 (en) * | 2015-10-01 | 2017-08-01 | Sorenson Media, Inc | Frequency capping for media content |
US9848214B2 (en) | 2015-10-01 | 2017-12-19 | Sorenson Media, Inc. | Sequentially overlaying media content |
US10432994B2 (en) | 2016-02-17 | 2019-10-01 | The Nielsen Company (Us), Llc | Content comparison testing on linear media streams |
US10063918B2 (en) | 2016-02-29 | 2018-08-28 | Gracenote, Inc. | Media channel identification with multi-match detection and disambiguation based on single-match |
US9930406B2 (en) | 2016-02-29 | 2018-03-27 | Gracenote, Inc. | Media channel identification with video multi-match detection and disambiguation based on audio fingerprint |
US10013614B2 (en) * | 2016-06-29 | 2018-07-03 | Google Llc | Using an image matching system to improve the quality of service of a video matching system |
US10623800B2 (en) | 2018-07-16 | 2020-04-14 | Gracenote, Inc. | Dynamic control of fingerprinting rate to facilitate time-accurate revision of media content |
CN111753181A (zh) * | 2019-03-28 | 2020-10-09 | 北京京东尚科信息技术有限公司 | 基于图像的搜索方法、装置、服务器、客户端及介质 |
CN112399209B (zh) * | 2019-08-13 | 2022-06-07 | 大唐移动通信设备有限公司 | 一种视频业务识别处理方法及装置 |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004341940A (ja) * | 2003-05-16 | 2004-12-02 | Fujitsu Ltd | 類似画像検索装置、類似画像検索方法、および類似画像検索プログラム |
JP2011059504A (ja) * | 2009-09-11 | 2011-03-24 | Victor Co Of Japan Ltd | コンテンツ同定方法、コンテンツ同定システム、コンテンツ検索装置及びコンテンツ利用装置 |
WO2014006903A1 (ja) * | 2012-07-06 | 2014-01-09 | パナソニック株式会社 | コンテンツ制御方法、コンテンツ制御装置およびプログラム |
Family Cites Families (111)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH04245552A (ja) | 1991-01-31 | 1992-09-02 | Canon Inc | データベース検索方法 |
JPH09185720A (ja) | 1995-12-28 | 1997-07-15 | Canon Inc | 画像抽出装置 |
JP3731952B2 (ja) | 1996-10-18 | 2006-01-05 | シャープ株式会社 | 動画像検索用情報生成装置 |
JPH10214258A (ja) | 1997-01-28 | 1998-08-11 | Victor Co Of Japan Ltd | データ処理システム |
JP2000287189A (ja) | 1999-03-31 | 2000-10-13 | Toshiba Corp | テレビ番組の視聴管理装置 |
JP3485020B2 (ja) | 1999-04-09 | 2004-01-13 | 日本電気株式会社 | 文字認識方法及び装置ならびに記憶媒体 |
US8205223B2 (en) | 2000-04-12 | 2012-06-19 | Lg Electronics Inc. | Method and video device for accessing information |
WO2001080553A1 (fr) | 2000-04-14 | 2001-10-25 | Nippon Telegraph And Telephone Corporation | Procede, systeme et appareil permettant d'acquerir des informations concernant des informations diffusees |
JP3994682B2 (ja) | 2000-04-14 | 2007-10-24 | 日本電信電話株式会社 | 放送情報送受信システム |
US7712123B2 (en) | 2000-04-14 | 2010-05-04 | Nippon Telegraph And Telephone Corporation | Method, system, and apparatus for acquiring information concerning broadcast information |
JP2002016885A (ja) | 2000-06-30 | 2002-01-18 | Pioneer Electronic Corp | 映像再生装置及び映像再生方法 |
US20020126990A1 (en) | 2000-10-24 | 2002-09-12 | Gary Rasmussen | Creating on content enhancements |
JP2002175311A (ja) | 2000-12-08 | 2002-06-21 | Nippon Telegr & Teleph Corp <Ntt> | 映像情報登録方法、映像情報登録装置、映像情報検索方法及び映像情報検索装置 |
US6829015B2 (en) | 2001-01-19 | 2004-12-07 | Samsung Electronics Co., Ltd. | Device and method for realizing transparency in an on screen display |
