CN1320822C - Method and relative system for cross detecting ad fragment using different detection principle - Google Patents
Method and relative system for cross detecting ad fragment using different detection principle Download PDFInfo
- Publication number
- CN1320822C CN1320822C CNB2004100617169A CN200410061716A CN1320822C CN 1320822 C CN1320822 C CN 1320822C CN B2004100617169 A CNB2004100617169 A CN B2004100617169A CN 200410061716 A CN200410061716 A CN 200410061716A CN 1320822 C CN1320822 C CN 1320822C
- Authority
- CN
- China
- Prior art keywords
- frame
- advertisement
- information
- advertising segment
- vision signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Lifetime
Links
- 238000000034 method Methods 0.000 title claims abstract description 14
- 238000001514 detection method Methods 0.000 title abstract description 33
- 239000012634 fragment Substances 0.000 title description 30
- 230000010354 integration Effects 0.000 abstract description 3
- 230000005236 sound signal Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 10
- 230000000052 comparative effect Effects 0.000 description 6
- 238000006243 chemical reaction Methods 0.000 description 5
- 239000000126 substance Substances 0.000 description 4
- 241001269238 Data Species 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 240000004859 Gamochaeta purpurea Species 0.000 description 1
- 238000003780 insertion Methods 0.000 description 1
- 230000037431 insertion Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Landscapes
- Television Systems (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
The present invention provides a method and a relevant system for detecting advertisements from video signals. The invention can automatically integrate the results of advertisement detection under different detection principles so as to crossly evaluate the most possible position of an advertisement segment. The different principles for the advertisement detection can comprises: detecting the positions where a video signal image is not continuous; detecting the playback positions where an image repeatedly occurs in the video signals; detecting the position where an image with specific contents occurs in the video signals; or detecting the positions where sound signal segments are positioned in the video signals. Using the accuracy rate of detection under the different detection principles, the invention can measure the integration according to different weight ratios so as to carry out the cross comparison so that the position the advertisement segment is figured out.
Description
Technical field
The present invention relates to a kind of method and related system that carries out purposes of commercial detection, the particularly a kind of commercial detection method and related system that can effectively integrate different detection principle.
Background technology
The image and sound program service that is provided by wired or wireless broadcasting and TV medium is one of most important information source of advanced information society.Spectators can obtain useful news, knowledge, information or can express the audiovisual entertainment of separating body and mind from the image and sound program service.Yet under the consideration of commerce, the image and sound program regular meeting that the broadcasting and TV medium are provided has advertising segment to intert between normal program.Concerning spectators, these advertising segment regular meetings disturb the continuity of normal program, and spectators' time is also wasted in the puzzlement when causing spectators' rating normal program.When spectators will record the reference that be used as down in the future with normal program (maybe will record when playing after a while), these advertising segments more can expend the resource of user's recording video signal, and cause that the user can't retrieve easily and quickly, management, its vision signal of being recorded of access.And in known technology, existing technology also is difficult to detect the advertising segment in the vision signal.
Summary of the invention
Therefore, main purpose of the present invention promptly is to propose a kind of method and related system of purposes of commercial detection, to detect the advertising segment in the vision signal, and then assist user's filtering or skip these advertising segments, the image and sound program service that allows the user can more effectively use the broadcasting and TV medium to provide.
In general, after the broadcasting and TV medium are inserted in normal program with advertising segment, can also can on voice signal, form the conversion of paragraph forming discontinuous on the picture between normal program and the advertising segment.In addition, be connected the content of normal program in order to assist spectators, the broadcasting and TV medium also can finish a bit of normal program content of back playback at advertising segment.Also have, the broadcasting and TV medium can be when normal program comes to an end and will begin advertising segment, or finish and normal program will continue to broadcast the time at advertising segment, point out the user with linking fragment (similarly being the Corporate Identity sign of broadcasting and TV medium, specific statement or the like) with specific picture.The present invention promptly utilizes above-mentioned these features, detects the possible insert division of advertising segment with different detection principle respectively, integrates the testing result of different detection principle again, to reason out the advertising segment insert division.The present invention can utilize following detection principle: carry out a diversity ratio to detect the discontinuous part of picture in the vision signal; Carry out a ratio of similitude detecting the fragment of playback in the vision signal, or have appearance place of certain content picture in the vision signal; Also can carry out a sound relatively, to detect the paragraph of voice signal in the vision signal.After detecting respectively according to above-mentioned detection principle, resulting testing result can give suitable weight respectively according to its accuracy rate, to integrate every testing result, reasons out the place of advertising segment.
Description of drawings
What Fig. 1 to Fig. 4 illustrated respectively is the different characteristic of advertising segment insert division in the vision signal.
Fig. 5 is the schematic diagram of signal processing circuit one embodiment of the present invention.
Fig. 6 to Fig. 9 is the schematic diagram of relevant data signals when each comparison circuit operates among Fig. 5.
Figure 10 is the schematic diagram of each relevant data signals when the advertisement estimation module is worked among Fig. 5.
