DE60105975T2 - METHOD FOR DETECTING AND IDENTIFYING TIMELY CONSTANT PATTERNS IN VIDEO SIGNALS, IN PARTICULAR FOR DETECTING ADVERTISING POTS IN TV TELEVISIONS, AND DEVICE FOR CARRYING OUT THIS METHOD - Google Patents

Description

The The invention relates to a method for finding and recognizing time-stable patterns in video signals from any video source, in particular for automatic recognition of advertisements in television broadcasts or the like transmissions by video signal analysis according to the preamble of claim 1, a specific use of such a method according to claim 9 and an apparatus for performing a Such method according to claim 10.

private and now also increasing in public legal television broadcasters use almost every possibility for interruption of the program by advertising inserts money to earn. Annoying are there for many viewers not only the duration and frequency of commercials, but also the times when the current program is interrupted becomes.

The here presented invention is capable of advertising contributions in television broadcasts independent to recognize and almost every device dependent on from the advertising recognition over to control any bus system or by infrared command sequences. This can be used, among other advertising-free video recordings, by a video recorder at the beginning and end of a recognized advertising sequence is switched by appropriate control commands.

It Already methods and devices are known, the recognition of advertising contributions in TV (TV) broadcasts and on the basis of which a system for the suppression of advertising sequences should be realized in a video recording. Part of this Method is based on that within a central monitoring device TV channels monitored by humans and control signals are sent to a decoder box at the customer should. The other part of the method is based on circuitry, either integrated directly into the video recorder or designed as an accessory are and are based on a fully automated detection of TV advertising allow a certain feature present in the television signal should.

following will be first a short overview of the previous state of development of advertising recognition procedures and on it given based video switching systems:

From the DE 195 15 604 A1 a method for controlling a video recorder in response to a contained in television signals identification of the transmission format is known. According to the solution proposed there, it is assumed that commercials are not created in the so-called PALplus format. If an ordinary PAL signal is detected in the context of a PALplus program, this is defined as unwanted advertising and a connected video recorder is paused. The problem with this method is the fact that so far as no broadcasts are broadcast in PALplus format, so this process can currently be described as irrelevant.

Likewise, methods such as in the DE 295 11 499 U1 described, which are to distinguish using the black boundary bars feature films in the widescreen format of advertising, not generally satisfactory solution. Only a small percentage of broadcast television contributions are currently feature films in widescreen format (about 8% in Germany).

Also in the DE 42 04 689 A1 a circuit for suppressing advertising images during the video recording of a television program will be described. In order to distinguish the transmission of advertisements to be recorded, the presence of a so-called transmitter emblem is checked, which normally disappears completely or partially during an advertising interruption. For the detection of this transmitter embryo, the brightness values of a specific section of a picture line are integrated over a certain period of time (about 5 seconds) and then evaluated. If the result exceeds a certain threshold value, the presence of a transmitter emitter in this image area is concluded. Depending on this feature recognition, a command sequencer in the form of a learnable remote control for controlling a video recorder should then be activated. Obviously, this assumes that a sender emblem is characterized by particularly high brightness values compared to its image context. Colored or even dark sender emblems are thus ruled out from the outset for this type of advertising recognition. However, the main criticism of this method lies in the fact that most sender emblems are transparent and thus have fluctuating brightness values. Reliable advertising recognition based on brightness thresholds is therefore not possible. Furthermore, as will be explained in more detail, the problem of video recorder control remains unresolved, even if the on Apparently, with the help of a learnable remote control, an easy-to-implement solution could be found.

Also in the DE 44 17 481 A1 For the detection of advertisements in television broadcasts, reference is made to the detection of sender logos. Concretely, the video signal is first converted analog-digitally, whereby the digital signal is subjected to a special feature analysis. The feature recognition devices check, among other things, screen lines as to whether they consist wholly or at least partially of pixels that do not change over a longer period of time in their luminance and / or color information. If such points exist, then according to this method either the presence of a transmitter logo or the existence of a black bar, that is to say a feature film in the widescreen format, can be concluded. In both cases the found pixels are stored for subsequent checks. When changing the luminance and / or color information of the previously found pixels is concluded on an advertising sequence, and a connected control logic is to put a video recorder to pause. Just like the DE 42 04 689 A1 The same points of criticism are also to be criticized in this procedure. Transmitter logos are transparent in many cases and can therefore at least not reliably be detected by this method. Also, the control logic for the external control of a video recorder is not explained in detail, so that the method for the practical use is unlikely to be suitable.

In the DE 43 03 942 A1 . DE 195 28 344 C2 . DE 197 54 252 A1 and DE 198 07 765 A1 is referred to the idea that monitored in a central monitoring device all the television channels of people around the clock and that in the case of a detected advertising sequence certain commands are to be sent in an independent of the television signal transmission. A special decoder box at the customer should finally be able to receive these signals and control the video recorder accordingly. The central problem is the control of the video recorder through the decoder box.

In all solutions that as an accessory are provided for the external control of a video recorder remains the question unanswered, such as this for the purpose of advertising-free video recordings can be switched. An external control of a video recorder just seems over Infrared remote control commands possible to be. In some procedures, for this purpose is a learning Remote control provided with the commands for recording start and recording stop can be programmed. This solution seems at first Look practical, but many problems are ignored been left.

For example many video recorders switch on these days after a few minutes Recording interruption from pause to stop mode, and many devices after another short period of time finally in the standby mode. It is usual, that a recording start, starting from all three different operating modes, is to accomplish with different command sequences. Consequently a stored command sequence for the recording start is not always to the desired Result.

One Another problem is timer or timer programmed video recordings In order to prevent an accidental recording interruption, Most VCRs will not work in this recording mode controlled by infrared commands. Furthermore, in the case of a timer-programmed Video recording the auxiliary device be able to automatically detect the start time of a video recording then also with the advertising recognition to start. This problem also finds little mention such as the shutdown of the video recorder and the advertising recognition system at the end of the recording.

Alles the problems mentioned in the VCR control and the many shortcomings the systems in the advertising recognition show that so far no satisfying solution has been found to be a reliably functioning advertising recognition and video switching or control system can guarantee.