JP2002232372A (ja) | 2001-01-31 | 2002-08-16 | Matsushita Electric Ind Co Ltd | 放送番組関連情報提供装置 |
KR20020072453A (ko) | 2001-03-10 | 2002-09-16 | 삼성전자 주식회사 | 재생장치 및 부가정보 서비스 서버 시스템 |
JP2002334010A (ja) | 2001-05-11 | 2002-11-22 | Tsubasa System Co Ltd | 広域ネットワークシステム |
JP2004007323A (ja) | 2001-06-11 | 2004-01-08 | Matsushita Electric Ind Co Ltd | テレビジョン放送受信機 |
US20030084462A1 (en) | 2001-10-26 | 2003-05-01 | Junichi Kubota | Digital boradcast reception device and method thereof, and printing device and method thereof |
US20030149983A1 (en) | 2002-02-06 | 2003-08-07 | Markel Steven O. | Tracking moving objects on video with interactive access points |
JP2004104368A (ja) | 2002-09-06 | 2004-04-02 | Sony Corp | 画像データ処理方法、画像データ処理プログラム及び立体画像表示装置 |
KR100930351B1 (ko) | 2003-02-25 | 2009-12-08 | 엘지전자 주식회사 | 고밀도 광디스크의 그래픽 및 서브타이틀 데이터관리방법 |
US20040194130A1 (en) | 2003-03-07 | 2004-09-30 | Richard Konig | Method and system for advertisement detection and subsitution |
US7738704B2 (en) | 2003-03-07 | 2010-06-15 | Technology, Patents And Licensing, Inc. | Detecting known video entities utilizing fingerprints |
US7809154B2 (en) * | 2003-03-07 | 2010-10-05 | Technology, Patents & Licensing, Inc. | Video entity recognition in compressed digital video streams |
JP4109185B2 (ja) | 2003-12-01 | 2008-07-02 | 日本電信電話株式会社 | 映像シーン区間情報抽出方法,映像シーン区間情報抽出装置,映像シーン区間情報抽出プログラムおよびそのプログラムを記録した記録媒体 |
JP2005167894A (ja) | 2003-12-05 | 2005-06-23 | Matsushita Electric Ind Co Ltd | コンテンツ切替方法、およびチャンネル切替方法 |
JP2005347806A (ja) | 2004-05-31 | 2005-12-15 | Toshiba Corp | 電子機器および該電子機器にて実行される番組情報取得方法 |
US7762760B2 (en) * | 2004-06-24 | 2010-07-27 | Paceco Corp. | Method of operating a cargo container scanning crane |
JP2006030244A (ja) | 2004-07-12 | 2006-02-02 | Matsushita Electric Ind Co Ltd | 情報端末及びプログラム |
WO2006030742A1 (ja) | 2004-09-14 | 2006-03-23 | Matsushita Electric Industrial Co., Ltd. | 機器制御システムおよび機器制御方法 |
EP1817908B1 (en) * | 2004-11-30 | 2010-12-22 | The University Court of the University of St. Andrews | System, method for video fingerprinting |
US20060200842A1 (en) | 2005-03-01 | 2006-09-07 | Microsoft Corporation | Picture-in-picture (PIP) alerts |
JP2006303936A (ja) | 2005-04-21 | 2006-11-02 | Matsushita Electric Ind Co Ltd | コンテンツデータ、受信装置、生成装置および送信装置 |
US8886014B2 (en) | 2005-08-11 | 2014-11-11 | Mitsubishi Electric Corporation | Video recording apparatus, scene change extraction method, and video audio recording apparatus |
JP3894940B2 (ja) | 2005-08-11 | 2007-03-22 | 三菱電機株式会社 | 映像音声記録装置 |
EP1954041A4 (en) | 2005-09-30 | 2010-01-27 | Pioneer Corp | SUMMARY PRODUCTION EQUIPMENT AND PROGRAM THEREFOR |
JP2007134948A (ja) | 2005-11-10 | 2007-05-31 | Matsushita Electric Ind Co Ltd | 駐車車両搭乗者撮像装置 |
US20070157242A1 (en) | 2005-12-29 | 2007-07-05 | United Video Properties, Inc. | Systems and methods for managing content |
JP4871668B2 (ja) | 2006-08-01 | 2012-02-08 | キヤノン株式会社 | 画像処理システム、画像処理方法及びプログラム |
JP2008040622A (ja) | 2006-08-02 | 2008-02-21 | Sharp Corp | 広告提供システム、広告提供装置、広告表示装置、広告提供方法およびプログラム |
JP2008116792A (ja) | 2006-11-07 | 2008-05-22 | Sharp Corp | 表示制御装置、映像表示機器及び表示制御方法 |
