FR3054396B1 - SYSTEM AND METHOD FOR MEASURING CENTERED-AUDIENCE AUDIENCE, CAPTURING AND ANALYZING IMAGES DISPLAYED BY A TERMINAL ASSOCIATED WITH AT LEAST ONE PANELIST. - Google Patents

Abstract

There is provided a user-centered audience measurement system comprising: at least one server (4) for measuring content consumption by a terminal (1); image capture means (2) configured to generate, from images displayed by the terminal, digital images and / or intermediate signals (allowing to reconstruct at least partially digital images) and / or digital prints ; analysis means (5) configured to determine at least one information relating to the consumption of content by the terminal, as a function of the digital images and / or intermediate signals and / or digital fingerprints generated by the capture means; images; and associating means (6) configured to associate the at least one information relating to the content consumption by the terminal, with a single panelist, if the terminal is of the non-shared type, or a list of panelist (s) declared (s) if the terminal is of the shared type.

Description

System and method for user-centric audience measurement, by capturing and analyzing images displayed by a terminal associated with at least one panelist.

1. TECHNICAL FIELD

The field of the invention is that of audience measurement.

More specifically, the invention relates to a user-centric ("user-centric") solution (system and method), in a context where at least one audience measurement server is configured to measure a user-centric audience. content consumption by a terminal associated with at least one panelist.

It applies in the case where the measured consumption is a consumption of content (by the terminal) on a computer network (including but not exclusively Internet). In the rest of the description, this first type of audience measurement is called (for simplification and abuse of language) "Internet audience measurement".

It also applies in the case where the measured consumption is a consumption (by the terminal) of TV and / or video content. In the rest of the description, this second type of audience measurement is called "TV and / or video audience measurement".

2. TECHNOLOGICAL BACKGROUND 2.1 Internet audience measurement

According to a first approach, the "user-centric" measure is based on the monitoring of the navigation (also called "surfing" or "network traffic") on the Internet (or another computer network) of a panel (sample of Internet users also called panellists) representative of a population of Internet users (for example French Internet users). By abuse of language, the network traffic of a terminal is sometimes also called "Internet audience" of this terminal. Panel navigation is measured and a scoping survey is done in parallel to extrapolate their navigation to the Internet population studied. By precisely identifying the panellist, the "user-centric" measure is used to analyze the number of unique visitors who visited a site during a given period. A single panel visitor is a qualified individual identified. In extrapolated value, a panelist corresponds to thousands of unique visitors. The "user-centric" measure provides agencies, publishers and marketing professionals with accurate and reliable reports on the audience of Internet sites and the use of Internet users, which information can be used for strategic planning, competitive studies or media planning. For example, the results are updated every month.

To measure navigation (surfing, network traffic) on the Internet, or another computer network, a terminal panelist (sometimes speaking, by language abuse, measurement of Internet audience of a panelist terminal), different techniques may be used: 1. Browser extension: A browser extension (plug-in) is installed on each of the Internet browsers of the terminal in order to intercept all the URLs (for "Uniform Resource Locator" or "Uniform resource locator" ) HTTP requests (for "HyperText Transfer Protocol" or "HTTPS" for "Secured HTTP" or "Secure HTTP") issued by the browser. URLs for HTTP / HTTPS requests are also referred to as "traffic information" in the rest of the description. 2. Home router equipment: a router equipment is connected to the fireplace box (ADSL modem-router or cable); Panellist households are asked to connect their terminals to this router rather than to the box so that the router, modified specifically by the audience measurer, collects in passing the URLs of requests sent to the Internet. 3. Proxy configuration: The network configuration of the terminal is set to pass traffic through a proxy (proxy server) that establishes the connection to the destination Internet servers and collects the HTTP request URLs in passing. Note: The proxy can be a remote server operated by the audience measurer in its back office or an internal software process to the terminal to be measured that is set up during the installation of a measurement application ( meter). 4. VPN + proxy configuration: A VPN ("Virtual Private Network" or "VPN") configuration is established between the terminal and a VPN server behind which a proxy is placed to collect traffic URLs. Unlike the previous option which requires a configuration of each Wifi connection (or more generally each connection to a wireless LAN meeting one of the IEEE 802.11 standards) or cellular connection (for example type 2G, 3G, etc.), the VPN has the advantage once configured, to channel traffic from all connections into a single secure channel leading to the VPN server and proxy. Note: The VPN server and proxy may also be internal to the terminal, although this implementation is technically complex to implement. With the exception of the browser extension (solution 1 above) that captures the URLs at the source, the other techniques (solutions 2 to 4 above) proceed by capturing the network traces. In the case of HTTPS traffic, once the connection is established to the server, the traffic is encrypted so HTTPS URLs are inaccessible. The measurement is limited to the knowledge of the recipient server. To work around this difficulty, we proceed to a "man-in-the-middle" type interception type attack using a proxy that intercepts the server certificate and sends another certificate to the client application. . This certificate sent by the proxy will be verified by a root certificate from the audience measurer and that the panelist must install on his terminal. In this way, it is possible to register HTTPS URLs at the proxy level.