Furthermore, a diploma thesis is known from the Münster University of Applied Sciences, Department of Electrical Engineering, entitled "Optimization and Implementation of a System for Suppression of Advertising Blocks in Recordings with the Video Recorder" in the fall of 1999. This diploma thesis deals with the possibilities to record TV programs without advertising with the video recorder For the recognition of advertising blocks, various algorithms based on optical detection of the logo are introduced, implemented in a software solution, tested and evaluated.The K-Develop development environment under Linux was the basis for an open-soure C ++ program that uses feature films the logo recognition is to be ad-free, but in its recognition security still has significant deficiencies and that in terms of recorder control as well as the other presented methods and devices no effective represents a practicable solution.

It It is therefore an object of the invention to specify both a method which allows advertising contributions in TV broadcasts or the like transmissions reliably recognize, in dependence this advertising recognition any device, in particular a video recorder, to indicate a specific use of the method as well as an apparatus for carrying out the method.

The Invention should be characterized by the following goals over others Mark procedure:

The Procedure is to television advertising on the basis of a, present in the video signal Certainly recognize features and not on special cases, such as For example, the recognition of a particular feature film format, restrict.

The In terms of its reliability, the method should be significantly different from the previous ones distinguish well-known fully automatic advertising recognition. Even with disturbed Sender reception should no error detections occur.

The Arrangement should be able to almost any device, even a VCR in timer-programmed recording mode, through Control commands via to switch a bus system or via infrared.

The Arrangement should be able to start a timer programmed Independent video recording to determine then from a power-saving mode of operation to self-activate and start with the advertising recognition.

The Arrangement should be able to the end of a video recording independent to detect both the VCR and itself stop and into a low-power standby mode, below as well Sleep mode called, move.

The Device should with as possible be low hardware effort feasible.

The The aforementioned object is achieved according to the method by a Method with the features of claim 1, use side by use with the features of claim 9 and On the device side by a device with the features of Patent claim 10 solved. In the dependent claims are expedient embodiments and further developments of the invention.

Before the invention is described in detail, first takes a brief overview:

The Invention operates on the basis of a detection of transmitter-specific Logos. The inventive method is based on the observation that TV broadcasters their respective transmission by the insertion of a transmitter-typical logo in one of the four Identify the screen corners and that this marking during a Advertising interruption is missing in whole or in part. Assuming that Shape, size and position a transmitter logo time invariant, i. constant, are, be within a so-called reference measurement time the data of a digitized Video signals continuously read and pattern recognition algorithms analyzed. From the multitude of images read in one after the other can Using statistical methods, the edge profile directions of the Logos filtered and for Be stored for comparison purposes. In the subsequent test measurement is by similarity comparison following pictures with the saved reference template the others Presence of the sender logo checked for commercial breaks using this identifier. As according to the invention neither Color still luminance but edge trajectories analyzed can, can with the help of this procedure also transparent transmitter logos for the advertising recognition be used.

Furthermore, a working on the basis of this method circuitry, hereinafter referred to as device described for advertising-dependent device control. In particular, a device for controlling a video recording device is considered here in particular. This device has the purpose to recognize even during the recording of a television program advertising sequences and optionally interrupt the video recording. On the one hand, the device can be designed as an assembly integrated in the video recorder, and on the other hand as a supplementary device. The cradle is simply connected to the video output of the recorder and then automatically takes over the recording control at the beginning of the recording. For this purpose, the device has both an infrared transmitter, as well as an infrared receiver unit, with the help of any remote control commands store and at the appropriate time issue point depending on the advertising recognition again. Alternatively, the control of a video recorder via any bus system is possible. Another inventive approach results from the possibility of data exchange via SCART bus. According to the widespread P50 standard, for example, timer data of a video recorder can be read out and also changed. By manipulating the recording start or stop time, video recordings can thus be advantageously controlled.

According to the invention that Procedure for advertising recognition as a basis for a program used for post-processing digitally recorded TV broadcasts be on a rewritable DVD or similar Medium were saved. The program reads the image data from a corresponding drive, analyzes it and deletes detected Advertising sequences in the video recording or copies them.

In addition, can with the aid of the method according to the invention realized a program for automatic logging of advertising contributions become. This program can control both the time slot, the broadcast date, but also the context in which a particular advertising message is sent has been save.

To the primary Basic approach of the invention is for this purpose after activity testing of Video source a predetermined image section of an incoming video signal digitized by an image processor or by an AD converter and the resulting luminance values as so-called brightness matrix written in a memory. By averaging over several consecutive Read in brightness matrices of the same image section disorders in the form of image noise first reduced. In the case that a digital Video source is predetermined, both the digitization process and the averaging operation are omitted.

Out the resulting brightness matrix are by means of a, in one digital signal processor running image processing program, the edge courses determined in the current image section and in the form of a directional field stored as a matrix. By comparing the direction field matrices following pictures of the same picture section with the already stored direction field matrix become time stable parts in the stored direction field matrix or train by train certainly. After a certain number of measurements have been completed the time-stable portions of the direction field as an adjusted direction field matrix stored for reference purposes and for further comparative measurements used.

The In the method implemented reference measurement is dynamic, that is, that these have a different time despite the fixed number of measuring cycles can take. This is because image sequences in which the image content does not change substantially, with just one each single image into the reference measurement. This can be self Such foreign edges are eliminated, otherwise as unwanted time-stable Patterns would have emerged from the reference measurement.

As soon as within the reference measurement, the time-stable parts of the stored Direction field matrix have dropped below a certain level, becomes one started another reference measurement in a new screen section, and it begins the above sequence from the beginning. In difference The known state of the art is therefore not all Screen corners examined simultaneously, but it is a sequential Scan the screen corners for the sender logo.

The obtained after a reference measurement and adjusted for all foreign edges Directional field matrix then finds its actual use in the subsequent test measurement. Aim of the test measurements is it, the further presence of the transmitter logo in the as relevant Detect captured image detail, for example, an advertising-dependent device control to be able to make.

To become the direction field matrices of the image section to be evaluated with the direction field array stored as a reference continuously compared. In analogy to the formation of a path integral is the Sum of the angular deviations of the individual elements of the current one Directional field matrix over the Reference direction field matrix formed. The educated from this sum mean angle deviation is a measure of the probability of Presence of the sender logo.