JP4746568B2 (ja) | 2007-01-16 | 2011-08-10 | ヤフー株式会社 | 情報提供装置、情報提供方法、及びプログラム |
JP4957960B2 (ja) | 2007-01-29 | 2012-06-20 | ソニー株式会社 | 画像処理装置、画像処理方法、及びプログラム |
US8281332B2 (en) | 2007-05-02 | 2012-10-02 | Google Inc. | Animated video overlays |
US8229227B2 (en) | 2007-06-18 | 2012-07-24 | Zeitera, Llc | Methods and apparatus for providing a scalable identification of digital video sequences |
US8433611B2 (en) | 2007-06-27 | 2013-04-30 | Google Inc. | Selection of advertisements for placement with content |
WO2009011030A1 (ja) | 2007-07-17 | 2009-01-22 | Pioneer Corporation | 情報処理システム、並びに情報処理装置及び情報処理方法 |
JP5203648B2 (ja) | 2007-07-20 | 2013-06-05 | オリンパス株式会社 | 画像抽出装置および画像抽出プログラム |
JP2009088777A (ja) | 2007-09-28 | 2009-04-23 | Dentsu Inc | 広告選択最適化処理装置及びその処理方法 |
US8060609B2 (en) | 2008-01-04 | 2011-11-15 | Sling Media Inc. | Systems and methods for determining attributes of media items accessed via a personal media broadcaster |
US8621505B2 (en) | 2008-03-31 | 2013-12-31 | At&T Intellectual Property I, L.P. | Method and system for closed caption processing |
JP4623135B2 (ja) | 2008-05-08 | 2011-02-02 | 株式会社デンソー | 画像認識装置 |
JP4870820B2 (ja) | 2008-05-28 | 2012-02-08 | 京セラ株式会社 | 通信端末、検索サーバ及び通信システム |
US8509571B2 (en) | 2008-06-11 | 2013-08-13 | Panasonic Corporation | Pattern recognition apparatus, pattern recognition method, image processing apparatus, and image processing method |
US8195689B2 (en) | 2009-06-10 | 2012-06-05 | Zeitera, Llc | Media fingerprinting and identification system |
US9367544B2 (en) | 2008-06-18 | 2016-06-14 | Gracenote, Inc. | Method for efficient database formation and search on media devices acting synchronously with television programming |
JP4640456B2 (ja) | 2008-06-25 | 2011-03-02 | ソニー株式会社 | 画像記録装置、画像記録方法、画像処理装置、画像処理方法、プログラム |
US20110313856A1 (en) | 2008-08-18 | 2011-12-22 | Ipharro Media Gmbh | Supplemental information delivery |
KR101380777B1 (ko) | 2008-08-22 | 2014-04-02 | 정태우 | 영상에 포함된 객체를 색인하는 방법 |
WO2010026745A1 (ja) | 2008-09-02 | 2010-03-11 | パナソニック株式会社 | コンテンツ表示処理装置及びコンテンツ表示処理方法 |
AU2009287465B2 (en) | 2008-09-17 | 2014-09-11 | Panasonic Corporation | Recording medium, playback device, and integrated circuit |
JP2010164901A (ja) | 2009-01-19 | 2010-07-29 | Sharp Corp | Osd信号処理装置、av装置、映像表示システム、及びosd表示方法 |
JP4725690B2 (ja) | 2009-01-23 | 2011-07-13 | 日本電気株式会社 | 映像識別子抽出装置 |
US8260055B2 (en) * | 2009-03-27 | 2012-09-04 | The Nielsen Company (Us), Llc | Methods and apparatus for identifying primary media content in a post-production media content presentation |
CN102461066B (zh) * | 2009-05-21 | 2015-09-09 | 数字标记公司 | 鉴别内容信号的方法 |
JP5434265B2 (ja) | 2009-05-22 | 2014-03-05 | 三菱電機株式会社 | 領域分類装置、画質改善装置、映像表示装置、およびそれらの方法 |
JP5320204B2 (ja) | 2009-07-31 | 2013-10-23 | 日本電信電話株式会社 | 情報端末装置、情報検索方法、および情報検索プログラム |
US8421921B1 (en) | 2009-07-31 | 2013-04-16 | Pixelworks, Inc. | Post processing displays with on-screen displays |
US8582952B2 (en) | 2009-09-15 | 2013-11-12 | Apple Inc. | Method and apparatus for identifying video transitions |