Figure 11 is the schematic diagram of another embodiment of signal processing circuit of the present invention.
The reference numeral explanation
10A-10D, V vision signal 10E, Av voice signal
20,50 signal processing systems, 22 difference comparison modules
24,26 similar comparison module 28 sound comparison modules
30,52 advertisement estimation module, 32 frame buffer modules
34 with reference to frame logging modle 36 audio frequency buffer modules
38A-38D weighting block 40A-40D testing result
Ss fragment P, P2 advertisement prompting information
Ad1-Ad4 advertising segment R (1)-R (3) is with reference to frame
Sd1-Sd6 sound paragraph
F(a1)-F(a12)、F(b1)-F(b8)、F(c1-F(c8)、F(i-2)-F(i+2)、F(k-2)-F(k+2)、
F (j-1)-F (j+1), F (t1-1)-F (t6-1), F (t1)-F (t6) frame
Pa1-Pa2, Pb1-Pb2, Pc1-Pc2 normal program
Embodiment
Please refer to Fig. 1 to Fig. 4.In the vision signal that existing broadcasting and TV medium are provided, its advertising segment is interspersed in the situation between normal program, promptly is illustrated in Fig. 1 to Fig. 4.At first, as shown in Figure 1, can be in regular turn among the vision signal 10A in the different time provide frame F (a1), F (a1+1) to F (a2), F (a3) to F (a4) ... F (a11) presents dynamic image to the frame of F (a12) or the like with the tableaux combination that utilizes each frame.Wherein, frame F (a1) is to F (a2) ... be used for presenting the dynamic image of a normal program Pa1 to F (a4) to frame F (a3), frame F (a11) is used for presenting the dynamic image of another normal program Pa2 to the frame of F (a12) or the like, be inserted between normal program Pa1, the Pa2 advertising segment Ad1 then with frame F (a5) to F (a6), F (a7) to F (a8) ... make up the dynamic image that presents advertisement to F (a9) to F (a10).
As be familiar with known to the technology personage, can be with a series of dynamic image (similarly being the dynamic image of so-called same camera lens, Same Scene) by presenting with a series of a plurality of frames with gradual change picture; No matter and be normal program or advertising segment, all be the set different series dynamic image to present its content.Picture in Fig. 1, frame F (a1) to F (a2), frame F (a3) to F (a4), frame F (a5) to F (a6), frame F (a7) to F (a8), frame F (a9) promptly is used for presenting the dynamic image of different series to F (a10), frame F (a11) to F (a12) or the like respectively.For instance, between F (a2), frame F (a1) and the picture of a time frame F (a1+1) are gradual change and similar (just difference is little between the two) at frame F (a1), and frame F (a1+1) and the picture of an inferior frame also are gradual changes and similar, by that analogy.So, frame F (a1) just can be combined into a series of dynamic images of smooth-going variation to F (a2).With respect to the similarity degree between each frame in a series of dynamic images, between the frame of different series dynamic image, just have bigger difference, cause discontinuous on the picture.Picture is in the example of Fig. 1, and frame F (a1) is used for presenting the dynamic image of different series respectively to F (a2), frame F (a3) to F (a4), so between adjacent two frame F (a2), F (a3), will form discontinuous on the picture.In addition, compared to normal program Pa1, Pa2, advertising segment Ad1 also can present its content with the dynamic image of different series, so also be bound to form discontinuous on the picture between advertising segment Ad1 and normal program Pa1, the Pa2.Picture between the F (a11), will form discontinuous on the picture at adjacent frame F (a4) and F (a5), frame F (a10) because of the linking between advertising segment/normal program.In other words, the part that is connected of advertising segment and normal program is bound to occur the discontinuous of picture.
In Fig. 2, vision signal 10B can provide in regular turn frame F (b1), F (b1+1) to F (b8) to present dynamic image; Wherein, frame F (b1) belongs to a normal program Pb1 to F (b3), frame F (b6) then is used for presenting the dynamic image of normal program Pb2 to F (b8) etc., and is inserted in the advertising segment Ad2 between normal program Pb1, the Pb2, promptly is to present its dynamic image with frame F (b4) to F (b5).For convenience spectators are connected the content of normal program, and modern broadcasting and TV medium can be after advertising segment finishes, and the normal program playback before the advertising segment is a bit of.Picture in Fig. 2, the fragment Ss that just will originally in normal program Pb1, broadcast among normal program Pb2 playback one time again; That is to say, frame F (b2) to the picture of F (b3) will be identical with frame F (b6) respectively to the picture of F (b7).Situation before and after advertising segment Ad2, has the identical playback frame of picture and occurs as can be known thus.
