JP2006157108A - Video recording / playback device - Google Patents

Abstract

A video recording / reproducing apparatus capable of more easily searching for a desired scene is provided.
A video recording / reproducing apparatus 10 records a video signal input from an external device 34 on a recording medium 14. Of the recorded video, an index is assigned to a portion where a predetermined keyword occurs. The presence / absence and generation amount of the keyword are detected by the keyword generation amount detection unit 16, and the detection result is output to the level determination unit 24. The level determination unit 24 generates index information to be assigned to the occurrence location of the keyword and outputs it to the recording unit 12 only when the keyword generation amount is equal to or greater than a predetermined reference value. The recording unit 12 records the generated index information on the recording medium 14 together with the video signal.
[Selection] Figure 1

Description

  The present invention relates to a video recording / reproducing apparatus that records video on a recording medium and reproduces the video recorded on the recording medium.

  Some video recording / playback apparatuses that record and play back video detect a location where a preset keyword occurs in the video to be recorded, and add an index to the location where the keyword occurs.

  For example, Patent Document 1 discloses a video processing apparatus that recognizes audio included in video, extracts a keyword from the recognized audio, and records an index indicating the occurrence of the keyword on a recording medium together with the video. Yes. Patent Document 2 also discloses a scene search in which words specific to a program to be recorded are extracted, an index for video search is created, and the video search index is recorded on a recording medium together with a video signal and an audio signal. An apparatus is disclosed.

  Thus, by assigning an index to the appearance location of a specific keyword, a desired scene can be easily searched. In other words, if you register a term or person's name that you are interested in as a keyword, you can view the indexing status and see the related situation for that term or person's name without having to watch all the videos recorded on the recording medium. You can search for a scene.

JP 2002-171481 A JP 2000-236494 A

  However, all of the above techniques determine whether or not an index can be assigned based on “the presence / absence of a keyword”. Therefore, the keyword is generated only once, and an index is given to a scene that is not related to the keyword. As a result, it may not be possible to clearly determine where the desired scene is based on the index assignment status. Also, if an index is assigned even if it occurs only once, the number of indexes tends to increase. Usually, the user views the video related to the indexed portion and determines whether or not the scene is a scene related to the keyword. However, as the number of indexed portions increases, the number of viewing scenes increases, and as a result, the scene search efficiency decreases.

  In the above technique, the same index is assigned to a scene in which a keyword occurs once within a unit time and a scene in which a keyword occurs multiple times. Therefore, it is not possible to determine which scene is closely related to the keyword only by looking at the index assignment status.

  Accordingly, an object of the present invention is to provide a video recording / playback apparatus that can more easily search for a desired scene.

  The video recording / reproducing apparatus of the present invention is a video recording / reproducing apparatus for recording a video on a recording medium, in which a storage unit for storing a keyword designated by a user in advance and an input video are analyzed to obtain a predetermined A detecting means for detecting a keyword generation amount per unit time, a determination means for determining whether or not a keyword generation amount is greater than or equal to a reference value every predetermined unit time, and a keyword generation amount being greater than or equal to a reference value And index recording means for recording the index information indicating the occurrence of the keyword on the recording medium together with the position information where the keyword is generated.

  In a preferred aspect, the detection means calculates the generation amount based on the number of occurrences of the keyword within a predetermined time. When the keyword is generated by voice, it is desirable that the detection means calculates the generation amount based on the number of times the keyword is generated and the volume of the keyword at each occurrence. When the keyword is generated as a character, it is preferable that the detection unit calculates the generation amount based on at least one of the generated keyword color, display time, and screen occupancy rate, and the number of occurrences of the keyword. .

  In another preferred aspect, the index recording means also records a keyword generation amount as index information. Further, it is desirable to provide a presentation means for presenting an index display screen showing a keyword occurrence location and the amount of occurrence based on the index information recorded on the recording medium to the user. Furthermore, it is desirable that the playback start position of the video recorded on the recording medium can be indicated on the index display screen.

  Another video recording / reproducing apparatus according to the present invention is a video recording / reproducing apparatus for recording a video on a recording medium, in which a storage unit for storing a keyword designated by a user in advance and an input video are analyzed. Detecting means for detecting a keyword generation amount per predetermined unit time; index recording means for recording index information indicating a keyword generation position and generation amount on a recording medium in association with the keyword generation position; The present invention is characterized in that it has presentation means for presenting an index display screen showing a keyword occurrence location and the generation amount to the user based on the index information recorded on the recording medium.