US8327407B2 (en) | 2009-10-27 | 2012-12-04 | Sling Media, Inc. | Determination of receiving live versus time-shifted media content at a communication device |
US8682145B2 (en) | 2009-12-04 | 2014-03-25 | Tivo Inc. | Recording system based on multimedia content fingerprints |
US20110137976A1 (en) | 2009-12-04 | 2011-06-09 | Bob Poniatowski | Multifunction Multimedia Device |
KR20110088334A (ko) | 2010-01-28 | 2011-08-03 | 삼성전자주식회사 | 3차원 멀티미디어 서비스를 제공하기 위한 데이터스트림 생성 방법 및 장치, 3차원 멀티미디어 서비스를 제공하기 위한 데이터스트림 수신 방법 및 장치 |
US8355910B2 (en) | 2010-03-30 | 2013-01-15 | The Nielsen Company (Us), Llc | Methods and apparatus for audio watermarking a substantially silent media content presentation |
US9264785B2 (en) * | 2010-04-01 | 2016-02-16 | Sony Computer Entertainment Inc. | Media fingerprinting for content determination and retrieval |
JP5594002B2 (ja) | 2010-04-06 | 2014-09-24 | ソニー株式会社 | 画像データ送信装置、画像データ送信方法および画像データ受信装置 |
JP5111627B2 (ja) | 2010-04-06 | 2013-01-09 | キヤノン株式会社 | 映像処理装置及び映像処理方法 |
JP5960133B2 (ja) | 2010-07-12 | 2016-08-02 | コーニンクレッカ フィリップス エヌ ヴェKoninklijke Philips N.V. | 3dビデオ放送における補助データ |
JP2012027687A (ja) | 2010-07-23 | 2012-02-09 | Casio Comput Co Ltd | 画像処理装置及びプログラム |
US9142122B2 (en) | 2010-11-25 | 2015-09-22 | Panasonic Intellectual Property Corporation Of America | Communication device for performing wireless communication with an external server based on information received via near field communication |
US9106871B2 (en) | 2011-04-01 | 2015-08-11 | Google Inc. | Detecting which channel a local STB is displaying using a local tuner |
JP2012231383A (ja) | 2011-04-27 | 2012-11-22 | Nippon Telegr & Teleph Corp <Ntt> | Iptvサービスを利用した情報表示制御装置、表示用情報提供サーバ、情報表示制御方法、情報配信方法、情報表示制御用プログラム、および情報配信用プログラム |
KR101778530B1 (ko) | 2011-06-14 | 2017-09-15 | 삼성전자 주식회사 | 영상 처리 방법 및 장치 |
KR101828342B1 (ko) | 2011-08-10 | 2018-02-12 | 삼성전자 주식회사 | 방송수신장치, 방송신호 관련 연관정보 제공방법 및 서버 |
KR101995425B1 (ko) | 2011-08-21 | 2019-07-02 | 엘지전자 주식회사 | 영상 표시 장치, 단말 장치 및 그 동작 방법 |
US8805827B2 (en) | 2011-08-23 | 2014-08-12 | Dialogic (Us) Inc. | Content identification using fingerprint matching |
US20130111514A1 (en) | 2011-09-16 | 2013-05-02 | Umami Co. | Second screen interactive platform |
JP5948773B2 (ja) | 2011-09-22 | 2016-07-06 | ソニー株式会社 | 受信装置、受信方法、プログラム、及び情報処理システム |
KR101310943B1 (ko) | 2011-09-26 | 2013-09-23 | (주)엔써즈 | 방송 콘텐츠와 연관된 콘텐츠 연관 정보를 제공하는 시스템 및 방법 |
US20130094590A1 (en) | 2011-10-12 | 2013-04-18 | Vixs Systems, Inc. | Video decoding device for extracting embedded metadata and methods for use therewith |
US9626798B2 (en) | 2011-12-05 | 2017-04-18 | At&T Intellectual Property I, L.P. | System and method to digitally replace objects in images or video |
US9204199B2 (en) | 2011-12-06 | 2015-12-01 | Lg Electronics Inc. | Image display apparatus and methods for operating the same |
JP2013125191A (ja) | 2011-12-15 | 2013-06-24 | Canon Inc | 映像表示装置、映像表示方法及びプログラム |
US8949872B2 (en) * | 2011-12-20 | 2015-02-03 | Yahoo! Inc. | Audio fingerprint for content identification |