As shown in Figure 3, vision signal 10C among Fig. 3 presents its dynamic image with frame F (c1) to F (c8), wherein, the content of normal program Pc1 is presented to F (c3) by frame F (c1), frame F (c6) is used for presenting the dynamic image of another normal program Pc2 to F (c8), be inserted in the advertising segment Ad3 between normal program Pc1, the Pc2, then present its dynamic image to F (c5) with frame F (c4).For beginning or the end that indicates advertising segment, modern broadcasting and TV medium also are everlasting and are connected normal program and advertising segment with the linking fragment with immobilized substance in the vision signal that it provided.For instance, in Fig. 3, the frame F (c2) of normal program Pc1 is one to F (c3) and is connected fragment, the content that it presented, may be the literal that presents " having a rest ... " with picture, with the interruption of prompting spectators normal program, and advertising segment Ad3 promptly will begin.In addition, after advertising segment Ad3 finishes, it similarly is " so-and-so program to be ready beginning ... " or the like literal that frame F (c6) also may present to the linking fragment of F (c7), or the Corporate Identity sign of broadcasting and TV medium itself or the like, with the end of prompting spectators advertising segment.These linking fragments that are connected when advertising segment begins or finish can have convention, thus appearance place of the fragment that these contents are fixed, the place of also just having represented advertising segment to insert.
As know known to the technology personage, except the picture of each frame, the corresponding voice signal of also can arranging in pairs or groups is comprehensively to present dynamically audio-visual effect in the vision signal.In Fig. 4, voice signal 10E is promptly corresponding to vision signal 10D; When vision signal 10D provide successively frame F (d1) to F (d2), F (d3) to F (d4), F (d5) or the like to be when presenting the dynamic image of normal program Pd1, advertising segment Ad4 and normal program Pd2 respectively, voice signal 10E also just can provide related sound signals such as sound frequency, amplitude of vibration, presents the effect of audio-visual multimedia with the collocation dynamic image.As know known to the technology personage, in voice signal 10E, also can include different sound paragraphs.Picture is in Fig. 4, and in normal program Pd1, Pd2, voice signal 10E can provide sound paragraph Sd1 to represent corresponding voice signal with Sd2, sound paragraph Sd5 with Sd6 respectively.Different sound paragraphs can be used for representing the melody of different series respectively, or different performers' voice dialogue.In like manner, in the advertising segment Ad4 that forms to F (d4) by frame F (d3), also have sound paragraph Sd3, Sd4 or the like, present the voice of advertising segment and dub in background music.When the broadcasting and TV medium are inserted in advertising segment between the normal program, can just insert advertising segment in the content of normal program, when the story of a play or opera comes to an end; Along with normal program comes to an end, voice signal also should come to an end.In other words, being connected part in normal program with advertising segment, also can be the linking part of alternative sounds paragraph in the voice signal.
By the discussion of Fig. 1 to Fig. 4 as can be known, the advertising segment insert division can have several features, comprise: the insert division that advertising segment begins and finishes can form fragment, the advertising segment that playback may appear before and after inserting in discontinuous, advertising segment on the picture may have the linking fragment with immobilized substance before and after inserting, and advertising segment begins/finishes the paragraph that part can form voice signal.In other words, if can in a vision signal, detect automatically on the picture discontinuous, detect place that whether playback fragment and playback fragment are arranged, detect the paragraph place that whether has some frame to meet the feature that is connected fragment and detect voice signal, also just can detect the place of advertising segment automatically, and then assist the user to skip or these advertising segments of montage according to every testing result.
Certainly, only according to the testing result of individual event feature, be the difficult insert division of confirming advertising segment.For instance, as shown in Figure 1, no matter among normal program or advertising segment, all may be because of the different series dynamic image discontinuous on the picture alternately take place; That is to say that the discontinuous part of picture is not must be the advertising segment insert division.If only detect the discontinuous part of picture in the vision signal, be difficult to then confirm whether it is the advertisement insert division really.But, just can intersect and compare out the correct insert division of advertising segment (or probability is the highest, most possibly be the place of advertising segment insert division) if can integrate the testing result of different characteristic.For instance, if discontinuous on the picture arranged between a certain given frame and its last frame, and this given frame meets the characteristic feature that is connected frame in the fragment, and this just represents between this given frame and its last frame very likely is exactly the insert division (this advertising segment of for example saying so finishes part) of advertising segment.And the present invention is exactly a testing result of specifically integrating various features in quantitative mode, so that can more correctly detect the insert division of advertising segment.
Please refer to Fig. 5.Fig. 5 is the function block schematic diagram of signal processing system one embodiment 20 of the present invention.Signal processing system 20 can build and place recording apparatus (similarly being the video tape recorder of different medium such as video tape, CD, hard disk) or can record/multimedia computer of broadcast video signal, so that from the vision signal of these device recordings, detect advertising segment.Can be provided with a frame buffer module 32, an audio frequency buffer module 36, a difference comparison module 22, similar comparison module 24 and 26, in the signal processing module 20 with reference to frame logging modle 34, a sound comparison module 28 and an advertisement estimation module 30.When signal processing system 20 will be carried out purposes of commercial detection to an audio-visual vision signal V, frame buffer module 32 can be by the information that obtains each frame picture among the vision signal V, and each frame is offered difference comparison module 22 and similar comparison module 24,26.36 energy of audio frequency buffer module receive its voice signal Av from vision signal V, to offer sound comparison module 28.