  Another video recording / reproducing apparatus according to the present invention is a video recording / reproducing apparatus for recording a video composed of a plurality of scenes on a recording medium, a storage unit for storing a keyword designated in advance by a user, and an input Analyzing the recorded video to detect a keyword generation amount per predetermined unit time, a determination unit determining whether the keyword generation amount exceeds a reference value every predetermined unit time, a reference Video recording means for recording a video related to the generation scene of the keyword on a recording medium when the value is greater than or equal to the value;

Another video recording / playback apparatus according to the present invention is a video recording / playback apparatus for playing back video recorded on a recording medium, and is recorded in advance on a storage unit for storing a keyword designated by a user, and on the recording medium. Detecting means for analyzing the generated video and detecting a keyword generation amount per predetermined unit time, a determination means for determining whether or not the keyword generation amount per predetermined unit time is greater than or equal to a reference value, Video deletion means for deleting a video whose generated amount is less than a reference value from a recording medium.
Here, the video refers to a signal composed of audio and moving images. The recording medium refers to all recording media capable of recording digital video signals, such as hard disks, as well as optical disks such as DVD, Blu-ray, and HD-DVD.

  According to the present invention, an index is assigned only to a portion where there is a keyword generation amount equal to or greater than a predetermined reference value. In other words, an index is assigned only to a portion that is reliably related to the keyword. As a result, a desired scene can be easily retrieved by looking at the index assignment status.

  Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a schematic block diagram of a video recording / reproducing apparatus 10 according to an embodiment of the present invention. The video recording / reproducing apparatus 10 is an apparatus that records and reproduces video on a recording medium 14 such as an optical disk such as a DVD, Blu-ray, or HD-DVD, or a hard disk. Therefore, a DVD recorder, a hard disk recorder, etc. correspond. The video recording / reproducing apparatus 10 can add an index to the recorded video. An index is a mark attached to a portion of the recorded video where attention is particularly needed. In the present embodiment, an index is attached to a place where a preset keyword is generated. Hereinafter, the video recording / reproducing apparatus 10 will be described in detail.

  The external device 34 corresponds to all devices capable of outputting some video signal, such as a television receiver, another video playback device, and a TV tuner. The video signal output from the external device 34 is sent to the recording unit 12 and the keyword generation amount detection unit 16. The recording unit 12 records the input video signal, that is, the moving image signal and the audio signal on the recording medium 14. The moving image signal and the audio signal are input to the recording unit 12 after A / D conversion MPEG encoding processing, if necessary. The recording unit 12 also records index information generated by an index generation unit described later on the recording medium 14.

  The keyword generation amount detection unit 16 divides the input video signal into scenes for each predetermined unit time (for example, 1 minute), and detects whether or not a keyword is generated for each scene and the generation amount thereof. The video signal may be input from an external device during video recording, or may be input as a video signal reproduced by the reproduction unit 30 described later. The keywords are set in advance by the user, and are stored in the keyword storage unit 26 including a RAM or the like.

  The keyword is detected by the voice recognition unit and the image recognition unit. The voice recognition unit 18 analyzes the input voice signal and determines whether or not a voice indicating a keyword is included in the input voice. When the voice indicating the keyword is included, the recognition result and the volume of the voice signal related to the keyword are output to the level determination unit 22.

  The image recognition unit 20 analyzes the input moving image signal and determines whether or not character information indicating a keyword is included in the moving image signal. This image analysis method will be described in detail later. When it is determined that character information indicating a keyword is included, the display time, color, screen occupancy rate, and the like of the character information are calculated and output to the level determination unit 22.

  Based on the recognition results output from the voice recognition unit 18 and the image recognition unit 20, the level determination unit 22 calculates a keyword generation amount in the scene, and determines the scene level based on the generation amount. The generation amount calculation method will be described in detail later. Basically, the generation amount is calculated based on the keyword occurrence frequency. Then, the level of each scene is determined based on the obtained generation amount. The level is a parameter indicating the amount of keywords generated in each scene. The larger the numerical value, the larger the amount of keywords generated. If a plurality of keywords are set, the generation amount and level are determined for each keyword. The level determination result, the voice recognition result, and the image recognition result are output to the index generation unit 24.

  The index generation unit 24 generates index information to be given to a scene where a keyword has occurred. Whether or not index information can be generated is determined based on the level of each scene. If the level of the target scene, that is, the amount of generated keywords is equal to or less than a predetermined reference value, no index information is generated because no index is assigned to the scene. On the other hand, if the level is equal to or higher than the reference value, the index information of the scene is generated and output to the recording unit 12. The index information includes a keyword generation position (scene), a generation amount of each keyword, a generation mode (whether it is generated by voice or characters), and the like. The generated information is output to the recording unit 12 and recorded on the recording medium in association with the video signal of the scene.

  The reproduction unit 30 reproduces the video signal recorded on the recording medium 14. The reproduced video signal is usually output to a display 36 such as a television monitor. In addition, when an index is already assigned to the video recorded on the recording medium 14, the reproduced video signal is output to the keyword generation amount detection unit 16.

  The reproducing unit 30 can also reproduce the index information recorded on the recording medium 14, and the reproduced index information is output to the index screen generating unit 32. The index screen generation unit 32 generates an index screen indicating an index generation status based on the index information. The index screen is a screen in which keyword occurrence locations, generation amounts, and the like are visually expressed using time bars, bar graphs, and the like. The generated index screen is output to the display 36.

  The control unit 28 controls the entire video recording / reproducing apparatus 10 in accordance with a user instruction input via the user interface (U / I) 29. The user interface 29 corresponds to a remote controller or a switch, and the user operates the video recording / reproducing apparatus 10 by operating these remote controller or switch. The control unit 28 performs video recording, reproduction, index display screen display, and the like in accordance with an instruction from the user.

  Next, keyword detection will be described in detail. There are cases where the keyword is generated as a voice and when the keyword is generated as a character (image). Various known voice recognition techniques can be used to detect keywords generated by voice. For example, the voice recognition unit 18 calculates the frequency or the like of the input voice signal and acquires the voice feature pattern of the voice signal. Subsequently, the obtained voice feature pattern is compared with a keyword voice feature pattern recorded in advance, and if both match or approximate, the voice signal is recognized as a keyword. . If the audio signal is a keyword, the volume of the audio signal is detected. Then, the recognition result and the volume are output to the level determination unit 22.

  Keywords generated in characters (images) are detected by the image recognition unit 20. For this detection, various known image analysis techniques such as pattern matching can be used. Pattern matching is one of image analysis techniques for determining the matching between a feature pattern obtained for an input image and a feature pattern of a reference image stored in advance. A specific analysis method using this pattern matching will be described with reference to FIG. FIG. 2A is a diagram showing an example of an input image (hereinafter referred to as “target image 40”), and FIG. 2B is a reference character stored in the keyword storage unit 26 including a RAM or the like. It is a figure which shows the image.

  First, the image recognition unit 20 performs edge detection on the target image 40. In edge detection, the luminance value of each pixel is obtained. If the luminance difference between adjacent pixels is equal to or greater than a predetermined value, the pixel is extracted as an edge. By this edge detection, the contour of the target image 40 can be acquired. Next, a pixel corresponding to a corner portion is extracted as a feature point 40a from the outline. For example, in the case of an image of the character “ki”, 20 feature points 40a can be acquired as shown in FIG. Next, the feature point 40 a of the target image 40 is compared with the feature point 42 a of the reference character image 42 stored in the keyword storage unit 26. This comparison is performed in the following procedure. First, of the plurality of feature points 40a and 42a, reference points 40b and 42b serving as a reference are obtained for the target image 40 and the reference character image 42, respectively. Next, the target image and the reference character image are overlaid so that the reference points 40b and 42b coincide. Then, in a state where the reference points 40b and 42b are made coincident with each other, the target image 40 or the reference character image is set so that the deviation amount between the feature point 40a of the target image 40 and the feature point 42a of the reference character image 42 is minimized. Adjust the size and angle of 42. If the amount of deviation between the feature points 40a and 42a is minimized, it is determined whether or not the amount of deviation at that time is less than a predetermined reference value. When the amount of deviation is less than the reference value, the target image 40 is determined to be a character indicated by the reference character image 42.