US20120092248A1 (en) | 2011-12-23 | 2012-04-19 | Sasanka Prabhala | method, apparatus, and system for energy efficiency and energy conservation including dynamic user interface based on viewing conditions |
US9578378B2 (en) | 2012-01-05 | 2017-02-21 | Lg Electronics Inc. | Video display apparatus and operating method thereof |
WO2013112991A2 (en) | 2012-01-27 | 2013-08-01 | Xumo Llc | System and method of augmenting linear broadcast advertising |
US9319740B2 (en) | 2012-02-07 | 2016-04-19 | Turner Broadcasting System, Inc. | Method and system for TV everywhere authentication based on automatic content recognition |
JP5867133B2 (ja) | 2012-02-10 | 2016-02-24 | 富士通株式会社 | 画像処理装置、画像処理方法及び画像処理プログラム |
JP6294238B2 (ja) | 2012-02-10 | 2018-03-14 | エルジー エレクトロニクス インコーポレイティド | 映像表示装置及びその動作方法 |
WO2013130988A1 (en) | 2012-03-02 | 2013-09-06 | Care Cam Innovations, Llc | Apparatus, method and computer-readable storage medium for media processing and delivery |
US9516365B2 (en) | 2012-03-05 | 2016-12-06 | Lg Electronics Inc. | Video display apparatus and operating method thereof |
US9292894B2 (en) | 2012-03-14 | 2016-03-22 | Digimarc Corporation | Content recognition and synchronization using local caching |
JP6514505B2 (ja) * | 2012-05-23 | 2019-05-15 | サターン ライセンシング エルエルシーSaturn Licensing LLC | 情報処理装置、情報処理方法、並びにプログラム |
US8645994B2 (en) | 2012-06-28 | 2014-02-04 | Microsoft Corporation | Brand detection in audiovisual media |
CN103581705A (zh) * | 2012-11-07 | 2014-02-12 | 深圳新感易搜网络科技有限公司 | 视频节目识别方法和系统 |
JP5820986B2 (ja) | 2013-03-26 | 2015-11-24 | パナソニックIpマネジメント株式会社 | 映像受信装置及び受信映像の画像認識方法 |
US20150026718A1 (en) | 2013-07-19 | 2015-01-22 | United Video Properties, Inc. | Systems and methods for displaying a selectable advertisement when video has a background advertisement |
US9456237B2 (en) | 2013-12-31 | 2016-09-27 | Google Inc. | Methods, systems, and media for presenting supplemental information corresponding to on-demand media content |
US10616613B2 (en) * | 2014-07-17 | 2020-04-07 | Panasonic Intellectual Property Management Co., Ltd. | Recognition data generation device, image recognition device, and recognition data generation method |
-
2015
- 2015-08-19 US US15/302,460 patent/US10200765B2/en active Active
- 2015-08-19 WO PCT/JP2015/004112 patent/WO2016027457A1/ja active Application Filing
- 2015-08-19 CN CN201580020600.7A patent/CN106233746B/zh active Active
- 2015-08-19 EP EP15832982.1A patent/EP3185577B1/en active Active
- 2015-08-19 JP JP2016543816A patent/JP6432047B2/ja active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004341940A (ja) * | 2003-05-16 | 2004-12-02 | Fujitsu Ltd | 類似画像検索装置、類似画像検索方法、および類似画像検索プログラム |
JP2011059504A (ja) * | 2009-09-11 | 2011-03-24 | Victor Co Of Japan Ltd | コンテンツ同定方法、コンテンツ同定システム、コンテンツ検索装置及びコンテンツ利用装置 |
WO2014006903A1 (ja) * | 2012-07-06 | 2014-01-09 | パナソニック株式会社 | コンテンツ制御方法、コンテンツ制御装置およびプログラム |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2019511138A (ja) * | 2016-02-29 | 2019-04-18 | グレースノート インコーポレイテッド | 複数一致検出とロケーションに基づく曖昧性排除とを用いたメディアチャネル識別 |
JP2019201415A (ja) * | 2016-02-29 | 2019-11-21 | グレースノート インコーポレイテッド | 複数一致検出とロケーションに基づく曖昧性排除とを用いたメディアチャネル識別 |