In recent years, mobile devices (tablets and smartphones) are increasingly used to access the Internet everywhere. Measuring the audience on these mobile devices has become paramount. Unfortunately, the four known Internet audience measurement technologies listed above have various disadvantages, especially for mobile terminals equipped with operating systems (or OS) for mobile "iOS" (registered trademark) and " Android (registered trademark): 1. Browser extension: on the one hand, this solution is limited to the use of the browser, which excludes applications downloaded from an application store; on the other hand, browsers on mobile OS do not offer the possibility to install extensions like on a computer. 2. Home router equipment: this solution is limited to home Internet consumption, which excludes out-of-home consumption which is obviously very important for smartphones and to a lesser extent for tablets. 3. Proxy configuration: With the operating system "Android" it is now impossible to perform a proxy configuration in a programmatic way, probably in order to counter the ad-blockers that implement a proxy (usually local) to filter the connections to the advertisement servers (Ad-servers). As it would be very laborious to ask a panelist to manually configure the IP address and proxy port for each new network connection, this solution suddenly became obsolete for the Android mobile OS. 4. VPN + proxy configuration: To prevent Internet users from using a VPN server located abroad to access inaccessible movie catalogs in their territory, it is not impossible that the VPN functionality is also banned future versions of mobile OS, leaving this feature reserved for business use from devices dedicated to this purpose (eg an "iPad pro" (registered trademark)).

Another technological evolution threatens the measurement of Internet audience: the democratization of the HTTPS protocol (which also becomes mandatory from January 1, 2017 for any application hosted on the Apple application store). Until now proxy-based measurement systems could perform a "Man-In-The-Middle" attack at the proxy level in order to intercept the HTTPS connection and save the URL in passing. Unfortunately, application editors are increasingly using the "Certificate Pinning" technique to refuse any certificate different from that of the publisher; this allows them to prevent any "Man-In-The-Middle" attack.

We also see that the Google company will provide in the next version of Android ("Android N") tools facilitating the work of developers to enhance security ("Network Security Configuration"): • with this mechanism based on a simple XML file, an application developer can simply set / change the security settings of his application without having to recompile. This will simplify the task, which should further facilitate technologies to increase the level of security (eg Certificate Pinning); • This XML file will allow to simply define the Certification Authorities that will verify the certificates sent by the HTTPS servers (a priori a subset of the system certificates). By default, apps targeting Android N will trust only system certificates and not those added (such as the audience measurer root certificate added by the meter to make a Man in The Middle attack). • This XML file will also make Certificate Pinning by specifying which server certificate is expected by the application, which blocks any attack attempt required to measure the HTTPS. If it is no longer possible to perform a "Man-In-The-Middle" type attack, the number of sites and Internet applications accessible in HTTPS, which are becoming more numerous, will be limited to domain names and time spent on these areas. For example, it will no longer be possible to distinguish "facebook.com/lemonde.fr" and "facebook.com/lesechos". 2.2 TV and / or video audience measurement

To measure the consumption of TV and / or video content of a panelist terminal (sometimes speaking, by language abuse, measurement of TV audience and / or video of a panelist terminal), different techniques can be used according to the type of terminal (TV or Internet terminal).

In the case of a television, most of the content being historically broadcast by satellite, cable or terrestrial broadcasting networks, without a return path, the audience measurers have developed tools capable of identifying content via its channel. audio, regardless of the decoder used. Thus, two large families of detection technology are used: • "Watermarking audio" (tattooing digital audio) tattoo the audio channel in the publisher control, with a unique identifier of the editor and a time stamp. The panelist's audimeter has only to analyze the sound coming out of the television to determine the content (TV channel for example) and a possible time lag; • "Fingerprinting audio" does not require any intervention in the publisher's control room. It consists of continuously extracting the audio footprint of each string in order to constitute a reference base and to compare the fingerprint captured by the panelist's audiometer with the various possible reference fingerprints. The nearest print indicates the identity of the source (TV channel for example).

In the case of an Internet terminal (computer, tablet or smartphone), TV and / or video content (live TV, TV off-line, catch-up TV, video on demand (VoD) , etc.) are for example broadcast on the Internet in continuous mode (also called "streamed content" or "streaming content") or in download mode (for example pre-roll, mid-roll and post-roll downloaded in file). The contents are available on Internet sites (referenced content in web pages that can be viewed with a browser run by the terminal) or on third-party applications (also called "mobile applications" or "editor applications") run by the terminal.

If digital sound acquisition is possible, or if it is possible to use the microphone of the terminal to capture and analyze the sound coming out of it, then both families of detection technology ("Watermarking audio And "Fingerprinting audio") can be used (as for a TV).

Otherwise, another solution is to use a marking technology ("Web analytics") integrated into the video player ("video player"). The marker (for example "eStat streaming") then sends HTTP requests (called "hits") on the measuring nodes of the audience measurer to report on the operating status of the player (play / pause / stop) and its temporal index in the content. This technology works in "site-centric" mode, so it works for all users. In a panel context, a technique for measuring Internet traffic (like those mentioned above, in paragraph 2.1) is implemented in the panelist's environment, in order to capture the hits emitted by its terminal and to measure its audience.

For TV and / or video audience measurement, current solutions are more durable than for Internet audience measurement, but there are some limitations: • Some publishers refuse to perform a digital audio tattoo ("Watermarking audio" ) of their contents (ex: Netflix). Only the technology of "audio fingerprinting" is then possible, but in case of diffusion of the same content by several broadcasters (for example the wishes of the president broadcast on several television channels simultaneously, or a video content common to several services of VoD by subscription), the "Fingerprinting audio" technology does not make it possible to determine the broadcaster (TV channel or VoD service in the aforementioned examples) with certainty; • for some Internet terminals, access to the digital audio channel is not possible (typically with the "Android" and "iOS" operating systems) to use "Watermarking audio" or "Fingerprinting" technologies audio "; one possible solution is to use the terminal's microphone to capture and analyze the sound coming out of it. Unfortunately, when listening to headphones, no sound is accessible. Watermarking audio and audio fingerprinting technologies are therefore not entirely satisfactory; • The marking technology ("Web analytics") involves a lot of work from all content providers (publishers, aggregators such as Internet Service Providers ...) who must integrate the marker in their website and in their mobile applications. which means for aggregators that you have to access metadata about publisher content. Some, like for example Youtube or Netflix, do not want to do this work of integration with all the audience meters in the world.

3. OBJECTIVES The invention, in at least one embodiment, has the particular objective of overcoming these disadvantages of the state of the art.

More specifically, in at least one embodiment of the invention, one objective is to provide a user-centric audience measurement solution ("user-centric") that: • does not require the installation of an extension browser; • is not limited to home consumption, but includes mobile consumption; • does not require a proxy configuration; • does not require a VPN configuration; • does not require a "Man-In-The-Middle" type attack; • does not require access to the audio channel or use the terminal microphone; and • does not require the implementation of web analytics technology at the website and application level.

Another objective of at least one embodiment of the invention is to provide such a solution that makes it possible to determine the broadcaster of a TV and / or video content, even if the same content is broadcast by several broadcasters. .

Another objective of at least one embodiment of the invention is to provide such a solution that is compatible with the use of terminals shared by one or more panelists (and possibly one or more non-panelists) within the same home.

Another objective of at least one embodiment of the invention is to provide such a solution that is simple to implement and inexpensive. 4. SUMMARY

In a particular embodiment of the invention, a user-centered audience measurement system is proposed, comprising at least one audience measurement server configured to measure content consumption by a terminal associated with at least one panelist, the system comprising: image capturing means, included in or placed on or near the terminal, and configured to generate, from images displayed by the terminal, digital images and / or intermediate signals, at least partially reconstructing digital images, and / or digital fingerprints; analysis means, at least partially included in or cooperating with said at least one audience measurement server, and configured to determine at least one information relating to said consumption of content by the terminal, as a function of the digital images and / or intermediate signals and / or digital prints generated by the image capturing means; and association means, configured to associate said at least one piece of information relating to said content consumption by the terminal, with a single panelist, if the terminal is of the non-shared type, or a list of panelist (s) declared (s) ), if the terminal is of the shared type.

Thus, the proposed solution is based on a completely new and inventive approach, to capture and analyze the images displayed by the panelist's terminal in order to extract relevant information for the audience measurement. In this way, we know exactly what you're talking about (s) panelist (s).

As detailed below, and incorporating the definitions given above, this solution applies to TV and / or video audience measurement and Internet audience measurement.

This solution differs, in principle, from the known techniques of Internet traffic measurement discussed above (in section 2.1) and therefore has several advantages over these: • it does not require the installation of an extension browser; • it is not limited to home consumption, but includes mobile consumption; • it does not require a proxy configuration; • it does not require a VPN configuration; • It does not need to perform a "Man-In-The-Middle" attack (it allows to analyze the internet consumption of a panelist even when it is no longer possible to intercept the HTTPS traffic because the increase in the level of security on the Internet).

This solution is also distinguished, in its very principle, from the known techniques for measuring TV and / or video traffic discussed above (in section 2.2) and thus has several advantages over these: • it does not require access the audio channel or use the terminal microphone; and • it does not require the implementation of web analytics technology at the website and application level.

In a first particular implementation, the image capture means comprise a camera or camera-like camera, external to the terminal and configured to capture at least some of the images displayed by the terminal and generate digital images.

An advantage of this first implementation is that the camera can be selected from many existing models and inexpensive. Another advantage is that the camera can directly generate digital images (simplification of treatments), with sufficient resolution to achieve better levels of image analysis.

In a second particular implementation, the image capture means comprise, within the terminal, a screen capture module produced in hardware and / or software and configured to capture at least some of the images displayed by the terminal and generate digital images.

An advantage of this second implementation is that the screen capture module is internal to the terminal. Another advantage is that the screen capture module makes it possible to directly generate digital images (simplification of the treatments), with a sufficient resolution to reach better levels of image analysis.

Advantageously, in the case of the second particular implementation, the terminal comprises a traffic measurement module configured to activate the screen capture module in case of launching the execution by the terminal of a client application whose traffic does not can not be measured by the traffic measurement module.

Thus, the screenshot module is launched only if necessary.

In a third particular implementation, the image capture means comprise at least one transparent film, configured to be placed against a screen of the terminal, driver if the screen of the terminal is tactile, and comprising or cooperating with a network of photo- detectors generating digital images and / or intermediate signals.

An advantage of this third implementation is that it does not require a camera or camera-like camera external to the terminal (which can be bulky, not elegant and lead to distorted and / or poorly framed images, in which the terminal screen is missing or too small to extract information), nor screen capture module internal to the terminal. Another advantage is that the transparent film can be used to protect the screen of the terminal to protect it from scratches or breakage in case of drop of the terminal.

Advantageously, in the case of the third particular implementation, said at least one transparent film is configured to function as a luminescent concentrator and to cooperate with said array of photo-detectors, the photo-detectors being positioned on at least a portion of the outer edges. said at least one film, said photodetector array being configured to generate intermediate signals for at least partially reconstructing digital images.

This implementation makes it possible to avoid opacifying the film with sensors and circuits.

Advantageously, in the case of the third particular implementation, said at least one transparent film comprises said photodetector array which is configured to generate digital images.

This variant makes it possible to generate digital images directly (simplification of the treatments), with a sufficient resolution to reach better levels of image analysis.

In a first application (called "TV and / or video audience measurement", with the definition given above), said consumption of content by the terminal is a consumption of TV and / or video content, the means for capturing images comprise means for generating first fingerprints from the digital images and / or intermediate signals, and the analysis means comprise: means for calculating and storing second fingerprints, of reference, from TV contents and / known video or video; and means for identifying the TV and / or video contents consumed by the terminal among the known TV and / or video contents, as a function of a comparison between the first and second imprints.

Thus, it is possible to perform a TV and / or video audience measurement using a video digital fingerprint detection technique ("Fingerprinting video").

Advantageously, in the case of the first particular application, the intermediate signals are assimilated to first impressions extracted from the TV and / or video content consumed by the terminal.

In this way, in the case where the capture means generate intermediate signals (and not directly digital images), it is not necessary to systematically carry out an image reconstruction, which can be costly in terms of resources and / or or in computing time.

Advantageously, in the case of the first particular application, the image capture means or the analysis means comprise means for at least partial reconstruction of digital images from the intermediate signals generated by the image capture means. , and the analysis means comprise means for detecting, in the reconstructed digital images, at least one identification element of a broadcaster of the TV and / or video contents consumed by the terminal.

Thus, when the capture means generate intermediate signals, it is possible (by analysis of the content of the reconstructed images from the intermediate signals) to determine the broadcaster of a TV and / or video content, even in the case of broadcast. the same content by several broadcasters. The image reconstruction is preferentially carried out only if the technique of "Fingerprinting video" leads to an ambiguity that must be removed (ie the content identified having been broadcast by several broadcasters, it is sought to determine the broadcaster of the content consumed by the terminal ).

Advantageously, in the case of the first particular application, the analysis means comprise means for detecting, in the digital images generated by the image-capturing means, at least one identification element of a broadcaster. TV and / or video content consumed by the terminal.

Thus, when the capture means directly generate digital images, it is possible (by analyzing the content of these digital images) to determine the broadcaster of a TV and / or video content, even if the same content is broadcast. by several broadcasters.

In a second application (called "Internet audience measurement", with the definition given above), said consumption of content by the terminal is a consumption of content on a computer network, the image capture means or the means of analysis comprise means for at least partial reconstruction of digital images from intermediate signals generated by the image-capture means, and the analysis means comprise detection means, in the reconstructed digital images, of at least one information relating to the consumption of content on the computer network.

Thus, when the capture means generate intermediate signals, it is possible (by analyzing the content of the digital images reconstructed from the intermediate signals) to determine one or more information relevant for the measurement of the Internet audience.

In a variant of the second application, said content consumption by the terminal is a consumption of content on a computer network, the analysis means comprise detection means, in the digital images generated by the image capture means, at least one information relating to the consumption of content on the computer network.

Thus, when the capture means directly generate digital images, it is possible (by analyzing the content of these digital images) to determine one or more information relevant for the measurement of the Internet audience.

Advantageously, in the second application or its variant, said at least one information relating to the consumption of content on the computer network belongs to the group comprising: information relating to a support site on which a content is consumed; information relating to a support application on which content is consumed; information relating to an on-board multimedia player of a broadcaster, with which content is consumed; and information relating to a content.

The nature and relevance of the determined information may depend on the resolution of the digital images used (digital images reconstructed from the intermediate signals, or digital images generated directly by the capture means).

In another embodiment of the invention, there is provided a user-centered audience measurement method, implemented in a system comprising at least one audience measurement server configured to measure a content consumption by a user. terminal associated with at least one panelist, the method comprising: an image capture step, configured to generate, from images displayed by the terminal, digital images and / or intermediate signals, making it possible to reconstruct at least partially digital images, and / or digital prints; an analysis step, configured to determine at least one information relating to said content consumption by the terminal, as a function of the digital images and / or intermediate signals and / or digital fingerprints generated in the image capture step ; and an association step, configured to associate at least one captured image in the image capture step, in the form of intermediate signals or directly of a digital image, with a single panelist, if the terminal is of the type non-shared, or a list of panelist (s) declared, if the terminal is of the shared type.

This method has the same advantages as those discussed above for the system.

5. LIST OF FIGURES Other features and advantages of the invention will appear on reading the following description, given by way of indicative and nonlimiting example, and the appended drawings, in which: FIGS. IB (process) illustrate the general principle of the invention; FIG. 2 presents a block diagram of a system according to a first embodiment of the invention, for a TV and / or video audience measurement; Figures 3 and 4 illustrate a particular embodiment of the image capture device shown in Figure 2; FIG. 5 presents a block diagram of a system according to a second embodiment of the invention, for a TV and / or video audience measurement; FIG. 6 shows a block diagram of a system according to a third embodiment of the invention, for a TV and / or video audience measurement; FIG. 7 presents a block diagram of a system according to a fourth embodiment of the invention, for a TV and / or video audience measurement; and Figure 8 shows a block diagram of a system according to a fifth embodiment of the invention, for an Internet audience measurement;

6. DETAILED DESCRIPTION

In all the figures of this document, the elements and identical steps are designated by the same numerical reference.

As illustrated in FIG. 1A, the context is that of a user-centered audience measurement system, comprising (at least) an audience measurement server 4 configured to measure a consumption of content by a terminal 1 associated with (at least) a panelist.

Audience measurement is, for example, a measure of TV and / or video audience, or a measure of Internet audience. According to the definitions given above, in the first case the consumption of TV and / or video content by the panelist terminal is measured, whereas in the second case the network traffic (on the Internet or another computer network) of the panelist terminal is measured. .

In the case of a TV and / or video audience measurement, the terminal 1 is either a television receiving the TV and / or video contents via a broadcasting network (terrestrial, cable, satellite or ADSL), or an Internet terminal. (computer, tablet or smartphone) receiving the TV and / or video content (live TV, delayed TV, catch-up TV, VoD, etc.) via the Internet 3 or another computer network. To access the contents, the Internet terminal runs a browser (for viewing and displaying web pages of Internet sites) or a third-party application (also called "mobile applications" or "publisher applications").

In the case of an Internet audience measurement, the terminal 1 is an Internet terminal (computer, tablet or smartphone).

With the proposed solution, the system further comprises image capture means 2 (hereinafter also referred to as "image capture device"), analysis means 5 and association means 6.

The image capture device 2 generates, from images displayed by the terminal 1, digital images and / or intermediate signals for at least partially reconstructing digital images. The image capture device 2 is (at least partially) included in the terminal 1, or it is placed on or close to it (case illustrated in Figure IA).

The analysis means 5 determine at least one information relating to the consumption of contents by the terminal, as a function of the digital images and / or the intermediate signals generated by the image-capture device 2. The analysis means 5 are at least partially included in the audience measurement server 4, or cooperate with it (case shown in Figure IA).

The association means 6 associate the at least one information relating to the consumption of content by the terminal, with a single panelist, if the terminal is of the non-shared type, or a list of panelist (s) declared (s), if the terminal is of the shared type.

Thus, and as illustrated in FIG. 1B, the general principle of the proposed solution (new paradigm) consists in capturing (step S1) and analyzing (step S2) the images displayed by the terminal 1 of the panelist in order to extract information from them. relevant for the audience measurement, and to associate this information with one or more panelists (step S3). This new paradigm is based on the increase of computational powers and the bit rates of communication networks, as well as on the appearance of machine learning techniques ("machine leaming").

With reference to FIG. 2, a system according to a first embodiment of the invention is presented for a TV and / or video audience measurement.

In this first embodiment, the image capture device is an optical audimeter 22, generating intermediate signals (assimilated to first fingerprints extracted from the TV and / or video content consumed by the terminal) making it possible to reconstruct at less partially digital images. A particular embodiment of the optical meter 22 is shown below.

Moreover, the analysis means comprise: means for generating second fingerprints, referred to as reference fingerprints, from known contents (for example the contents (programs) of different TV channels); in the example presented, these means comprise, for each TV channel, a terminal 26, 26 'equipped with an optical meter 25, 25' (of the same nature as that 22 equipping the terminal 1); A database 27 containing the reference fingerprints; and in the audience measurement server 4, a fingerprint management module 28 (made in hardware and / or software form) which stores the reference fingerprints in the database 27, and collects (via the Internet 3 ) the first fingerprints and compares with the reference fingerprints, so as to identify TV and / or video content consumed by the terminal among known TV and / or video content.

It should be noted that in this first embodiment of the invention, it is not necessary to reconstruct the images (expensive operation in computing time). Indeed, the fingerprint management module 28 works with the first fingerprints, that is to say the intermediate signals generated by the optical meter 22.

It should also be noted that the comparison of fingerprints described above makes it possible to identify a TV and / or video content consumed by the terminal, but it does not make it possible to identify the broadcaster of the content consumed, in the event that this content is broadcast by several broadcasters (see examples above: President's greetings broadcast on several simultaneous television channels, or video content common to several VoD services by subscription). To overcome this uncertainty, one solution consists in performing the following complementary processing, for example in the fingerprint management module 28 (or in another module of the audience measurement server 4): • reconstruction (at least partial) of digital images from the intermediate signals generated by the optical meter 22; and detecting, in the reconstructed digital images, at least one identification element of the broadcaster of the TV and / or video content consumed by the terminal. The identification element of the diffuser is for example the logo of the television channel, which is embedded in a corner of the image.

FIGS. 3 and 4 show a particular embodiment of the optical audimeter 22. The optical meter 22 comprises (at least) a transparent film 38, extended on at least one of its sides by a border 39. The transparent film 38 is placed against the screen of the terminal 1, in the manner of a privacy filter. This solution applies preferentially to terminals with small screens, such as smartphones, tablets or laptops. In the case of touch screen terminals, the optical meter 22 is made of a conductive material to preserve the touch function of the screen.

The transparent film 38 has the function of capturing the images displayed by the screen, in order to be able to analyze them. It is not necessary to capture images at the video frame rate of 50 frames per second. For example, a sampling of one frame per second may be sufficient depending on the types of measure concerned.

An example of a transparent film was developed in 2013 by the Austrian Johannes Kepler University of Linz; see for example the links: • http://www.infohightech.com/a-new-captor-dimensions-transparent-plat-and-flexible-for-cars-to-general-control/, and • https: //www.osapublishing.org/oe/search.cfm?a=0%20%20Bimber).

Unlike the photodiode-based sensors implanted on the entire surface of the film, the originality of the technique proposed by this Austrian university is to use a film functioning as a luminescent concentrator, which makes it possible to avoid opacifying the film with sensors and circuits. A wavelength of incident light is absorbed by the film and retransmitted inside the film in all directions at a lower frequency. At the periphery of the film, pierced triangles 33 make it possible to select certain angles of incidence of the light rays propagating inside the film. The selected rays are projected on photo-sensor arrays 32. Each photo-sensor integrates all the light transported by the light rays at a given angle. The signals picked up by the photo-sensors 32 located on the four sides of the film (also called "intermediate signals") are collected by a processor 35a cooperating with memories 35b (ROM and RAM type for example). From these intermediate signals, tomographic image reconstruction techniques (also used in medical imaging, such as the Radon transform) are used to reconstruct the images displayed on the film. Indeed, when a light beam strikes the film at a point, part of the rays is transmitted inside the film to the periphery sensors. By combining the information received by all the sensors, it is possible to determine the coordinates of the light spots because it is known that the travel time of a ray to the sensors is a non-linear function of the distance traveled.

In theory, reconstruction can be done in real time. The system is precalibrated by linear regression on the basis of thousands of images. More intensive mathematical operations can achieve a better resolution provided that they are performed in non-real time.

In the case where an image reconstruction is necessary, it is preferable to transfer the raw data (intermediate signals) to the audience measurement server (back-office) in order to benefit from a large computing power and not not affect the optical meter 22 (in terms in particular of CPU power, battery ...).

To transmit the intermediate signals to the audience measurement server 4, the optical meter 22 also comprises an antenna 36 and a communication circuit 310 comprising for example a Wifi circuit (to take advantage of a home or business network) and / or a 3G / 4G type modem coupled to a SIM card (allowing access to the cellular network of a mobile operator when the terminal is mobile) and / or a Bluetooth circuit (to connect the optical meter 22 to the terminal 1 in the manner of a connected object (eg smart watch) and thus take advantage of its Internet connectivity). If the optical meter does not have a TCP / IP stack to access the Internet, it is necessary to install a dedicated application on the terminal to retrieve the measured data and send it to the audience measurement servers. Optical meter 22 also includes a battery 37 which allows the power supply. To avoid the stress of recharging, it is also possible to use photovoltaic cells placed on the screen frame or even on the surface of the transparent film. This makes it possible to take advantage of the ambient light, or even the light emitted by the screen of the terminal itself. This solution is however not possible for smartphones that are most often placed in the dark at the bottom of a pocket or bag. The optical meter 22 is fixed to the screen by simply adhering the transparent film 38 or with small hooks 34. Depending on the size of the screen, different sizes of optical meter 22 are manufactured. In a variant, the optical meter 22 is integrated in a protective case for smartphone or tablet.

With a single transparent film, the reconstructed image is monochrome which is generally sufficient to recognize shapes and / or text. To avoid altering the panelist's rendering of the screen, the film absorbs a wavelength emitted by the screens, but not visible to the human eye.

Moreover, by superimposing several absorbing films for example Red, Green, Blue colors, it is possible to reconstruct colored images, which provides richer information, in particular to perform a recognition of logos. In return, the light passing through this stack is then more attenuated. A panelist can simply compensate for the attenuation by increasing the brightness of his screen.

The aforementioned technique for producing an optical meter 22 (use of at least one transparent film configured to operate as a luminescent concentrator and to cooperate with a network of photo-detectors (not shown) is not the only solution.

A system according to a second embodiment of the invention, for a measurement of TV and / or video audience is now presented in connection with FIG.

In this second embodiment, the image capture device is an optical meter 52 which differs from that of the first embodiment (referenced 22 in FIG. 2) in that it is produced with (at least) a transparent film 53 which comprises a network of photodetectors for example based on graphene, this network being configured to directly generate digital images. Graphene is a material made up of carbon atoms that are assembled in very thin layers, which gives it lightness and transparency. Different teams are working on the development of graphene photo-detectors, to capture natively two-dimensional images without having to rebuild; see for example the links: • http://www.graphene-info.com/graphene-sensors; and • http://www.gizmag.com/graphene-imaging-sensor/27718/.

Since in this first variant the optical meter 52 directly generates digital images, adaptations are necessary with respect to the first embodiment described above in relation to FIGS. 2 to 4: the optical meter 52 comprises a module fingerprint generation system 54 (realized in hardware and / or software form), configured to generate the first fingerprints from the digital images; On the side of the analysis means, the means for generating reference fingerprints comprise a calculation module 55 (made in hardware and / or software form) which implements the same calculation algorithm as the fingerprint generation module 54 included in the optical meter 52.

To remove the indeterminability mentioned above (in the case where the same content is broadcast by several broadcasters), the complementary processing (performed for example in the fingerprint management module (referenced 28 '), or in another module of the audience measurement server 4) is also modified (with respect to the first embodiment): • reception of the digital images generated by the optical meter 52 (compression of the digital images can be performed before transmission); and detecting, in the digital images received, at least one identification element of the broadcaster of the TV and / or video content consumed by the terminal.

Like the optical audimeter 22 of the first embodiment, the optical meter 52 of the second embodiment comprises for example an antenna, a communication circuit and a battery, and it is fixed on the screen by simple adhesion of the transparent film. or with small hooks.

With reference to FIG. 6, a system according to a third embodiment of the invention, for a measurement of TV and / or video audience, is now presented.

In this third embodiment, the image capture device is a camera or camera-type camera 61, external to the terminal 1 and configured to capture at least some of the images displayed by the terminal 1 and to generate digital images.

The shooting device 61 is for example: a mini-camera carried by the panelist, for example on the chest; • a mini-camera integrated into glasses worn by the panelist; or • a camera attached to the end of an arm placed above the screen of the terminal 1 of the television type or computer.

Like the optical audimeter 52 of FIG. 5 (second embodiment), the camera 61 comprises a fingerprint generation module, an antenna, a communication circuit and a battery.

The other elements of the system of Figure 6, and the operation thereof, are identical to those of Figure 5.

A system according to a fourth embodiment of the invention, for a measurement of TV and / or video audience, is now presented in connection with FIG.

In this fourth embodiment, the image capture device is a screen capture module 72, included in the terminal 1. It is made hardware and / or software, and configured to capture (typically every second ) at least some of the images displayed by the terminal 1 and generate digital images.

The screen capture module 72 uses the connectivity of the terminal 1 to transfer the captured images (possibly after their compression) to the audience measurement server 4 (in the back office).

The other elements of the system of Figure 7, and the operation thereof, are identical to those of Figure 5.

In connection with FIG. 8, a system according to a fifth embodiment of the invention is presented for an Internet audience measurement.

In the case of Internet audience measurement, it is useful to have two-dimensional (2D) digital images in order to analyze them and thus derive relevant information for Internet audience measurement.

The relevant information is for example: • for the consumption of video content by the terminal 1: the support site or the support application, the multimedia embedded player of a broadcaster ("brand player") and the video content (" brand content "). Example: it is determined that there is a reading, with the video player of the Youtube broadcaster, of a video clip of a program of the channel TL1, diffused on the site Lacebook; and for the consumption of textual and / or image content by the terminal 1: the support site or the support application and the content ("brand content"). Example: it is determined that there is a display of a page of the newspaper Le Monde via the application Lacebook.

In this fifth embodiment, the image capture device is the optical meter 22 already described in relation to FIGS. 2 to 4. It generates intermediate signals and transmits them to the audience measurement server 4 so that last at least partially reconstruct digital images. This thus benefits from the high computing power of the audience measurement server 4 (or other back-office equipment) and does not affect the computing resources of the optical meter 22 (or the terminal 1). In a variant, the optical meter 22 has sufficient computing resources and performs the image reconstruction itself and then transmits the reconstructed images to the audience measurement server 4.

Moreover, the analysis means comprise: a database 82 containing elements to be recognized (logos, forms, images, text extracts, etc.); An image analysis platform 81; and within the audience measurement server 4, a processing module 83 itself comprising: a module for at least partial reconstruction of digital images from the intermediate signals generated by the optical meter 22 (this module reconstruction is not necessary in the case of the aforementioned variant where the optical meter 22 itself performs the reconstruction of images); and o an interrogation module of the image analysis platform 81, so that the latter detects, in the reconstructed digital images, at least one of the elements stored in the database 82. Each element actually detected is a information relating to the consumption of content by the terminal 1 on the Internet.

Depending on the resolution of captured images, different levels of analysis are possible. With a good resolution, it is possible to extract text, for example the URL located in the address bar of a browser. With a lower resolution, it is possible to recognize brand logos, for example recognize the page of the newspaper Le Monde in the Facebook page facebook .com / lemonde .fr

The image analysis platform 81 is for example the Cloud Vision API platform, proposed by the Google company, which makes it possible to analyze images to recognize logos, shapes, images, and extract text. Other technologies can be used, for example that of the company JASTEC France (formerly LTU) (see the link: http://www.ltutech.com/fr/).

For an Internet audience measurement, variants of the system of FIG. 8 are possible, for example by replacing the optical meter 22 by an image capture device capable of directly generating digital images, in particular: • the audimeter optic 52 of Figure 5; The shooting device 61 of FIG. 6; or • the screen capture module 72 of FIG. 7. In a particular implementation, the screen capture module 72 is a software module integrated into a traffic measurement application (meter) installed on the terminal to analyze the traffic with a proxy. The traffic measurement application is, for example, configured to activate the screen capture module 72 in the event of launching the execution by the terminal 1 of a client application whose traffic can not be measured by the application 71 (for example the client application "Lacebook" which makes the "Certificate Pinning"). Various implementations of the association means 6 (as well as of the corresponding step S3) are now described, enabling the audience measurement server 4 to associate the information relating to the consumption of content by the terminal (that is, that is to say the information determined by the analysis means 5, from the elements generated by the capture means 2, namely digital images and / or intermediate signals and / or digital prints), with a single panelist, if the terminal is of the non-shared type, or a list of panelist (s) declared, if the terminal is of the shared type.

To achieve such an association, the association means 6 performs another association, between the elements generated by the capture means 2 and the single panelist (if the terminal is of the non-shared type) or the list of panelist (s) declared (if the terminal is of the shared type). It is considered that if elements generated by the capture means 2 are associated with one or more panelists, then the information determined by the analysis means 5 from these elements are associated with this (s) panelist (s).

To describe this "other association" in detail, two situations are distinguished below, depending on whether the elements generated by the capture means 2 pass through the terminal 1 or not.

First situation: the elements generated by the capture means 2 pass through the terminal 1. This first situation corresponds to the case where the capture means 2 are integrated in the terminal or use the communication means of the terminal to transmit the generated elements to the terminal. audience measurement server 4.

In the case of a non-shared terminal, the terminal sends to the audience measurement server, in an initialization phase, an identifier of the terminal and an identifier of the panelist. The audience measurement server records the terminal as a panellist terminal and stores an association between the terminal identifier and the panelist's identifier. In a use phase, the terminal sends the audience measurement server messages containing elements generated by the capture means 2 and the identifier of the terminal. Thanks to the association previously stored, the audience measurement server knows which panel member to associate the generated elements received (and therefore also the information determined by the analysis means 5 from these elements).

In the case of a terminal shared by several members of the household (for example a computer or a tablet), the terminal sends to the audience measurement server, in an initialization phase, an identifier of the terminal, with a couple login / password of the home. The audience measurement server records the terminal as a panellist terminal and stores an association between the terminal identifier and the home identifier. It sends the home identifier to the terminal. In a use phase, a declaration application (executed by the terminal) displays a window with the list of the members of the household each time they start, log in, or wake up. For this purpose, the declaration application pre-queries the audience measurement server with the home identifier and receives in return the list of household members. Each person present indicates his presence in front of the terminal by checking for example a box corresponding to his first name (or a "guest" box for a person not listed as a member of the household). The terminal sends the audience measurement server a declaration message containing a list of panelist (s) declared via the declaration application, as well as the identifier of the terminal. The terminal also sends the audience measurement server messages containing elements generated by the capture means 2 and the identifier of the terminal. Thanks to the association previously stored, the audience measurement server knows, for each message containing elements generated by the capture means 2, which list of panelist (s) declared (s) is (or was) the current list at the time of receipt of this message, and therefore to which (s) panelist (s) to associate the generated elements received (and therefore also the information determined by the analysis means 5 from these elements).

Second situation: the elements generated by the capture means 2 do not transit through the terminal 1. This second situation corresponds to the case where the capture means 2 (hereinafter called "capture device") are not integrated into the terminal and use their own means of communication to transmit the generated elements to the audience measurement server 4. It is then necessary to reconcile the declaration made on the terminal and the generated elements transmitted by the capture means 2.

In the case of a non-shared terminal, a solution is for example to place a unique identifier of the capture device in the memory thereof, during its manufacture. This identifier is for example registered on the capture device and the panelist is invited to copy it at the first launch of the declaration application on the terminal, in order to pair the terminal and its associated capture device. In other words, in the initialization phase, the terminal sends the audience measurement server the identifier of the terminal, the panelist's identifier and the identifier of the capture device. The audience measurement server registers the terminal as a panellist terminal and stores an association between the terminal identifier, the panelist's identifier and the identifier of the capture device. In the use phase, the capture device transmits its identifier to the audience measurement server, with the elements generated by the capture device. Thanks to the previously stored association, the audience measurement server knows which panel member to associate the elements generated by the capture device (and therefore also the information determined by the analysis means 5 from these elements).

In the case of a shared terminal, a unique identifier of the capture device is also used. In an initialization phase, the terminal sends to the audience measurement server the identifier of the terminal and the identifier of the capture device (provided by one of the panelists), with a login / password pair of the home. The audience measurement server records the terminal as a panellist terminal and stores an association between the terminal identifier, the identifier of the capture device and the home identifier. It sends the home identifier to the terminal. In a use phase, a declaration application (executed by the terminal) displays a window with the list of the members of the household each time they start, log in, or wake up. For this purpose, the declaration application pre-queries the audience measurement server with the home identifier and receives in return the list of household members. Each person present indicates his presence in front of the terminal by checking for example a box corresponding to his first name (or a "guest" box for a person not listed as a member of the household). The terminal sends the audience measurement server a declaration message containing a list of panelist (s) declared via the declaration application, as well as the identifier of the terminal. For its part, the capture device also sends the audience measurement server messages containing elements generated by the capture device and the identifier of the capture device. Thanks to the association previously stored, the audience measurement server knows, for each message containing elements generated by the capture device, which list of panelist (s) declared (s) is (or was) the list current at the time of receipt of this message, and therefore to which (s) panelist (s) associate the elements generated by the capture device (and therefore also the information determined by the analysis means 5 from these elements).

Claims (10)

A user-centric audience measurement system comprising at least one audience measurement server (4) configured to measure content consumption by a terminal (1) associated with at least one panelist, characterized in that it comprises: image capture means (2), included in or placed on or near the terminal, and configured to generate, from images displayed by the terminal, digital images and / or intermediate signals, at least partially reconstructing digital images, and / or digital fingerprints; analysis means (5), at least partially included in or cooperating with said at least one audience measurement server, and configured to determine at least one information relating to said consumption of content by the terminal, according to the images digital and / or intermediate signals and / or digital prints generated by the image capturing means; and association means (6), configured to associate said at least one information relating to said content consumption by the terminal, with a single panelist, if the terminal is of the non-shared type, or a list of panelist (s) declared (s), if the terminal is of the shared type, and in that the image capturing means comprise at least one transparent film (38, 53), configured to be placed against a screen of the terminal, driver if the terminal screen is tactile, and comprising or cooperating with a network of photodetectors (32) generating digital images and / or intermediate signals. 2. System according to claim 1, characterized in that said at least one transparent film (38) is configured to function as a luminescent concentrator and to cooperate with said photodetector array, the photodetectors being positioned on at least one part of the outer edges of said at least one film, said photodetector array being configured to generate intermediate signals for at least partially reconstructing digital images. 3. System according to claim 1, characterized in that said at least one transparent film (53) comprises said photodetector array which is configured to generate digital images. 4. System according to any one of claims 1 to 3, characterized in that said consumption of content by the terminal is a consumption of TV and / or video content, in that the image capture means comprise means ( 22, 54) for generating first impressions from the digital images and / or intermediate signals, and in that the analyzing means comprises: means (25, 26, 25 ', 26', 55) for calculating and storage of second fingerprints, reference, from known TV and / or video content; and means (28, 28 ') for identifying the TV and / or video contents consumed by the terminal among the known TV and / or video contents, as a function of a comparison between the first and second imprints. 5. System according to claim 4, characterized in that the intermediate signals are assimilated to first impressions extracted from the TV and / or video content consumed by the terminal. 6. System according to any one of claims 4 and 5, characterized in that the image capturing means or the analysis means comprise means (28) for at least partial reconstruction of digital images from the signals. intermediates generated by the image-capturing means, and in that the analysis means comprise means (28) for detecting, in the reconstructed digital images, at least one identification element of a content diffuser. TV and / or video consumed by the terminal. 7. System according to any one of claims 4 and 5, characterized in that the analysis means comprise means (28 ') for detecting, in the digital images generated by the image-capturing means, from minus an identification element of a broadcaster of the TV and / or video content consumed by the terminal. 8. System according to any one of claims 1 to 3, characterized in that said consumption of content by the terminal is a consumption of content on a computer network, in that the image capture means or the means of analysis comprise means (83) for at least partial reconstruction of digital images from the intermediate signals generated by the image-capturing means, and in that the analysis means comprise means (81, 82, 83) detection, in the reconstructed digital images, of at least one information relating to the consumption of content on the computer network. 9. System according to any one of claims 1 to 3, characterized in that said consumption of content by the terminal is a consumption of content on a computer network, and in that the analysis means comprise means (81, 82, 83) for detecting, in the digital images generated by the image capturing means, at least one information relating to the consumption of content on the computer network. 10. System according to any one of claims 8 and 9, characterized in that said at least one information relating to the consumption of content on the computer network belongs to the group comprising: information relating to a support site on which is consumed a content; information relating to a support application on which content is consumed; information relating to an on-board multimedia player of a broadcaster, with which content is consumed; and information relating to a content.