  The image recognition unit 20 extracts characters included in the input moving image signal through image analysis such as pattern matching, and detects the presence or absence of a character string indicating a keyword. At this time, the presence / absence of a keyword is determined based on the continuity of characters so as not to make a mistake between a character with a similar shape such as “B” in katakana and a square of the graphic. If a character string indicating a keyword is included, the screen occupation rate, display time, and color of the character string are also detected and output to the level determination unit 22.

  Here, the screen occupation ratio is the ratio of the area occupied by the keyword to the screen. The area occupied by the keyword is the smallest rectangular area that can contain the entire keyword. For example, in the image shown in FIG. 3, when the keyword 50 is “AAAA”, the occupied area of the keyword 50 is the area of the rectangle 52 indicated by the broken line in FIG. The screen occupation ratio is the area of the rectangle 52 with respect to the area of the entire screen. For example, when the screen is 720 × 480 dots and the keyword occupation area (area of the rectangle 52) is 216 × 46 dots, the screen occupation ratio is (216 × 46) / (720 × 480) × 100% ≈3% It becomes.

  The image recognition unit 20 also obtains the keyword color and outputs it to the level determination unit 22. The keyword color is output as a numerical value using a color model such as RGB. At this time, when a plurality of colors are included in one keyword, the RGB value of each color is output. For example, when “DVD” is red in the keyword “DVD recorder” and “recorder” is black, both the RGB value indicating red and the RGB value indicating black are output to the level determination unit 22.

  Note that the keyword detection method described here is merely an example, and other methods may naturally be used as long as the presence or absence of the keyword can be detected. In this embodiment, in addition to the presence / absence of a keyword, the volume, display time, screen occupancy, color, and the like are also detected, but it is not always necessary to detect them. In the present embodiment, both sound generation and image generation are detected, but only one of them may be detected. Furthermore, in addition to the set keyword, an alias or a synonym of the keyword may be detected as a keyword. For example, when the keyword is “DVD”, the word “digital video disc” may be detected as a keyword in addition to “DVD”. In this case, the nickname of each term may be registered by the user himself or may be previously stored as a dictionary.

  Next, the level determination in the level determination unit 22 will be described in detail. As described above, the level determination unit 22 receives the keyword detection results, volume, screen occupancy, and the like by the voice recognition unit 18 and the image recognition unit 20. Based on these pieces of information, the level determination unit 22 first calculates the keyword generation amount for each scene. The amount of keywords generated is determined by multiplying the number of keyword occurrences by a weighting factor based on the volume and screen occupancy.

  Specifically, the keyword generation amount Da when the keyword is generated by voice is calculated from the number of occurrences of the keyword and a coefficient based on the sound volume at each generation. For example, a keyword occurs n times in one scene, and the volume at each occurrence is Ai (i = 1, 2,... N). The keyword generation amount Da at this time is a product sum of a reference value B (for example, 1) added every time the keyword is generated and a coefficient Kia based on the volume Ai. That is, the keyword generation amount Da is Da = ΣB × Kia. Various settings are possible as the set value of the volume coefficient Kia. In this embodiment, when the volume is less than a predetermined first reference value (for example, 24 dB), 0, the first reference value or more and the second reference The value is set to 1 when the value is smaller than the value (for example, 66 dB), and 1.5 when the value is equal to or larger than the second reference value. That is, when the volume is less than the first reference value, the keyword generation amount Da is 0, and processing is performed assuming that no keyword is generated. Conversely, when the keyword is generated at a high volume, the keyword generation amount Da is also increased. This is because when a keyword occurs at a low volume, the scene often has a small relationship with the keyword, and it is not desirable to add an index to such a scene. On the other hand, when a keyword occurs at a high volume, the scene often has a large relationship with the keyword, and it is desirable to make the scene more easily indexed.

  The keyword generation amount Dv when the keyword is generated in characters is a reference value B (for example, 1 or the like) added every time the keyword is generated, and a weighting factor Kib (hereinafter referred to as “time factor” based on the display time at each occurrence. )), A weighting factor Kic based on the screen occupancy rate (hereinafter referred to as “occupancy factor”), and a weighting factor Kid based on color (hereinafter referred to as “color coefficient”). That is, Dv = ΣB × Kib × Kic × Kic. The time coefficient Kib is 0 when the display time is less than a predetermined first reference value (for example, 1 sec), 1 when the display time is not less than the first reference value and less than the second reference value (for example, 3 sec), and the second reference value. In the above case, 1.5 is set. The occupation coefficient Kic is 0 when the screen occupation ratio is less than a predetermined first reference value (for example, 2%), and 1 when the screen occupation ratio is greater than or equal to the first reference value and less than the second reference value (for example, 5%). In the case of the second reference value or more, it is set to 1.5 or the like. The color coefficient Kid is set to 1 for black (RGB = 0, 0, 0), 1.5 for other colors, and the like. By using such a coefficient, when the character string of the keyword is displayed only for a moment or when it is displayed very small, it is processed that the keyword has not occurred. On the other hand, when the keyword is displayed for a long time, when it is displayed in a large size, or when it is displayed in color characters, the value of the keyword generation amount Dv increases, and it becomes easier to add an index.

  The level determination unit 22 determines the levels La and Lv of each scene based on the two types of calculated keyword generation amounts Da and Dv, respectively. Levels La and Lv are set in a plurality of stages, such as level 0, level 1, level 2... In order from the lowest keyword generation amount Da and Dv. The level determination unit 22 outputs the level determination result to the index generation unit 24.

  The index generation unit 24 adds the two types of levels La and Lv, and determines whether the value is equal to or greater than a predetermined reference value. When the keyword generation amounts Da and Dv are equal to or greater than a predetermined value, in other words, the index information of the scene is given. The reference value at this time may be constant or may be set as appropriate by the user. Or you may make it change according to the kind (for example, a music program, news, a sports broadcast, etc.) of the image | video to be recorded.

  In this way, by assigning an index only when the level (keyword generation amount) is equal to or greater than a predetermined value, it is possible to assign an index only to a scene that is more reliably related to the keyword. That is, even though the scene is originally not related to the keyword, a video character may ask for the keyword or erroneously display a telop related to the keyword. In addition, due to a voice / image recognition error, a word different from the registered keyword may be detected as a keyword. Even in such a case, when an index is assigned, the number of scenes to which the index is assigned increases, and the user cannot know which scene is really related to the keyword. However, as in the present embodiment, by assigning an index only when the level is equal to or higher than a predetermined value, that is, when keywords are frequently generated, it is possible to reliably assign an index only to a scene related to the keyword. As a result, a desired scene can be easily searched.

  Note that the method for calculating the keyword generation amounts Da and Dv described here is merely an example, and naturally, it may be calculated by another method. Therefore, when calculating the keyword generation amount, it is not necessary to use the volume, screen occupancy, etc. in addition to the number of keyword generations, and the keyword generation amount may be calculated based only on the number of keyword generations. Further, even when various weighting factors are used, various forms other than the above can be used as the set value. Further, it may be calculated using a weighting factor other than the above.

  Next, an index display screen displayed when the index information is reproduced will be described with reference to FIGS. 4 to 6 are diagrams illustrating an example of the index display screen 60. FIG. FIG. 4 is a diagram of the index display screen 60 when the indexing status is displayed with a time bar. In this example, a band-shaped time bar 62 is displayed for each keyword. The length of the time bar 62 indicates the video time. In the time bar 62, a mark 64 having a different color from the others is attached to the portion corresponding to the indexing time. Therefore, in the example of FIG. 4, the keyword A is generated at the times t1 and t4, the keyword B is generated at the times t1, t2, and t3, and the index is given. In addition, the color of the mark 64 attached to the indexing location is set according to the index level (keyword generation amount). In the example of FIG. 4, the color of the mark 64 is darker as the level is higher, and lighter as the level is lower. That is, as the darker color mark 64 is added, the amount of generated keywords is larger, and the scene is deeply related to the set keyword.

  In this way, by indicating the index assignment status with the time bar 62, the user can find out which part of the total video time the keyword has been generated, in other words, the content in which the user is interested. It is easy to check whether Also, by displaying the time bar 62 for each keyword, it is possible to easily confirm which keyword is generated at which time. Furthermore, by assigning the mark 64 by color according to the level, it is possible to easily estimate the relevance between each indexed scene and the keyword. As a result, the user can easily select a desired scene.

  The index display screen 60 also has a function as a playback position instruction screen. That is, if the user indicates the time to be reproduced using the time bar 62, the video is reproduced from the selected time. For example, if the user indicates the mark 64 at time t1, the video is reproduced from time t1. Thus, by using the index display screen 60 as a playback position instruction screen, the user can easily view a desired scene.

  FIG. 5 is a diagram of another index display screen 60. The index display screen 60 shows an index assignment status with a bar graph 66 in which the horizontal axis represents time and the vertical axis represents level. The data of the bar graph 66 (bar 68) is displayed in a different color for each keyword. That is, in the example of FIG. 5, it can be seen that the keyword A occurs at times t1, t2, t3, and t4, and the keyword B occurs at times t1 and t2. The keyword level is indicated by the height of the bar 68. Therefore, in the example of FIG. 5, it can be seen that many keywords A are generated at time t3 and many keywords B are generated at time t2. By indicating the indexing status with such a bar graph 66, the level (keyword generation amount) in each scene can be confirmed more easily. Note that this index display screen can also be used as a playback position instruction screen. When a desired playback start time is indicated on the bar graph 66, a video is played from the specified time.

  FIG. 6 is a diagram of another index display screen 60. Also on this index display screen 60, the indexing status is indicated by a bar graph 70 in which the horizontal axis indicates time and the vertical axis indicates level. However, on the index display screen 60, the color of the bar 72 is changed depending on the generation mode so that the generation mode of each keyword can be understood in more detail. For example, when a keyword is generated by voice, the level is indicated by a red bar, and when the keyword is generated by a letter, the level is indicated by a blue bar. Furthermore, when a keyword occurs in a special manner, for example, when it occurs at a volume higher than the reference value, or when the screen occupancy rate of characters indicating the keyword is higher than the reference value, it corresponds to the keyword generation in that special correspondence The level to be displayed is displayed in a different color. In the example of FIG. 6, it can be seen that at time t1, many keywords are generated by voice and some keywords are generated by characters. Also, at time t3, it can be seen that keywords are generated in a special manner. By using such an index display screen, the occurrence status of keywords in each scene can be understood in more detail. This index screen may be provided for each keyword, or a value obtained by adding the levels of all keywords may be displayed as a bar graph.

  Next, a flow when recording (recording a video signal) using the video recording / reproducing apparatus 10 will be described with reference to FIGS. The user sets a keyword via the user interface 29 before starting recording (S12). The set keyword is stored and held in the keyword storage unit 26. Next, recording is started in accordance with a user instruction (S14). The video signal input from the external device 34 is input to the recording unit 12 and the keyword generation amount detection unit 16 after being subjected to A / D conversion, MPEG encoding processing, and the like as necessary. The recording unit 12 sequentially records the input video signal on the recording medium 14 (S16). It is desirable that the video signal input from the external device 34 is input to the keyword generation amount detection unit 16 immediately after A / D conversion and noise reduction processing. With such a configuration, the data amount of the video signal is large, and the S / N ratio is good because it is not compressed. Furthermore, if the same video signal is sent to the keyword generation amount detection unit 16 and the MPEG encoding processing unit (not shown), the degradation of the video signal can be reduced.

  In parallel with the video recording, an index is generated and recorded (S18). This flow is shown in FIG. The voice recognition unit 18 and the image recognition unit 20 of the keyword generation amount detection unit 16 divide the input audio signal and video signal into scenes for each unit time, and determine whether or not a keyword is generated in each scene. Judgment is made (S22). This is performed using the voice recognition technology and image recognition technology described above. As a result, if no keyword is generated, the next scene is indexed and recorded (S32, S34). When a keyword is generated, the voice / image recognition result is output to the level determination unit 22. The level determination unit 22 calculates a keyword generation amount based on the sound / image recognition result, and determines the level of the scene (S24). The level determination result is output to the index generation unit 24 together with the sound / image recognition result.

  The index generation unit 24 determines whether or not the scene level is equal to or higher than a predetermined reference value (S26). If the level is less than the predetermined reference value, no index is assigned to the scene, and the process proceeds to index generation / recording of the next scene (S32, S34). If the level is equal to or higher than the predetermined reference value, index information is created based on the sound / image recognition result and the level determination result (S28). The created index information is passed to the recording unit 12. The recording unit 12 records the index information on the recording medium 14 in association with the video signal corresponding to the scene (S30). This flow is repeated for each scene, and when there is no next scene, index generation / recording ends.

  Next, a flow when a new index is added to the video signal recorded on the recording medium will be described with reference to FIG. When assigning an index to a recorded video, the user sets a keyword via the user interface 29 (S36). The set keyword is stored and held in the keyword storage unit 26. Subsequently, the user instructs a keyword search start (S38). Receiving this instruction, the control unit 28 instructs the reproduction unit 30 to reproduce the recorded video. The reproducing unit 30 reproduces the video recorded on the recording medium 14 according to the instruction (S40). However, the playback at this time is not playback for video viewing but playback for keyword search, so playback may be performed at a rate different from normal. Further, the video signal may be reproduced at the limit speed of the processing capability of the DSP in the keyword generation amount detection unit 16 without outputting a moving image or sound. The reproduced video signal is output to the keyword generation amount detection unit 16. Then, after the presence / absence of a keyword or the level is determined, the index generation unit 24 creates index information. Since this flow is the same as in FIG. 8, the description thereof is omitted here. The created index information is additionally recorded on the recording medium 14 by the recording unit 12. If index information for all scenes can be recorded, the process ends (S44).

  Next, the display flow of the index display screen will be described with reference to FIG. The user instructs display of the index display screen via the user interface 29 (S46). Receiving this instruction, the control unit 28 instructs the reproduction unit 30 to reproduce the index information. The reproducing unit 30 reproduces only the index information and passes it to the index screen generating unit 32 (S48). The index screen generation unit 32 creates an index display screen based on various information included in the index information, keyword generation position, level, generation mode, and the like (S50). The display screen created here is a time bar format, a bar graph format, or the like. The created index display screen is displayed on a display 36 such as a monitor. The user instructs a desired reproduction scene while viewing the displayed index display screen (S52). This instruction is made by specifying a position to be reproduced on the time bar or bar graph. The control unit 28 instructs the playback unit 30 to play back video from the scene corresponding to the instructed position. The playback unit 30 plays back the video from the instructed scene and outputs it to the monitor (S54).

  As described above, according to the present embodiment, since the index is assigned only when the keyword generation amount (level) is a certain level or more, the keyword generation amount is high, in other words, the relevance to the keyword is deep. Only scenes can be indexed. Also, by visually displaying the indexing status, the user can easily recognize which scene is closely related to the keyword. Furthermore, since the reproduction start position can be instructed by looking at the application status, a desired scene can be easily appreciated. As a result, a desired scene can be easily searched.

  In this embodiment, whether or not indexing is possible is determined based on the amount of generated keywords. However, whether or not video can be recorded may be determined based on the amount of generated keywords. That is, a video signal input from an external device is temporarily stored in a storage unit such as a buffer memory, and a keyword generation amount for each scene is calculated. If there is a scene in which the keyword generation amount is equal to or greater than a predetermined value, only the scene and scenes for a predetermined time before and after the scene are recorded on the recording medium. In this way, only the desired scene is recorded on the recording medium, so that the user can appreciate the desired scene without searching for the scene after recording. Further, the recording capacity of the recording medium can be saved. In this case, it is desirable to record not only the scene in which the keyword occurred but also scenes for a certain amount of time before and after that. The time may be a fixed value or may be set appropriately by the user. Further, it may be changed depending on the video type (for example, news, sports broadcast, song program, etc.).

  In addition, for a video that has already been recorded on a recording medium, a deletion scene may be determined based on a keyword generation amount. That is, the keyword generation amount of each scene of the recorded video is calculated, and scenes whose keyword generation amount is less than a predetermined value are deleted. However, even in this case, it is desirable to leave a few minutes of scenes before and after a scene with a keyword generation amount equal to or greater than a predetermined value.

  Thus, by performing video recording and video scene deletion based on the amount of generated keywords, it is possible to greatly reduce the trouble of editing video. As a result, a desired scene can be easily searched.

1 is a schematic block diagram of a video recording / reproducing apparatus according to an embodiment of the present invention. It is a figure which shows the example of the character recognition by pattern matching. It is a figure explaining a screen occupation rate. It is a figure which shows an example of an index display screen. It is a figure which shows an example of an index display screen. It is a figure which shows an example of an index display screen. It is a flowchart which shows the flow of recording of an image | video. It is a flowchart which shows the flow of an index production | generation and recording. It is a flowchart which shows the flow which provides an index to the recorded image | video. It is a flowchart which shows the flow of a display of an index display screen.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 Video recording / reproducing apparatus, 12 Recording part, 14 Recording medium, 16 Keyword generation amount detection part, 18 Voice recognition part, 20 Image recognition part, 22 Level determination part, 24 Index generation part, 26 Keyword storage part, 28 Control part, 29 user interface, 30 playback unit, 32 index screen generation unit, 34 external device, 36 display, 60 index display screen.

Claims (10)

A video recording / reproducing apparatus for recording video on a recording medium,
A storage unit for storing keywords designated by the user in advance;
A detecting means for analyzing the input video and detecting a keyword generation amount per predetermined unit time;
A determination means for determining whether or not a keyword generation amount is equal to or greater than a reference value every predetermined unit time;
Index recording means for recording index information indicating the occurrence of a keyword on a recording medium together with position information where the keyword is generated when the amount of generated keyword is equal to or greater than a reference value;
A video recording / reproducing apparatus comprising: The video recording / reproducing apparatus according to claim 1,
The image recording / reproducing apparatus, wherein the detecting means calculates the generation amount based on the number of occurrences of the keyword within a predetermined time. The video recording / reproducing apparatus according to claim 2,
If a keyword is spoken,
An image recording / reproducing apparatus, wherein the detecting means calculates the amount of occurrence based on the number of occurrences of the keyword and the volume of the keyword at each occurrence. The video recording / reproducing apparatus according to claim 2 or 3,
If the keyword occurs in letters,
The image recording / reproducing apparatus, wherein the detecting means calculates the generation amount based on at least one of the color of the generated keyword, the display time, and the screen occupation ratio, and the number of times the keyword is generated. The video recording / reproducing apparatus according to any one of claims 1 to 4,
The video recording / reproducing apparatus characterized in that the index recording means also records a keyword generation amount as index information. 6. The video recording / reproducing apparatus according to claim 5, further comprising:
An image recording / reproducing apparatus, comprising: a display unit that presents an index display screen showing a keyword occurrence location and an amount generated based on index information recorded on a recording medium to a user. The video recording / reproducing apparatus according to claim 6,
A video recording / reproducing apparatus characterized by being able to indicate a reproduction start position of a video recorded on a recording medium on an index display screen. A video recording / reproducing apparatus for recording video on a recording medium,
A storage unit for storing keywords designated by the user in advance;
A detecting means for analyzing the input video and detecting a keyword generation amount per predetermined unit time;
Index recording means for recording index information indicating a keyword generation position and generation amount on a recording medium in association with the keyword generation position;
Based on the index information recorded on the recording medium, a presentation means for presenting to the user an index display screen showing the occurrence location of the keyword and its generation amount;
A video recording / reproducing apparatus comprising: A video recording / reproducing apparatus for recording video composed of a plurality of scenes on a recording medium,
A storage unit for storing keywords designated by the user in advance;
A detecting means for analyzing the input video and detecting a keyword generation amount per predetermined unit time;
A determination means for determining whether or not a keyword generation amount is equal to or greater than a reference value every predetermined unit time;
Video recording means for recording the video related to the occurrence scene of the keyword on a recording medium when the reference value is greater than or equal to the reference value;
A video recording / reproducing apparatus comprising: A video recording / playback apparatus for playing back video recorded on a recording medium,
A storage unit for storing keywords designated by the user in advance;
A detecting means for analyzing a video recorded on the recording medium and detecting a keyword generation amount per predetermined unit time;
A determination means for determining whether or not a keyword generation amount is equal to or greater than a reference value every predetermined unit time;
Video deletion means for deleting videos whose keyword generation amount is less than the reference value from the recording medium;
A video recording / reproducing apparatus comprising:

Images