In signal processing system 20, difference comparison module 22, similar comparison module 24 and 26, and sound comparison module 28 are exactly to detect according to the advertising segment feature among Fig. 1 to Fig. 4 respectively, and produce corresponding detection result 40A to 40D.For further specifying the situation of above-mentioned each module running, please refer to Fig. 6 to Fig. 9 (and in the lump with reference to figure 5).Fig. 6 to Fig. 9 promptly is used for illustrating the operation situation of difference comparison module 22, similar comparison module 24,26 and sound comparison module 28 respectively.
At first, as shown in Figure 6, at a succession of frame F (i-2), F (i-1), F (i), F (i+1) and F (i+2) of vision signal V or the like, difference comparison module 22 can more adjacent in regular turn two frame F (i-2) and F (i-1), frame F (i-1) and F (i), frame F (i) and frame F (i+1), frame F (i+1) and F (i+2) between difference, and the comparative result of correspondence is recorded among the testing result 40A.In preferred embodiment of the present invention, the present invention can calculate the characteristic of each frame quantitatively according to default characteristics algorithm.For instance, the characteristic of one frame can be the summation of all pixel pixel datas in this frame (similarly being data such as brightness, color), or the distribution scenario of all pixel datas (similarly is the distribution map of brightness or color in this frame, histogram), or even according to the result after this frame frequency domain conversion (similarly being the 2-D discrete cosine conversion) produce this frame characteristic of correspondence data.And difference comparison module 22 with regard to can according to adjacent two frames separately the characteristic of correspondence data carry out quantitative comparison; If the difference that a certain frame is adjacent characteristic between frame has surpassed one and faced the limit difference degree, just can judge that this frame is adjacent discontinuous on the picture arranged between the frame.Picture the present invention indicates the difference comparative result between two adjacent frames at testing result 40A with simple " 0 ", " 1 " exactly in the embodiment of Fig. 6.
For instance, if difference is little between frame F (i-2) and the frame F (i-1) (do not surpass and face the limit difference degree), represent this two frame to should be with two gradual change frames in a series of dynamic images, do not have picture discontinuous between the two, can in testing result 40A, represent this situation with sign " 0 " accordingly.In like manner, in testing result 40A, in the sign " 0 " between frame F (i) and the F (i+1), between frame F (i+1) and the F (i+2), also just represent between frame F (i) and the F (i+1) respectively, do not have the discontinuous of picture between frame F (i+1) and the F (i+2).Relatively, suppose between adjacent frame F (i-1) and F (i), the difference of both characteristics has been higher than faces the limit difference degree, just represents to have between this two frame discontinuous on the picture, can represent the discontinuous generation of picture with sign " 1 " accordingly in testing result.In other words, in this kind embodiment, the sign among the testing result 40A " 1 " just can be considered a difference information, is used for the discontinuous nidus of hint image.
As shown in Figure 7,24 energy of the similar comparison module among Fig. 2 are searched similar frame in each frame of vision signal V, and accordingly with the outcome record of searching in testing result 40B.For instance, when similar comparison module 24 finds that frame F (i) is similar to F (j), just can in testing result 40B, do record by the sign " 1 " with a correspondence.Relatively, when finding not have frame similar in the vision signals, just can in testing result 40B, write down the sign " 0 " of a correspondence to frame F (i-1) as if similar comparison module 24.As once mentioning in Fig. 2 and the related description, begin playback fragment preceding, after finishing at advertising segment and can be used as one of feature of advertising segment insert division, so if find that frame F (i) is similar in appearance to frame F (j), frame F (i) probably is exactly the playback of frame F (j), and advertising segment just is inserted between frame F (j) and the F (i), so in testing result 40B, the sign of available " 1 " is represented a similar information, the possible insert division of prompting advertising segment.In other words, similar comparison module 24 can be considered as the frame before a certain given frame with reference to frame, with relatively each with reference to the similarity degree between frame and this given frame, thereby carry out the detection of playback fragment.
Be similar to the implementation of difference comparison module 22, when realizing similar comparison module 24, equally also can carry out the quantitative comparison of similarity degree with each frame characteristic of correspondence data; If the difference degree between the characteristic of two frames faces limit difference less than one, in the equivalence, just represent the similarity degree between these two frames to face the limit similarity degree greater than one quantitatively, and can be considered frame like the two-phase.In addition, when searching the frame similar in appearance to a certain frame in the function that realizes similar comparison module 24, also can set the scope of search according to the characteristic that advertising segment inserts.For instance, when searching when whether having frame similar, can come carry out the comparison (just frame F (i-M-N) being used as to F (i-N) is with reference to frame) of similarity degree to the frame between F (i-N) at frame F (i-M-N) with frame F (i) at a frame F (i) to it.And wherein the setting of parameter M, N just can be decided by the actual characteristic that advertising segment inserts.For example, can not be shorter than 30 seconds, the longlyest can not surpass 5 minutes, just can decide the size of N and M respectively according to these characteristics (and frame rate, the frame number of unit interval correspondence) if the length of advertising segment is the shortest.
As for shown in Fig. 8, then be the schematic diagram of another similar comparison module 26 (Fig. 2) running.As discussing in Fig. 3 and the relevant narration, may meaningful fixing linking fragment before or after advertising segment, can be used as is one of feature of advertising segment; And when vision signal V was carried out purposes of commercial detection, similar comparison module 26 was exactly to be used for detecting the linking fragment that whether has these contents fixing among the vision signal V, and produced a corresponding detection result 40C.Because these image contents that are connected fragments are fixed, with reference to frame logging modle 34 can note in advance these image contents that are connected fragments as default with reference to frame (similarly be among Fig. 8 reference frame R (1) to R (3) or the like); When vision signal V is carried out purposes of commercial detection, similar comparison module 26 just can compare mutually with each frame among the vision signal V and with reference to the reference frame that writes down in the frame logging modle 34.With Fig. 8 is example, if similar comparison module 26 compares frame F (k) a certain reference frame R (2) of record in advance in the reference frame logging modle 34, similar comparison module 26 just can be used as a similar information with sign " 1 " accordingly in its testing result 40C, representing this frame F (k) to locate probably is exactly the insert division of advertising segment.Relatively, if the frame F (k-1) among the vision signal V does not meet with reference in the frame logging modle 34 each with reference to frame, just can be accordingly react its comparative result with the sign of " 0 ".
Be similar to similar comparison module 24, similar comparison module 26 also can be to carry out the comparison of similarity degree according to each frame characteristic of correspondence data.In such cases, with reference to frame logging modle 34 records be exactly respectively with reference to frame characteristic of correspondence data, similar comparison module 26 then is that the characteristic of each frame among the vision signal V and each characteristic with reference to frame are compared.In preferred embodiment of the present invention, one frame characteristic of correspondence data should be less than all phase prime numbers in this frame according to the data of summation, make with reference to frame logging modle 34 and can write down more, and each comparison module 22,24 and 26 also can compare more efficiently with reference to frame.In addition, can specify and upgrade by the user with reference to the reference frame of record in the frame logging modle 34.For instance, the signal processing system among Fig. 2 20 can be implemented in the recording apparatus; When the user when watching the vision signal of this recording apparatus record, if seeing, the user has immobilized substance, fixed mode when being connected fragment between advertising segment and the normal program, the user just can be when this be connected the fragment broadcast, controlling the frame that this recording apparatus will be connected fragment captures, and the frame that captures transferred to reference to frame logging modle 34, with as the reference frame.So, in follow-up running, similar comparison module 26 just can carry out purposes of commercial detection with reference to frame according to this.And in preferred embodiment of the present invention, can be a non-volatile storage device with reference to frame logging modle 34, various to write down constantly with reference to frame (or its corresponding characteristic of sending out).
As for what anticipate shown in Fig. 9, then be the schematic diagram of sound comparison module 28 operation situations.Such as in Fig. 4 and the relevant narration discussion, the insertion meeting of advertising segment causes the paragraph of voice signal in the voice signal of vision signal correspondence.When vision signal V was carried out purposes of commercial detection, sound comparison module 28 can carry out the detection of paragraph to the voice signal Av of vision signal V correspondence, finding out the paragraph place among the voice signal Av, and produced a corresponding detection result 40D.For instance, as shown in Figure 9, suppose that sound comparison module 28 detects voice signal Av and between frame F (k-1) and F (k) paragraph arranged, just can in testing result 40D, be used as a sound paragraph information with sign " 1 " accordingly; Relatively, if between frame F (k), F (k+1), corresponding voice signal Av does not have the appearance of paragraph, just can represent with a sign " 0 " in testing result 40D.
After producing each testing result 40A to 40D, the advertisement estimation module 30 of the present invention in Fig. 5 be these testing results of energy cross reference just, to infer the place that advertising segment.For instance, if difference comparison module 22 detects discontinuous on the picture arranged between certain two adjacent frame in vision signal V, this picture is discontinuous may to be the insert division of advertising segment, may be discontinuous that scene transitions is caused in the normal program (or advertising segment), be not the insert division of advertising segment yet.But, if detecting the frame at this discontinuous place of picture, similar comparison module 24 also meets the reference frame (similarly being the frame F (k) among Fig. 8) that is connected fragment in addition, then this discontinuous place is the advertisement insert division probably just.In other words, each testing result 40A to 40D is compared in cross reference, integration, just can improve the accuracy rate of purposes of commercial detection.And the commercial detection module 30 of the present invention in Fig. 5 is exactly to be used for integrating each testing result, to reach the purpose of purposes of commercial detection.
In Fig. 5, also drawn the schematic diagram of commercial detection module 30 1 embodiment; Can be with four weighting block 38A to 38D in the advertisement estimation module 30, respectively each sign among the testing result 40A to 40D is weighted, and the results added after the weighting drawn an advertisement prompting information P with integration, be used for reflecting that each frame is the probability of advertisement insert division.Please refer to Figure 10 (and in the lump with reference to figure 5).Figure 10 is the schematic diagram of each related data when advertisement estimation module 30 operates among Fig. 5.Weighting block 38A to 38D can multiply by the sign among the testing result 40A to 40D weighted value w1 to w4 (each weighted value can be on the occasion of number) respectively, and obtains the advertisement prompting information of each frame correspondence after addition.For instance, as shown in Figure 10, suppose in vision signal V, difference comparison module 22 frame F (t1-1) to F (t1), F (t2-1) to F (t2), F (t3-1) to F (t3), F (t4-1) to F (t4), F (t5-1) to F (t5), F (t6-1) to detecting discontinuous on the picture between the F (t6), so in testing result 40A, can represent with the sign " 1 " of correspondence.Similar comparison module 24 then detects frame F (t6) similar in appearance to frame F (t2-1), and also the sign " 1 " with correspondence is represented in testing result 40B.Another similar comparison module 26 detect frame F (t5) afterwards to the frame between frame F (t6-1) similar in appearance to being connected the frame that has immobilized substance in the fragment, so the sign " 1 " with correspondence is represented in testing result 40C.28 of sound comparison modules frame F (t2-1) to F (t2), frame F (t5-1) to F (t5), frame F (t6-1) is to the paragraph that detects voice signal between the F (t6), also " 1 " sign with correspondence is represented in testing result 40D.And advertisement estimation module 30 with the sign weighting summation in each testing result after, the advertisement prompting information P of its gained is just as shown in Figure 10.
As shown in figure 10, the value of the advertisement prompting information P (t1) that frame F (t1) is corresponding is w1, frame F (t1) is though be the discontinuous part of picture in expression, but do not meet among the vision signal V yet and be connected the picture that has certain content in the fragment with the picture of its similar frame, frame F (t1), do not form the paragraph place of voice signal yet, so the discontinuous very likely journey that scene conversion is made just of the picture of frame F (t1) is not the advertising segment insert division.In like manner, distinguish corresponding advertisement prompting information P (t3), P (t4) as frame F (t3), F (t4), its value also all only is w1, and representative says that these frames only meet the discontinuous feature of picture, does not meet the further feature that advertising segment inserts.
Relatively, locate at frame F (t2),, also meet the feature (testing result 40B) of playback fragment except discontinuous (the testing result 40A) on the picture arranged, the feature (testing result 40D) that also has the sound paragraph is so its corresponding advertisement prompting information P (t2) just becomes w1+w2+w4; And frame F (t2) also probably is exactly the insert division of advertising segment.Its corresponding advertisement prompting information P (t6) in like manner, locates, also met the various features of Fig. 1 to Fig. 4, so will be adding up of weighted value w1 to w4 at frame F (t6).In other words, each weighted value be all on the occasion of situation under, if a certain frame place meets the feature of multinomial more advertising segment, its corresponding advertisement prompting information is also just big more.In the equivalence, the numerical values recited of the advertisement prompting information of a frame correspondence has just reflected that also this frame is the probability of advertising segment insert division.The advertisement prompting information of one frame correspondence is big more, and this place might be the insert division of advertising segment more just.And advertisement estimation module 30 of the present invention just can be judged the place of advertising segment according to the advertisement prompting information of each frame correspondence.
In the present invention, can set the size of each weighted value accordingly according to the pointer meaning of advertising segment various features.For instance, the discontinuous place of picture may be the advertisement insert division, but also might be the scene conversion place, so the testing result 40A at the discontinuous place of picture can not reflect the insert division of advertising segment clearly, its index meaning is less, accuracy rate as purposes of commercial detection is lower, so the value of weighted value w1 can be made as a less numerical value.Relatively, testing result 40B, 40C may be able to reflect the insert division of advertising segment comparatively clearly, and its index meaning is just bigger, and corresponding weighted value w2, w3 just can be made as bigger numerical value.By statistical analysis, just can set out the size of each weighted value quantitatively to actual video signal.
Certainly, advertisement estimation module 30 of the present invention also can be integrated different testing results with other algorithm, not necessarily will send out algorithm with weighted sum among Fig. 5.For instance, the present invention can find out the discontinuous place of picture among the vision signal V according to testing result 40A earlier, carry out the comparison of similarity degree with similar comparison module 26 at the frame at discontinuous place again, whether have the discontinuous frame of picture also to meet the feature that is connected fragment in addition to detect.In the equivalence, so also can integrate the testing result of different characteristic.
In the embodiment of Fig. 5, the present invention detects four kinds of features of advertising segment insert division respectively with four comparison modules, and integrates four testing result 40A to 40D that it produces, and compares out the insert division of advertising segment with intersection.But technology of the present invention also can further be simplified, only according to the wherein several purposes of commercial detection of carrying out in these four kinds of features; As long as these several features can be integrated out enough index meanings, still can suitably reach the purpose of purposes of commercial detection.Please refer to Figure 11; Embodiment among continuity Fig. 5, illustrated in Figure 11 promptly is the function block schematic diagram of another embodiment 50 of signal processing circuit of the present invention.Signal processing circuit 20 in Fig. 5, signal processing circuit 50 has been omitted sound comparison module and relevant audio frequency buffer module, only used difference comparison module 22 and similar comparison module 24,26 to detect three kinds of features of advertising segment, to produce three comparative result 40A to 40C respectively.And Figure 10 once in advertisement estimation module 52 also can integrate the detection that these three comparative results carry out advertisement, similarly be respectively the sign in each comparative result to be weighted with weighting block 38A to 38C, the value with summation produces advertisement prompting information P2 again.
In the embodiment of Fig. 5 and Figure 11, signal processing system of the present invention can realize with the form of hardware, software or firmware.For instance, if signal processing system is built and is placed recording apparatus, then the mode of available firmware realizes the comparison module of each detected characteristics, similarly is to carry out different firmware program code with single treatment circuit, realizes out the function of different comparison modules and advertisement estimation module respectively.Perhaps, when signal processing system of the present invention is a framework in multimedia computer the time, just can utilize central processing unit to carry out different software program codes, to realize out the function of each comparison module, advertisement estimation module respectively.
In known technology, effective method does not positively detect the advertising segment in the vision signal, causes the user can't use programme information useful in the vision signal efficiently.In comparison, the present invention then is the different characteristic that detects the advertising segment insert division respectively, and the testing result that can integrate various features is automatically intersected the insert division of comparing out advertising segment.After the relevant information of insert divisions such as obtaining advertising segment and begin/finish, just can assist that the user skips, montage or filtering advertising segment, allow the user can use programme information useful in the vision signal more convenient, more efficiently.
The above only is preferred embodiment of the present invention, and all equalizations of doing according to the present patent application scope of specially reflecting change and modify, and all should belong to the covering scope of patent of the present invention.
Claims (10)
1. method that can detect advertising segment in a vision signal comprises:
Obtain this vision signal, this vision signal can provide a plurality of different frames according to a preset order, to present dynamic image;
Carry out a difference comparison step,, wherein, then provide the difference information of a correspondence at this frame if a frame is adjacent difference degree between frame to be surpassed one and face the limit difference degree with the difference degree between each frame in this vision signal relatively;
Carry out a similar comparison step, with each frame and a similarity degree with reference to frame in this vision signal relatively, and face the limit similarity degree with reference to the similarity degree between frame above one as if a frame and this, the similar information of one correspondence then is provided at this frame, should be frame before this frame wherein with reference to frame, and a frame with default image content; And
Carry out an advertisement estimating step, to judge having which frame to belong to this advertising segment in this vision signal according to this difference information and this similar information.
2. the method for claim 1 more comprises:
When this vision signal be according to this preset order when different time provides a voice signal of unlike signal, carry out the voice signal of a sound comparison step with different time in this vision signal relatively, and find out the paragraph of voice signal, so that the sound paragraph information of a correspondence to be provided; And
When carrying out this advertisement estimating step, be to judge have which frame to belong to this advertising segment in this vision signal according to this difference information, this similar information and this sound paragraph information.
3. the method for claim 1, wherein, when judging the frame of this advertising segment according to this similar information, be to judge that the frame of this advertising segment is between this is with reference to the corresponding frame of the similar information of frame and this carrying out this advertisement estimating step.
4. when the method for claim 1, wherein judging the frame of this advertising segment according to this similar information, be that the frame of judging this advertising segment is positioned at before the corresponding frame of this similar information carrying out this advertisement estimating step.
5. when the method for claim 1, wherein judging the frame of this advertising segment according to these two difference informations, be to judge that the frame of this advertising segment is between the corresponding frame of this two difference information carrying out this advertisement estimating step.
6. the method for claim 1, wherein, when judging the frame of this advertising segment according to this similar information and this difference information carrying out this advertisement estimating step, be judge this advertising segment frame between this is with reference to the corresponding frame of the similar information of frame and this, and between two difference informations correspondence frame.
7. the method for claim 1, wherein, when carrying out this advertisement estimating step, be the advertisement prompting information that a correspondence is provided at each similar information and each difference information, so that each advertisement prompting information has the advertisement probable value of a correspondence, wherein, then make the advertisement prompting information of this difference information correspondence have bigger advertisement probable value if it is same frame that the frame frame corresponding with a difference information of a similar information correspondence arranged.
8. method as claimed in claim 7, wherein, carry out this advertisement estimating step and more comprise:
The advertisement probable value that compares each advertisement prompting information correspondence; And
Find out the big advertisement prompting information of advertisement probable value, and the frame of judgement advertising segment is between the frame of these advertisement prompting information correspondences.
9. signal processing system that detects advertising segment in a vision signal comprises:
One frame buffer module is used for keeping in this vision signal, and this vision signal provides a plurality of different frames according to a preset order, to present dynamic image;
One difference comparison module, be used for the difference degree between each frame in this vision signal relatively, wherein if the difference degree that a frame is adjacent between frame faces the limit difference degree above one, then this difference comparison module provides the difference information of a correspondence at this frame;
One similar comparison module, be used for relatively each frame and a similarity degree with reference to frame in this vision signal, wherein face limit similarity degree with reference to the similarity degree between frame above one as if a frame and this, then this similar comparison module provides the similar information of a correspondence at this frame; And
One advertisement estimation module is used for judging have which frame to belong to this advertising segment in this vision signal according to this difference information and this similar information.
10. signal processing system as claimed in claim 9, wherein, if this vision signal is the voice signal that unlike signal is provided in different time according to this preset order, this signal processing system more comprises:
One audio frequency buffer module is used for keeping in the voice signal in this vision signal; And
One sound comparison module, its relatively in this vision signal the voice signal of different time and provide the sound paragraph information of a correspondence to judge have which frame to belong to this advertising segment in this vision signal finding out the paragraph of voice signal.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2004100617169A CN1320822C (en) | 2004-06-30 | 2004-06-30 | Method and relative system for cross detecting ad fragment using different detection principle |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2004100617169A CN1320822C (en) | 2004-06-30 | 2004-06-30 | Method and relative system for cross detecting ad fragment using different detection principle |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1589004A CN1589004A (en) | 2005-03-02 |
CN1320822C true CN1320822C (en) | 2007-06-06 |
Family
ID=34603645
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB2004100617169A Expired - Lifetime CN1320822C (en) | 2004-06-30 | 2004-06-30 | Method and relative system for cross detecting ad fragment using different detection principle |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1320822C (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100580693C (en) * | 2008-01-30 | 2010-01-13 | 中国科学院计算技术研究所 | Advertisement detecting and recognizing method and system |
CN102890950B (en) * | 2011-07-18 | 2016-08-03 | 大猩猩科技股份有限公司 | Media automatic editing device, method, media transmissions method and its broadcasting system |
CN102523482B (en) * | 2011-12-07 | 2014-07-23 | 中山大学 | Advertisement monitoring technology based on video content and regression method |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1130004A (en) * | 1993-08-31 | 1996-08-28 | 联合企业股份有限公司和无线电产业保护两合公司 | Process and device for detecting undesirable video scenes |
WO2001035409A2 (en) * | 1999-11-10 | 2001-05-17 | Thomson Licensing S.A. | Commercial skip and chapter delineation feature on recordable media |
-
2004
- 2004-06-30 CN CNB2004100617169A patent/CN1320822C/en not_active Expired - Lifetime
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1130004A (en) * | 1993-08-31 | 1996-08-28 | 联合企业股份有限公司和无线电产业保护两合公司 | Process and device for detecting undesirable video scenes |
WO2001035409A2 (en) * | 1999-11-10 | 2001-05-17 | Thomson Licensing S.A. | Commercial skip and chapter delineation feature on recordable media |
Also Published As
Publication number | Publication date |
---|---|
CN1589004A (en) | 2005-03-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11477156B2 (en) | Watermarking and signal recognition for managing and sharing captured content, metadata discovery and related arrangements | |
US8855796B2 (en) | Method and device for detecting music segment, and method and device for recording data | |
TWI242376B (en) | Method and related system for detecting advertising by integrating results based on different detecting rules | |
US8068719B2 (en) | Systems and methods for detecting exciting scenes in sports video | |
CN1256588A (en) | Information processing system and method and distributing medium | |
EP2017827B1 (en) | Music section detecting method and its device, data recording method, and its device | |
CN1836287A (en) | Video abstracting | |
CN1973536A (en) | Video-audio synchronization | |
CN1922863A (en) | Video trailer | |
CN1703083A (en) | Moving image processing apparatus and method | |
US7646818B2 (en) | Method and related system for high efficiency advertising detection | |
CN1722280A (en) | CD, compact disk recording method and optical disk recording device | |
CN1777265A (en) | Image-sound synchronous recording and playing method | |
CN1320822C (en) | Method and relative system for cross detecting ad fragment using different detection principle | |
US8234278B2 (en) | Information processing device, information processing method, and program therefor | |
CN118018676B (en) | Playback interaction method, device and system for twin video conference | |
CN1180403C (en) | CD reproducing system and method of reproducing static pictures | |
CN1905045A (en) | Information playback method using information recording medium | |
CN116033096B (en) | Picture content dubbing method and device and terminal equipment | |
CN1992863A (en) | Apparatus for automatic separating chapter of CD video recorder and method thereof | |
CN118044206A (en) | Event source content and remote content synchronization | |
JP2005223794A (en) | Apparatus and method of recording audio-visual content | |
CN1564261A (en) | MP3 playback device having multi-file synchronous playing function and its method | |
CN1870156A (en) | Disk play device and its play controlling method and data analysing method | |
CN1641626A (en) | Music television making and broadcasting system and method thereof |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CX01 | Expiry of patent term | ||
CX01 | Expiry of patent term |
Granted publication date: 20070606 |