As already mentioned, The procedure for advertising recognition can also be digital for post processing recorded videos, such as commercials later to eliminate or to record the sent advertising contributions and to statistical To evaluate purposes.

the device side is the video source over an image processor connected to a signal processor. The image processor digitized after being called by the signal processor the video signal of the video source and formats these signals. To Request by the signal processor are processed accordingly Signals forwarded to the latter, the signal processor based a stored in a non-volatile memory digital image processing program performs a signal evaluation. The signal processor provides at the output the switching signal, e.g. to the Operating a video recorder, so that this in commercial breaks the Recording automatically stops.

In addition exists in one embodiment of the invention, the possibility of the audio signal output the video source also to lead the signal processor to Based on a level evaluation an additional signal for detecting a commercial break to obtain.

The inventive device can be used as a separate so-called (external) auxiliary module But also already on the hardware side in a video recorder or another storage device be integrated. In the former case, the device is by a Plug connection in the manner of a scart, cinch or similar Principle connected to the output of any video source. The device in question performs that Video signal over another, second interface back out.

Over a integrated infrared transmitter and receiver unit is the device able to learn command sequences for device control and dependent on to be sent out by the advertising notice. Additionally or alternatively to Control over Infrared commands also have the option of commands over existing bus systems and / or transfer data to the devices to be controlled. A completely new Idea for this arises from the possibility of data exchange via SCART bus. For example, according to the widely used P50 standard Timer data of a video recorder can be read out and also changed. By Manipulation of the recording start or stop time can thus record video be controlled specifically. The control logic of the recorder ensures this even for the correct behavior when reaching the timer stop time.

Concrete contains the arrangement for detecting time-stable patterns in video signals any video source, a memory device, e.g. one Flash memory, for permanent storage of program code and Remote control command data, an image processor for digitizing analog video signals if required, and a digital signal processor for data analysis, data storage, as well as for the input and output of Signals over corresponding ports. In the case of an external device, there is an infrared transmitting and receiving unit for device control intended.

The inventive device remains after a system startup and an automatic bump test all components in a low-power standby state, too Sleep mode, until activated by valid video signals. After this Recognition of such signals is then an automatic transition in the active mode with the consequence of the execution of the presented method for recognizing commercials in TV programs or the like Video transmissions.

The Device sends it at the beginning and end of the advertising recognition process a command for initialization or shutdown of the controlled Process or device over the Infrared transmitter unit and / or an activated or connected Bus interface. Furthermore are each at the beginning and end of a recognized advertising sequence corresponding control commands sent out.

The Arrangement of the infrared transmitting and receiving unit for the purpose of advertising recognition dependent device control has at least five memories for recording infrared command sequences or sets them each of which is at least one, preferably three different Infrared commands over store the infrared receiver unit and the infrared transmitter unit can send one after the other. The infrared command sequences can be accessed via a standard remote control be entered learning.

embodiments of the subject invention are described below with reference to the drawing explained in more detail. It demonstrate:

1A . 1B a schematic representation of a correction of the vertical position of a measurement section due to blanking lines before the start of the reference measurement;

2 an exemplary representation of the cyclic Eckenabsuche after the transmitter logo during the reference measurement using a monitor image;

3 a schematic representation of the measuring range for detecting image content changes during the reference measurement;

4 a graphical representation for determining the local edge direction Ar [i] [j] from the eight different directional gradients;

5A . 5B a graphical representation for determining the statistical weighting factor Ag [i] [j] based on the difference between the two largest directional gradients;

6 a block diagram of a trained in the form of an internal or external ad block device;

7 a flowchart for starting the device designed as an internal ad blocker device;

8th a flow chart for the start operation of the designed as an external ad block device;

9 a flowchart for the user interface of the device designed as an external ad blocker device;

10 a flowchart for learning infrared remote control commands;

11 a flow chart for testing learned infrared remote control command sequences;

12 a flowchart for the process of the advertising recognition program;

13 a flowchart for the course of the reference measurement; and

14 a flow chart for the course of the test measurement.

The method for finding and recognizing time-stable patterns in video signals of any video source can be divided into two different measurement phases. The first phase is about finding a time-stable pattern that can be interpreted as a sender logo. This phase is called reference measurement below and is in the form of a flow chart in FIG 13 specified. In the second phase, the further presence of the time-stable pattern resulting from the reference measurement is checked. This named as test measurement and in the form of a flow chart in 14 phase shown represents the actual advertising recognition. To explain the method also in connection with the advertising-dependent device control, it is assumed in the following of a device for controlling a video recorder.

The task of reference measurement according to 13 is the finding of own logo in digitized video signals. For this purpose, the edge profiles of an image section are examined over a certain period of time in order to find time-stable parts using statistical methods. In this case, "edge courses" is to be understood as a directional field whose individual direction vectors reflect the local edge direction in each pixel.

immediate before the start of a new reference measurement is for initialization a device to be controlled, a command sequence by infrared signal or via the control lines of a Bus system sent. For example, many VCRs can be used during one Timer-programmed recording not controlled by remote control commands become. To circumvent this safety mode, there is for each Device one special remote control command that turns off this control function. The recorder leaves the timer programmed recording mode and must now in manual Recording mode continue to be operated. For this purpose become immediate before starting a new reference measurement, the necessary control commands sent.

For detection of a transmitter logo 1 (please refer 1 ) does not require the entire image area 2 (please refer 2 ) to investigate. The station logos 1 are basically in the corners 3 to 6 of the visible image. It will be the corner first 3 , for example, top left, as a picture to be examined 7 established. If a time-stable pattern could not be found during a reference measurement, the next corner will be displayed 4 to the active image section 7 and a new reference measurement begins. This process is repeated according to 2 cyclically until a time-stable pattern is determined within a reference measurement could.

In contrast to the known state of the art, therefore, not all screen corners 3 to 6 investigated simultaneously, but it is carried out according to the invention, a sequential screening of the screen corners 3 to 6 after the transmitter logo 1 , which is associated with a significant storage space savings.

As a sender logo 1 In order to serve as a detector for the advertising recognition, in its Kantenverlaufsrichtungen does not need to be fully captured, can by selecting the correct size of the image section 7 a further memory reduction can be made. In practice, dimensions of 90 lines × 90 columns have a meaningful size for the image section 7 proved.

However, the smaller the dimensions chosen, the more important is the correct positioning of the image section 7 (see. 1A . 1B ). Because the vertical position of some transmitter logos 1 is adapted to the respective image format of a TV program (4: 3, 16: 9, etc.), is also the vertical position of the image to be examined 7 to correct. This is before any new reference measurement, an investigation of the black blanking lines 10 required at the top or bottom of the picture. Exist such blanking lines 10 so they can according to 1A on the basis of one, through the entire current image section 7 extending, horizontal edge 11 be detected, depending on the corner just examined above or below this detected edge 11 no further edge courses must occur and the average brightness value must be smaller than that in the rest of the picture section. The current image section 7 is then corrected accordingly in its vertical position (see 1B ).

Ah[i][j]:= Element i, j der aktuellen Helligkeitsmatrix
i:= Zeilenindex
j:= SpaltenindexIt is necessary for further processing within a computer processor that the incoming video signal is in digital form. In the case of an analog video source, for this purpose, a video decoder, also called image processor, is used, which digitized on command of a computer processor, preferably a digital signal processor, an image area of the analog video signal and the computing processor available. For further processing, the brightness values are written by the arithmetic processor in the form of a matrix into a memory. By adding a few consecutively read brightness matrices of the same image section and subsequent averaging noise in the form of noise are significantly minimized. The resulting brightness matrix serves as the basis for the further analysis process. Example:

Ah [i] [j]: = element i, j of the current brightness matrix
i: = row index
j: = column index

Rh[i][j]:= Element i, j der Referenz – Helligkeitsmatrix
i:= Zeilenindex
j:= SpaltenindexThe first brightness matrix, which was formed by averaging a plurality of brightness matrices read in succession, is of particular importance in this case. It is copied as a so-called reference matrix into another memory area and serves as a comparison object for the subsequent measurements. This reference matrix should be formed by a sufficiently high number of averages, since it represents an important starting point for the further course of the reference measurement. Example:

Rh [i] [j]: = element i, j of the reference brightness matrix
i: = row index
j: = column index

To Storing the reference matrix becomes a new brightness matrix read in and the search for the transmitter logo begins.

For filtering the edge profile of a transmitter logo, the reference measurement searches for time-stable on divide in the direction field of the current image section. However, broadcasts that do not change their image content over a longer period of time may lead to incorrect results in the reference measurement. For example, the rigid studio background of a news program provides undesirable edge trajectories that would remain as time stable portions after the reference measurement.

therefore an image in the reference measurement is only considered if its image content opposite that of the previous picture noticeably changed Has. The read-in process is accordingly repeated as long as until a change of the image content could be determined.

E(D) < S ⇒ Keine Veränderung des Bildinhaltes
[Ah[i][j]]_n :=Element i, j der Helligkeitsmatrix des aktuellen Bildes
[Ah[i][j]]_n-1 := Element i, j der Helligkeitsmatrix des letzten gültigen BildesA simple way of checking for changes in the image content of the current image to the previous image is to calculate the average brightness change E (D). If the mean brightness change E (D) drops below a certain threshold S, then it is possible to conclude that there is static image content.

E (D) <S ⇒ No change in the picture content
[Ah [i] [j]] _n : = element i, j of the brightness matrix of the current image
[Ah [i] [j]] _n-1 : = element i, j of the brightness matrix of the last valid image

The range of blanking lines 10 This is not taken into account, as they can not display any changes in the image content. Also the pixels in which the transmitter logo 1 in all likelihood should be expected, should be excluded from the calculation. Another limitation arises from the observation that in the case of the already mentioned example of a news program the head of the newscaster temporarily in the range of the current image section 7 can protrude and thus despite a rigid studio background image changes would be detected. Because of this, only a small area remains 12 to study for changes in image content (see 3 ).

With Help of so-called compass operators becomes the respective edge direction determined in a specific pixel. A compass operator is a gradient operator that gives a statement about the brightness gradient in the Environment of a pixel with respect a certain direction. Perpendicular to the direction of an edge course generally the strongest Gradient magnitudes. So it is the task, with the help of the compass operators the direction with the strongest brightness gradient for each To find pixel.

Example for 3 × 3 compass operators:

wobei i den Zeilenindex, j den Spaltenindex, p den Zeilenindex des Kompassoperators und q den Spaltenindex des Kompassoperators bezeichnen.The definition for the calculation of a gradient amount Grad [n] in direction n with the help of a 3 × 3 compass operator is:

where i denotes the row index, j the column index, p the row index of the compass operator, and q the column index of the compass operator.

The determined eight direction gradients are degrees [n] of each pixel sorted and evaluated according to their amount.

The direction perpendicular to an edge basically provides the largest gradient amount. This direction is stored in the form of a direction code in the direction matrix Ar [i] [j] (see 4 ). Ar [i] [j] = n with degrees [n]> = degrees [m] and n, m ∈ {1, ..., 8} Ar [i] [j]: = element i, j of the current direction matrix
i: = row index
j: = column index

Not always there are the edges of a station logo, determined by the determination the strongest Direction gradients are determined. Especially transparent logos sometimes do not lift up enough from the image background or are in a texture rich background difficult to recognize. Therefore, the direction codes to be stored become still after the trustworthiness evaluated their direction statement. So it becomes a weighting factor for the determined individual elements of the directional field and in a separate (current) weighting matrix Ag stored. The weighting factors can for example, by the difference of the two strongest directional gradients be determined.

For example, in the left picture, the 5 ( 5A ) the gradient in the "direction 3" is much more pronounced than in the right image ( 5B ). In the end, this means that statistical evaluations in this case should give the direction code a higher significance, in other words a higher weighting. Ag [i] [j] = g1 - g2 Ag [i] [j]: = element i, j of the current weighting matrix
i: = row index
j: = column index
g1: = largest direction gradient
g2: = second largest directional gradient

Consequently Now there are three matrices, which are the currently imported image section characterize. The brightness matrix describes the luminance the individual pixels, the directional matrix the associated directional field and the weighting matrix the quality of the directional field elements in statistical evaluations.

As with the currently read brightness matrix, it is also possible to determine a directional and weighting matrix for the reference matrix. This results in the following matrix system:
Ah: = Current brightness matrix
Ar: = Current direction matrix
Ag: = Current weighting matrix
Rh: = reference brightness matrix
Rr: = reference direction matrix
Rg: = reference weighting matrix

Around the edge courses of the sender logo in the stored reference matrix of edges of the It is necessary to distinguish about a background image longer Time the direction fields of the currently incoming pictures with those of the reference picture. Only time-stable patterns like that Sender logo remain over the total duration of the reference measurement seen in their edge curves almost constant.

The directional deviation of an element of the current directional matrix with respect to the same element of the reference directional matrix is determined by the difference of the stored directional codes. The directional deviation w _ij is given here in the form of an angle in units of 45 degrees: w ij = | Rr [i] [j] -Ar [i] [j] | for w ij ≤ 4 w ij = 8 | Rr [i] [j] ar [i] [j] | for w ij ≥ 5 with Rr [i] [j] and Ar [i] [j] as defined on the previous page.

The sum of all directional deviations that results over a certain number of measurements for a pixel is a measure of the temporal constancy of the directional field at this point. Viewed over the number of n measurements, the result is an average directional deviation E (w _ij ), also called the expected value of the directional deviation, from:

wobei Rg[i][j] das Element in der Zeile i, Spalte j der Referenz-Gewichtungsmatrix und Ag[i][j] das Element in der Zeile i, Spalte j der aktuellen Gewichtungsmatrix bedeuten.An improvement of this statistical statement is achieved according to the invention in that the different quality of the individual contributions w _ij in the calculation of the mean directional deviation E (w _ij ) is taken into account as follows:

where Rg [i] [j] is the element in line i, column j of the reference weighting matrix and Ag [i] [j] is the element in of line i, column j of the current weighting matrix.

sowie die Summe der Gewichtungsfaktoren

in eigenen Matrizen festgehalt werdenIn order to be able to calculate this mean deviation of direction after n measurements, the weighted angle sum must be used for this purpose

and the sum of the weighting factors

become fixed content in own matrices

The mean deviation E (w _ij ) of the directional field vectors in a pixel after n measurements is therefore calculated from:

By specifying a suitable threshold value S, the time-stable components of the directional field can thus be determined at each time of the reference measurement. The following applies:
E (wij) <= S => Time Stable Directional Field Element
E (wij)> S => time-varying direction field element

Around to the existence of a transmitter logo in the viewed image detail shut down to be able to allowed to the geometric expansions of the time-stable direction field component not dropped below a certain level be. Once due to the dimensions of the time-stable directional field component the existence of a transmitter logo can be excluded started a new reference measurement in a new image section.

Before Now read in a new brightness matrix for further analysis is, the reference matrix Rh by addition to the current Brightness matrix Ah and subsequent averaging further improved. This minimizes noise and edges that are not for possibly existing transmitter logo belong already in the reference brightness matrix weakened.

The reference measurement ends when a time-stable pattern has been detected after a predetermined number of valid measurements. If this is not the case, the reference measurement is, as mentioned, repeated in a new image section (see 2 ). After a successful reference measurement, the time-stable components in the directional field of the reference matrix are marked in such a way that only these are taken into account in the subsequent test measurement for the purpose of recognizing the time-stable pattern.

The test measurement serves to detect the transmitter's own logos found in the reference measurement for the purposes of advertising recognition. To do so, according to 14 Just as in the reference measurement, the digitized video signals of the same image section are continuously read in and analyzed. By comparison with the time-stable portions of the directional field resulting from the reference measurement, it is possible to deduce the further presence or absence of the transmitter logo. Connected thereto are for controlling a video recording at the beginning and end of a recognized advertising sequence corresponding control signals via a bus system or transmitted via infrared transmitting unit to the recording device.

In one of the non-visible video lines of a television picture information about the transmitter and the originally planned broadcast date of the currently broadcast television program are transmitted (PDC-line). At the beginning of the test measurement, the information of this video line is read in and stored in order to detect by permanent comparison with the current data of this video line the end of a television program to be recorded and to send a corresponding control command to the video recorder.

in the Case of VTR control via the P50 protocol for the SCART bus be at the beginning of the test measurement also read out and save the timer data stored in the recorder. Was the VCR not in a timer programmed Recording, can this generated for the purpose of recorder control and transmitted via SCART bus become. By manipulating the timer stop time can now be the video recording to control specifically. The control logic of the recorder itself ensures this for the correct behavior when reaching the timer stop time.

Beginning and end of an advertising sequence are generally due to an image content change characterized. On the one hand, the sender logo appears or disappears, on the other hand it comes through the transition between advertising and film ultimately to an exchange of the whole Image content. That's why only images are evaluated, with which a picture content change goes hand in hand.

Exactly As in the reference measurement, the mean change is also suitable here the brightness values of the last image read and current Image. Unlike in the reference measurement, however, this time only the pixels along the logo edges are taken into account, those from the reference measurement have emerged. That makes sense, because by the appearance or disappearance of the transmitter logo will be in these places Make picture content change particularly clear.

E(D) < S ⇒ Keine Veränderung des Bildinhaltes
[Ah[i][j]]_n := Element i, j der Helligkeitsmatrix des aktuellen Bildes
[Ah[i][j]]_n-1 := Element i, j der Helligkeitsmatrix des letzten gültigen BildesIn this context, the previously stated relationship applies:

E (D) <S ⇒ No change in the picture content
[Ah [i] [j]] _n : = element i, j of the brightness matrix of the current image
[Ah [i] [j]] _n-1 : = element i, j of the brightness matrix of the last valid image

to presence inspection the transmitter logo, the edge curves of the current image with compared to the stored reference image. A measure of the deviations the edge courses of the current picture with respect to those of the stored reference image results from the calculation the mean directional deviation that the individual elements of the form the current direction matrix with those of the reference direction matrix. Just as in the reference measurement also becomes from a brightness matrix first calculated a directional and weighting matrix.

The directional deviation of an element of the current directional matrix with respect to the same element of the reference directional matrix is determined by the difference of the stored directional codes. The directional deviation w _ij is again given in the form of an angle in units of 45 degrees: w ij = | Rr [i] [j] -Ar [i] [j] | for w ij ≤ 4 w ij = 8 | Rr [i] [j] ar [i] [j] | for w ij ≥ 5

In The reference measurement before were temporal changes of the directional field viewed in one pixel. After n measurements an average value was obtained from:

In the test measurement will now be the current directional deviations of all those Pixels of the current image viewed from the reference measurement emerged as time-stable. The mean of the direction deviations E (W) of all pixels to be considered is calculated from:

Of the mean weighting factor E (G), with which the pixels to be considered are received, calculated from:

E (W) determines the deviations of the current direction field with respect to the Reference direction field and can be used as a probability statement about the presence of the sender logo be interpreted. The question of whether currently broadcast an advertising sequence can be answered by a suitable threshold. The state of the advertising recognition will be after every new evaluation compared with the previous state and stored. Change the Condition, so will, depending on whether it is the beginning or the the end of a recognized advertising sequence is a corresponding one Control command sent out.

In practice, it has been shown that it is beneficial to operate with two different thresholds, namely a threshold for the detection of an advertisement start and a threshold for the recognition of an end of advertising. Age condition = movie: E (W)> S ₁ ⇒ Advertising start Old condition = advertising: E (W) <S ₂ ⇒ End of commercial

E (G) is a measure of the quality of using E (W) statement made about the presence of the sender logo. If E (G) is very small, it means that for most of the pixels examined asked the question about the edge direction not clearly clarified could be. In such cases the old state of the advertising recognition is maintained and thus also no control command sent.

In spite of the enormous reliability Of the method described above, it can happen every now and then Bad decisions in logo recognition come, especially then if the sender logo is completely or partially in a same color or threatens to disappear strong textural image background. For this reason, the recognition of an advertisement start only applies as assured, if the sudden Missing the transmitter logo in the immediately following measurement approved could be. This approach has the consequence that in conjunction with the test described above on image content changes even hard-to-evaluate image sequences mastered without errors can be.

For example could the transmitter logo over disappear for a while in a background of the same color, without the advertising recognition makes a misinterpretation because only with a recognizable change of picture another picture for Evaluation arrives and for confirmation is used. The next evaluable image then shows the sender logo in a new context, in which it shows in any case but then the sender logo in a new context in which it will definitely be clearer again has to stand out from the background. The previously determined absence the sender logo would so not confirmed become.

The Evaluation security of the advertising recognition method according to the invention elevated by doing so, to almost 100%, without doing so the reaction rate is significantly impaired.

The End of the test measurement can be either manually or automatically by comparison with the previously saved PDC line information triggered become. With the end of the test measurement If necessary, a control command is sent and the entire advertising recognition process completed.

In the following, the circuit arrangement based on this method will now be described in the form of the device according to the invention 13 for advertising-dependent device control with respect to 6 to 14 described.

The device 13 is preferably used to control a video recording device and has the purpose to recognize even during the recording of a television program advertising sequences and optionally interrupt the video recording. The device 13 On the one hand, as a built-in video recorder assembly (internal ad blocker 14 ), on the other hand also as an additional device (external ad blocker 15 ) be formed. The cradle is simply connected to the video output of the recorder and automatically takes over the recording control at the beginning of the recording. For this the external ad blocker has 15 both via an infrared transmitter unit 16 as well as via an infrared receiver unit 17 , with the help of which any remote control commands can be stored and played back at a given time depending on the advertising recognition.

alternative The control of a video recorder is also about Any standard bus system conceivable. A completely new idea for this arises from the possibility of data exchange via SCART bus. After the widespread P50 standard can be For example, read timer data from a VCR and also change. By Manipulation of the recording start or stop time can thus record video be controlled specifically.

The following are the components of such a device 13 for controlling a video recorder for the purpose of advertising-free video recordings of TV broadcasts described. Here, between the conceptions of the internal and external ad blocker 14 . 15 according to 6 distinguished.

A digital signal processor 20 is the central processing unit. It is used for data analysis, data storage and generation of control signals. A permanent memory 21 , eg a so-called flash memory, holds the data consisting of program code and stored infrared command sequences without the need for a voltage source.

An image processor 22 , also called video decoder, is located between a video source 23 and the signal processor 20 and fulfills the following tasks: detection of a video source, digitization and decoding of analog video signals as well as image data acquisition of a desired image detail and synchronized output to the digital signal processor. It is clear that the image processor 22 in the case of the presence of digital signals of the video source 23 can be omitted.

A video trigger 24 is between the video source 23 and the signal processor 20 arranged and, provided the image processor 22 is provided, parallel to this connected to said components. The video trigger 24 constantly monitors the status of the video source in sleep mode 23 and signals the digital signal processor 20 the current state (active / inactive), one to the signal processor 20 connected watchdog circuit 25 the reset of the digital signal processor 20 supervised.

The infrared receiver unit 17 It is used to acquire infrared signals from any remote control command and send them in digital form to the signal processor 20 forward. It analyzes the received digital data and stores it in compressed form in non-volatile memory 21 from.

The infrared transmitter unit 16 is capable of control signals from the signal processor 20 to generate modulated infrared sequences. Digitally stored infrared command sequences are from the non-volatile memory 21 in the signal processor 20 loaded, decoded there and into control signals for the infrared transmitter unit 16 transformed. Both the video trigger 24 as well as the infrared transmitting and receiving unit 16 . 17 are as well as a control panel mentioned below 26 only with the external ad blocker 15 intended.

The control panel 26 is the user interface and contains switches for selecting the program mode (infrared (IR) program, IR test program, advertising recognition program) as well as buttons for selecting the IR memory locations. This component is, as mentioned, only in the external ad blocker 15 meaningful.

The internal ad blocker 14 (please refer 6 ) is a, in a video recording device not shown in detail integrated assembly, which from the previously in the external ad blocker 15 but not the IR units 16 . 17 and the control panel 26 consists. The purpose of the internal ad blocker 14 it is to recognize advertising sequences during the recording of a television program and to communicate this to the video recording device via a signal output. This can then optionally interrupt the video recording and continue with the end of advertising again. For the purpose of communication of ad blocker, hereinafter also referred to as module, and video recorder is not spoken below by a standard bus system, but by two special signal inputs and signal outputs.

At the module input START is according to 7 by a pulse of the video recorder in a power-saving sleep mode module located (internal ad blocker 14 ) and started the advertising recognition program.

At the Module input END is triggered by a pulse of the video recorder Advertising detection program ended and the module into a power-saving Sleep mode reset.

At the Module output STATUS indicates the functional readiness of the module.

At the Module output DETECT becomes the recognition state of the ad blocker program signaled to the outside. This will be an advertising-dependent VCR control enabled.

To the first time the power supply to the ad blocker a reset is carried out by the implemented watchdog circuit. Of the digital signal processor performs then do a self-test of the existing components. If one these tests fails, the signal processor signals to the outside the inoperability of the ad blocker the module output STATUS. After successful testing is the internal Advertising blocker in a power-saving sleep mode, from which he always from the VCR via a pulse at the module START input can be awakened. After the START pulse the advertising recognition program is executed until the video recorder has a Impulse at the module input END the ad blocker to end the advertising recognition process invokes and thus the module back into the power-saving sleep mode added. While The current advertising recognition program will be the state of the advertising recognition the video recorder notified the module output DETECT.

The external ad blocker 15 is a device in the form of a set-top box connected to the video output of any video source 23 is connected and has one or more device control options. It is preferably for the purpose of recognizing during the recording of a television program advertising sequences and notify the video recording device or any other device via a control bus or via infrared command sequences. This can then optionally interrupt the video recording and continue with the end of advertising.

After the first application of the power supply to the ad blocker according to the flowchart after 8th executed by the implemented watchdog circuit a reset. The digital signal processor then performs a self-test of the existing components. After the self-test, the signal processor signals to the outside the functional state of the ad blocker. When the external ad blocker is ready for operation 15 are via the control panel 26 Three different programs can be selected, the IR tutorial, the IR test program and the advertising recognition program (see 9 ).

The IR tutorial according to 10 The task is to analyze and store video recorder-specific remote control commands for controlling a video recording. Upon startup, the IR tutorial waits for a user selection from one of the five infrared command sequence locations. Each of these five memory locations is capable of storing at least one, preferably three, in the case of the first memory location, four different remote control commands, since some actions require the playback of multiple infrared commands. For example, a start of recording often only by the command sequence RECORD and PAUSE be executed. Despite the fact that the video recorder is supposed to start or stop only the video recording, two memory locations for receiving the associated remote control command sequences are generally not sufficient. As a rule, a video recorder can actually be controlled only by providing the five memory locations described below for receiving remote control commands.

Of the first memory space is to receive IR commands for initialization provided by the video recorder. This initialization leads the VCR using a special, device-specific remote control command from any timer programmed recording in the manually started recording mode. Only leave in this state all of them to disposal the remote control commands of the video recorder. These Initialization commands are sent out at the beginning of the reference measurement.

Of the second memory for recording the recording pause command is for provided the IR commands, which are played at the beginning of each recognized advertising sequence. These Commands are used to interrupt a live video recording.

Of the third memory is for recording IR commands for repetition the recording interruption provided. This is necessary because the Most VCRs have a recording interruption only for a particular one Hold the time (about 3 minutes) and then change to another operating state switch.

Of the fourth space is for holding IR commands to continue one interrupted video recording provided.

Of the fifth Memory space is available for recording IR commands for limit shutdown a video recording provided.

Once the user selection of a memory location has been made for receiving remote control commands, the program will signal its readiness. The infra-red signals of the remote control commands to be learned can now be transmitted through the infrared receiver unit 17 digitized and sent to the digital signal processor 20 to get redirected. The IR tutorial analyzes the received digital data, structures it and places it in non-volatile memory in compressed form 21 from. If one of the received IR commands could not be correctly received and processed, the operator control forces a re-entry.

The purpose of the IR test program is to send the previously stored VCR-specific remote control commands for controlling a video recording for test purposes. After the start the IR test program waits according to 11 to a user selection for one of the memory locations for infrared command sequences. The infrared command sequences digitally stored in the selected memory location become the non-volatile memory 21 in the digital signal processor 20 loaded, decoded there and into control signals for the infrared transmitter unit 16 transformed. The infrared transmitter unit generates infrared sequences modulated by the control signals of the signal processor.

The advertising recognition program is used to detect advertising contributions in TV programs by detecting the station's own logos, as described above. After the start of the advertising recognition program, this checks according to 12 whether a valid video signal from the image processor 22 is delivered. If no valid video signal is present, the advertising recognition program first puts the image processor 22 and then the digital signal processor 20 in a power-saving sleep mode. The video trigger 24 in sleep mode, constantly monitors the video source 23 and awakens the digital signal processor 20 from sleep mode if valid video data is present. Now, the ad detection program continues to run until it finally ends and the system is put back into sleep mode.

Around possible Make improvements to the advertising recognition program easily accessible, The following are several methods for updating program versions described.

In the four corners of the screen of a special television advertisement at the same time a specific logo, which is the start of an incipient data transfer signaled. The ad blocker switches by recognizing this startup logo in the update mode and awaits the receipt of the to be updated program version in the form of specific black and white patterns. Similar to the information transfer In the PDC line, this data is evaluated and stored in the non-volatile memory loaded. The end of such a transfer is signaled accordingly and leave the update mode. alternative could do this the program version to be updated also on a commercial Video storage medium have been applied.

One Updating via the internet presents itself with increasing spread of this Technology to update the program versions more and more at. This would have to the ad blocker too a corresponding PC interface, such as RS232, CENTRONICS, USB, etc. have.

Complementary exists the possibility, the advertising recognition by additional To improve parameters in cases where, in which an evaluation of the transmitter logo should not be possible, another Detection parameter is consulted. A possible detection parameter could for example, be the sound level. Obviously, advertisements are on television through a higher volume labeled as the actual program itself. For sound level analysis For this purpose is continuously in constant periods of calculated average sound level and with a predetermined threshold compared.

Claims (21)

A method for finding and detecting time-stable patterns in video signals of any video source, in particular for the detection of advertisements in television broadcasts or similar transmissions, by video signal analysis and detection of transmitter-specifying video signals, such as logos, characterized in that: (a) the activity of the video source for determining at least one image section for reading in digitized form is checked as a brightness matrix in a memory; (b) the edge profiles of the current image section in the form of a direction field are determined from the brightness matrix previously read in according to step (a) and stored as direction field matrix; (c) by comparing the direction field matrices of subsequently read-in images of the same image detail with the stored direction field matrix, referred to below as comparison measurement, the time-stable portions of the direction field are determined and analyzed according to certain criteria with regard to the presence of a time-stable pattern; (d) if, after one of the comparison measurements, a time-stable pattern can not be recognized, the method steps (a) to (c), referred to hereinafter as reference measurement, are repeated for a new image section; (e) if after a certain number of comparative measurements the existence of a time-stable pattern has not been excluded, the presence of a time-stable pattern is concluded, the reference measurement is terminated and the determined time-stable portions of the directional field are stored as an adjusted direction field matrix and used for further comparison measurements; (F) by comparing the direction field matrices of subsequent images of the same image detail with the adjusted direction field matrix using statistical methods on the further presence or absence of the time-stable pattern is closed. Method according to claim 1, characterized in that that the statistical statement about the further presence of the time-stable pattern from the sum of Deviations of the individual elements of the current direction field matrix across from the adjusted direction field matrix is obtained. Method according to claim 1 or 2, characterized that each element of the direction field matrices has a weighting factor is assigned such that these elements with their respective Weightings into the statistical calculations. Method according to one of claims 1 to 3, characterized that a review of the current image section on image content changes takes place in such a way that image sequences with approx steady Image content does not reach the evaluation. Method according to claim 4, characterized in that that the disappearance or reappearance of the from the reference measurement emerged time-stable pattern only then as proven applies if the image data of the immediately following measurement this Confirm state change. Method according to one of the preceding claims, characterized characterized in that updated as a result of an external updating Evaluation algorithms for execution of the method can be applied. Method according to one of the preceding claims, characterized marked that in addition to the video signal, an audio signal is received, which is for evaluation preferably arrives at a digital signal processor in order to significant level and / or frequency differences due to changes in Close video signal. Method according to one of the preceding claims, characterized characterized in that based on a recognized advertising contribution a switching or control signal for influencing a connected device, preferably a video recorder. Use of a method according to one of the preceding claims for the post-processing of digital or analogue stored video data by means of a personal computer or the like computing device and / or for automatic logging of advertising contributions. Device for finding and detecting time-stable patterns in video signals of any video source, in particular for the detection of advertisements in television programs or the like About transmissions, by video signal analysis and detection of transmitter-specifying video signals, such as logos, characterized by: (a) means for checking the activity of the video source to determine at least one image portion for reading in digitized form as a brightness matrix into a memory; (b) a device for determining the edge profiles of the current image section from the brightness matrix previously read in accordance with feature (a) in the form of a direction field and storing this as a direction field matrix; (c) a device for comparing the directional field matrices of subsequently read-in images of the same image section with the stored directional field matrix, referred to below as comparison measurement, and for determining the time-stable components of the directional field and for analyzing the time-stable components according to specific criteria with regard to the presence of a time-stable pattern; (d) means for, if after one of the comparison measurements, a time-stable pattern can not be recognized, for repeating the effecting of checking, determining, comparing and analyzing according to features (a) to (c), hereinafter referred to as reference measurement, in a new image detail ; (e) means, if after a certain number of comparison measurements the existence of a time stable pattern has not been excluded, to conclude the presence of a time stable pattern, terminate the reference measurement and store the determined time stable portions of the direction field as an adjusted direction field matrix and to use these shares for further comparison measurements; (F) means for comparing the direction field matrices of subsequent images of the same image detail with the cleaned direction field matrix and for determining the further presence or absence of the time-stable pattern using statistical methods. Apparatus according to claim 10, characterized in that one with a video source ( 23 ) connected digital signal processor ( 20 ) is provided for data analysis, data storage and for controlling the input and output of signals, which processor with a non-volatile memory ( 21 ) is connected to the storage of program code and remote control command data and is designed in such a way that the processor can be used in the non-volatile memory ( 21 ) a signal evaluation of image sections ( 7 ) and at its output provides a control signal, wherein the control signal from the non-volatile memory ( 21 formed remote control command data is that an infrared receiver unit ( 17 ) by means of which the required remote control command data is sent to the signal processor ( 20 ), which analyzes these data and stores them in non-volatile memory ( 21 ) and that an integrated infrared transmitter unit ( 16 ) provided from the control signals of the signal processor ( 20 ) generates modulated infrared command sequences for controlling another device. Device according to claim 11, characterized in that the signal processor ( 20 ) is designed such that it transmits the infrared command data from the non-volatile memory ( 21 ), decoded and transmitted via the infrared transmitting unit ( 16 ). Apparatus according to claim 11 or 12, characterized in that in the presence of analog signals at the output of the video source ( 23 ) between the video source and the signal processor ( 20 ) an image processor ( 22 ), which image processor is designed such that it extracts image sections ( 7 ) of the video signal of the video source is digitized, formatted and sent to the signal processor ( 20 ) when prompted by the signal processor. Device according to one of Claims 11 to 13, characterized by a video trigger ( 24 ) between the video source ( 23 ) and the signal processor ( 20 ) and, if an image processor ( 22 ) is provided, is provided in parallel to this, so that the device from a power-saving standby or sleep mode out into an active state is then convertible when relevant video signals are detected. Device according to one of claims 11 to 14, characterized through an output video interface, that of the device used video interface for other devices accessible power. Device according to one of claims 11 to 15, characterized in that in addition an audio signal output of the video source ( 23 ) with the signal processor ( 20 ) connected is. Device according to one of Claims 11 to 16, characterized by a connection interface to the output of the video source ( 23 ) and a watchdog circuit ( 25 ) for monitoring the reset of the digital signal processor ( 20 ). Device according to one of claims 11 to 17, characterized in that the infrared transmitting and receiving unit (16, 17) has a plurality of storage locations for receiving infrared command sequences, each storage space of which receives a plurality of infrared commands via the infrared receiving unit (16, 17). 17 ) which commands can be sent via the infrared transmitting unit ( 16 ) can be discontinued successively. Device according to claim 18, characterized in that that five memories for each at least one, preferably three instruction sequences are provided, the first storage space for Commands for initializing preferably a connectable VCR via its IR interface, the second location for initial recognition commands, the third storage space for Commands for repeating the recording interruption of the recorder, the fourth storage space for Commands to continue the interrupted video recording and the fifth memory space for commands is provided for the end shutdown of the recording. Device according to one of claims 11 to 19, characterized that instead of infrared interfaces and commands a standard bus system is provided. Device according to claim 20, characterized in that that the further device is a video recorder and that over the Bus system allows data exchange, so for example the timer data of the video recorder and read for the purpose of Recording control are specifically manipulated.