CN111133765A (zh) * | 2017-09-13 | 2020-05-08 | 尼尔森(美国)有限公司 | 通过自动内容识别进行冷匹配 |
KR20200053543A (ko) * | 2017-09-13 | 2020-05-18 | 더 닐슨 컴퍼니 (유에스) 엘엘씨 | 자동 컨텐츠 인식에 의한 콜드 매칭 |
KR20200103117A (ko) * | 2017-09-13 | 2020-09-01 | 더 닐슨 컴퍼니 (유에스) 엘엘씨 | 자동 콘텐츠 인식에 대한 광고 프레임들의 플래깅 |
JP2020533918A (ja) * | 2017-09-13 | 2020-11-19 | ザ ニールセン カンパニー ユーエス リミテッド ライアビリティ カンパニー | 自動コンテンツ認識によるコールドマッチング |
JP2020533920A (ja) * | 2017-09-13 | 2020-11-19 | ザ ニールセン カンパニー ユーエス リミテッド ライアビリティ カンパニー | 自動コンテンツ認識のための広告フレームのフラッギング |
KR102272767B1 (ko) | 2017-09-13 | 2021-07-06 | 로쿠, 인코퍼레이티드 | 자동 콘텐츠 인식에 대한 광고 프레임들의 플래깅 |
JP2021122117A (ja) * | 2017-09-13 | 2021-08-26 | ザ ニールセン カンパニー ユーエス リミテッド ライアビリティ カンパニー | 自動コンテンツ認識のための広告フレームのフラッギング |
CN111133765B (zh) * | 2017-09-13 | 2022-02-15 | 六科股份有限公司 | 通过自动内容识别进行冷匹配 |
US11303974B2 (en) | 2017-09-13 | 2022-04-12 | Roku, Inc. | Flagging advertisement frames for automatic content recognition |
KR102406304B1 (ko) * | 2017-09-13 | 2022-06-10 | 로쿠, 인코퍼레이티드 | 자동 컨텐츠 인식에 의한 콜드 매칭 |
JP7116786B2 (ja) | 2017-09-13 | 2022-08-10 | ロク インコーポレイテッド | 自動コンテンツ認識によるコールドマッチング |
JP7240437B2 (ja) | 2017-09-13 | 2023-03-15 | ロク インコーポレイテッド | 自動コンテンツ認識のための広告フレームのフラッギング |
Also Published As
Publication number | Publication date |
---|---|
US20170026718A1 (en) | 2017-01-26 |
JPWO2016027457A1 (ja) | 2017-06-01 |
CN106233746B (zh) | 2019-07-09 |
EP3185577B1 (en) | 2018-10-24 |
EP3185577A4 (en) | 2017-06-28 |
US10200765B2 (en) | 2019-02-05 |
JP6432047B2 (ja) | 2018-12-05 |
EP3185577A1 (en) | 2017-06-28 |
CN106233746A (zh) | 2016-12-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP6432047B2 (ja) | コンテンツ認識装置およびコンテンツ認識方法 | |
JP6471359B2 (ja) | 認識データ生成装置、画像認識装置および認識データ生成方法 | |
US10798448B2 (en) | Providing restricted overlay content to an authorized client device | |
US10057657B2 (en) | Content replacement with onscreen displays | |
CN109906611B (zh) | 用于内容特征图谱化的特征图谱布局的服务器及介质 | |
JP2020025322A (ja) | 自動コンテンツ認識フィンガープリントシーケンス照合 | |
US11039222B2 (en) | Targeted content placement using overlays | |
US20100217671A1 (en) | Method and apparatus for extracting advertisement keywords in association with situations of video scenes | |
US11330329B2 (en) | System and method for detecting and classifying direct response advertisements using fingerprints | |
CN105430429A (zh) | 一种媒体数据的优化方法和装置 | |
CN110099298B (zh) | 多媒体内容处理方法及终端设备 | |
JP6424339B2 (ja) | 表示制御装置、および表示制御方法 | |
US20160125472A1 (en) | Gesture based advertisement profiles for users | |
WO2016172715A1 (en) | Content replacement with onscreen displays | |
US20150106828A1 (en) | Method and apparatus for identifying point of interest and inserting accompanying information into a multimedia signal |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 15832982 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2016543816 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 15302460 Country of ref document: US |
|
REEP | Request for entry into the european phase |
Ref document number: 2015832982 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2015832982 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |