Key Benefits:
Rectificatif au Journal officiel du 21 août 2007, édition électronique, texte n° 4, et édition papier, page 13889, après la signature, insérer les annexes suivantes :
A N N E X E T E C H N I Q U E 1
The Order sets out technical standards that relate, on the one hand, to cameras and transmission and storage systems (articles 1 and 2), on the other hand, to the interoperability of data storage and export systems to the police and gendarmerie forces (Article 3).
In order to facilitate the use of this circular, the requirements of the order which it comments are repeated in italics.
1. Cameras
"The cameras are set, equipped and connected to the visualization system and, where applicable, to the storage system, so that the images returned during the real-time or deferred-time visualization allow us to respond to the purposes for which the video surveillance system has been authorized. »
1.1. The technical features of the video surveillance system must be used to achieve the security objectives that presided over its installation.
The first implication is that the objectives of the system and each camera, in terms of security, must be clearly stated. These objectives relate to the global system (accepted unavailability rates, storage system characteristics...) and the cameras themselves, whose roles must be defined. For example, such a group of cameras may have as the main role of allowing the lifting of doubt before a door opening, while another group of cameras will have as the main objective of allowing the analysis of the image in real time as, for example, the recognition of individuals who have accessed a given area.
1.2. The quality of the images.
The second fundamental implication is that the technical characteristics of the system must be consistent with the stated objectives. This point is essential because, if the diversity of situations prohibits the absolute definition of what a system's technical characteristics must be to achieve a certain result, it is always possible to check the coherence of a system with the objectives assigned to it. The order specifies that this verification should not be done exclusively on the various elements of the system (quality of the cameras, quality of the data links, quality of the compression of the images...) but on the quality of the images returned.
This coherence requires the operator to adapt the poor or poorly dimensioned elements of the system when the quality of the images returned is incompatible with the objectives of the system.
Control of this consistency during the review of the pre-installation authorization application, so "on file", may be difficult. However, a technical annex provides some benchmarks whose prefecture services may be inspired during the review of the files.
However, attention should be drawn to the fact that this first limitation has a certain interest in the assumption of a posteriori system control, at the request for renewal of the authorization for example.
"The cameras have the technical characteristics adapted to the illumination conditions of the video surveillance site".
It is simply to check that the operator has taken into account the specificities related to the illumination of video scenes when choosing the cameras. Indeed, if it is to be able to record quality images in night vision, then either use high-sensitivity cameras or provide extra lighting, infrared for example. These elements must go hand in hand with the special lighting conditions of the filmed scenes, which must also be specified (in intense lighting can, in particular, allow the use of a less sensitive camera).
2. The transmission of images
"The networks on which video streams are transiting offer bandwidth compatible with the flow rates necessary for the transmission of images of sufficient quality to meet the purposes for which the video surveillance system has been authorized. »
Camera images, to be transmitted to the networks, must be coded and generally compressed to be communicated in real time through transmission arteries. The maximum flow of these transmission channels, called bandwidth, thus directly condition the quality of the received images. Thus, an insufficient bandwidth will automatically result in a loss of quality (compression of images that are too important inducing a detrimental loss of information) or performance (decrease in the number of images per second or choice not to transmit all streams).
The diversity of use cases (e.g. fixed image or with a lot of movement) and technical devices (e.g. MPEG 2, MPEG 4, JPEG, JPEG 2000...) does not allow to define in advance the minimum bandwidth required for the digital transmission of a quality image, this quality also depends on the fixed safety objective. It is recalled for memory that the average weight of an image of excellent quality is 45 Ko. However, the table below provides an overview of the size order of the bandwidth used to transmit images with certain characteristics for the different data compression classes.
You can see the table in the OJ
n° 196 of 25/08/2007 text number 7
Thus, if an operator declares to forward 8 simultaneous streams of images, in 4 CIF format to 12 images per second tablets in MPEG 2 format (2 Mbits x 8 = 16 Mbits of theoretical flow required for transmission of these streams), on a system with a 4Mbit bandwidth, it is necessary to question the relevance of this choice.
"The networks on which video streams transit take into account the security of video streams, ensuring their availability, confidentiality and integrity. »
Data from video surveillance systems must have three types of essential features:
- they must conform to the original images. The latter should therefore not have been corrupted or modified during their transfer. The transmission system must provide a guarantee for the integrity of the reported data;
- they must be accessible in case of solicitation. To do this, the transmission system must first be robust to malfunctions and possible external attacks. It must provide a guarantee of the availability of the data provided;
- they must be accessible only to persons authorized to dispose of them. This implies that specific devices must be implemented to prevent the interception and reading of transmitted data. The transmission system must therefore offer a guarantee of data privacy exchanged, most often through appropriate encryption functions.
It is not a matter of engaging in comprehensive security expertise that ensures these three criteria, or even soliciting the operator for a formal safety certificate. However, it is important to ensure that these criteria have been taken into account and that the solutions implemented address these three topics.
The case of wireless digital transmissions (e.g. Wi-Fi or Wi-Max) certainly deserve particular attention. Indeed, the interception of the flow is by easy nature and, to a lesser extent, the "denial of service". It is therefore appropriate that, on the one hand, the operator guarantees the confidentiality of the data by using a suitable and reliable encryption and that, on the other hand, it limits the use of these technologies to terminal or unfit network segments to wired technologies.
3. Storage
"The storage of video streams is done on a digital basis for video surveillance systems with eight or more cameras. This storage can also be done on another type of support. Storage of video streams is performed on analog or digital support for video surveillance systems with less than eight cameras. »
When a video surveillance installation becomes important, it is not conceivable, in a service quality objective, to use analog-type storage. Digital storage is therefore imperative. It should be noted that this constraint is limited to the image recording module, which in particular implies that nothing prohibits the use of analog cameras whose streams will be digitized thereafter. However, it is specified that storage can also be performed on another type of support to allow operators to maintain their analog recording system (VHS cassette type), in addition to the digital recording system they will be required to set up.
To limit the cost of installing small video surveillance systems, it is possible to use analog storage media providing greater ease of installation and use. The systems covered here, which have seven or less cameras, must be understood as those designed to secure an autonomous geographical entity. For example, a company that files a request for authorization for video surveillance systems in each of its agencies (independent and autonomous circumstances) can design these systems as autonomous for each of them. In this case, all seven or less camera agencies are allowed to maintain an analog storage system. However, it should be noted that these systems are considered autonomous only if storage and/or visualization is carried out in each agency. If agency videos are repatriated to one or more common sites, then the systems of each agency can no longer be considered independent.
"Every digitally recorded video stream is stored with information to determine at any time of the video sequence its date, time and location of the camera. »
In order to be able to use video images stored in court proceedings, it is necessary to be able to certify spatial and temporal information associated with images. Section 2, second paragraph, of the September 26, 2006 Order explicitly refers only to the ability of the recording system to associate these three data with the images. Its spirit, however, is to allow an investigator to effectively use the digital data transmitted, which has a dual involvement:
- the date and location parameters must be accessible to the investigator with the visualization system available to him;
- parameters must be accurate.
It is therefore necessary to ensure, to the extent possible, that these two constraints have been taken into account.
There is a simple method to mark this information directly on the video image. However, this method has the disadvantage of masking parts of the image. Another method is to associate the information with the video stream and then create a software link between the images and the associated information file. In this particular case, readers provided to investigative services will need a specific capacity to re-associate data and images during their operation.
The operator of the video surveillance system will also need to clarify how it ensures the reliability of the time repository that will be associated with the images.
"For analog video stream recording systems, a device can determine at any time the date, time and location of the camera corresponding to the recorded images. »
The need for police forces is identical regardless of the nature of the system, only the storage mechanism of information associated with the images will be different. In the case of analog recording (of the VHS type), the information must exist but their format (paper or digital file) is not specified. However, it is necessary to ensure that investigators will be able to have this information when analyzing the images. The data associated with analog supports must therefore be available to them with video cassettes.
"The digital recording ensures the integrity of video streams and associated data related to the date, time and location of the camera. "
The means to ensure the integrity of video streams and associated data relating to the date, time and location of the camera are not specified. In particular, it is not required here that systems incorporate electronic image marking devices (sometimes called watermarking or watermarking), even if these devices are welcome and should be encouraged. Indeed, a robust digital system for recorded traces (any intervention that changes the data is unmistakably recorded) and the operating environment (which guarantees, in particular, the integrity of the trace system) is likely to achieve the integrity objectives.
"The stored video streams from the cameras that, given their positioning and orientation, operate mainly in a narrow plane, excluding those of traffic regulation, have an image format greater than or equal to 704 x 576 pixels. This format can be lower if the system allows the extraction of face thumbnails with a minimum resolution of 90 x 60 pixels. »
The purpose of section 2, fifth paragraph, is to promote the existence of images of satisfactory accuracy for the work of the investigators. It therefore sets the principle of a minimum level of quality of the images stored when they are from cameras operating in a narrow plane.
The balance sought here is to guarantee a good quality of the images only when it is necessary for the police and gendarmerie forces, without imposing too much technical constraints on the parts of the device that are less directly concerned with the investigative work.
To do this, we distinguish two large types of video surveillance cameras, the ones whose main function is to analyze the information about the individuals or objects present in the camera field (which are said to be working in close plan) and those whose main function is to provide a global view of the situation (which are said to be operating in a broad plan).
This classification calls two remarks and deserves to be illustrated by some examples.
First of all, it is understood that cameras that make up a video surveillance device usually have multiple missions. This is all the more true given that some cameras have zoom and fast orientation functions that allow them to offer a global plan and move the next instant in close-up. However, it remains that at each camera is most often assigned a main operating objective: raising doubt, managing a queue, monitoring a sensitive target, controlling the flow...
It is necessary that these main objectives be specified for each camera in the files transmitted by the operators. In most cases, they must be able to decide on the classification of large or narrow cameras.
Then, it is legitimate to question the possible correlation between the technical characteristics in terms of camera focal or zoom and their use in a broad or narrow plan (as defined above). Given the diversity of uses of video surveillance, this link does not seem to be relevant. Indeed, a camera designed to guarantee the safety of an automatic ticket distributor or to secure the inbounds in a bus can, due to the low distance to the target, function with an important angular aperture, while in the sense of the 26 September 2006 stoppage it is well, given the expected accuracy of the image, a close operation. In the same way, some cameras designed to secure railway tracks can operate with a small angular aperture but in a wide plan within the meaning of the stop, if intended for the regulation of railway traffic.
The resolution of 704 x 576 corresponds to the so-called 4 CIF format, standardized in the video field, compatible with the performance of the majority of installed cameras and constituting the high standard for image definition pending the generalization of the so-called high definition cameras. The definition referred to in this article concerns images stored on the recording system. This implies that the entire video chain must display features compatible with these recording formats: the resolution of the sensors (technical features of the cameras), the camera output image format, the compression ratio of the images during the transfer and storage. Another consequence is that storage spaces must be compatible with the overall features of the system. It is therefore important that the technical specifications (definition, compression ratio, number of images per second, data retention period, number of stored fluxes) of the system be specified as well as the calculation leading to the size of storage spaces.
In some cases, it is not necessary to have an image of 704 x 576 pixels to offer satisfactory resolution of filmed subjects. Operators therefore have all latitude to retain a lower format if it proposes, in the nominal shooting zone, a resolution allowing the identification of a face. In particular, digital cameras in VGA format (640 x 480 pixels) that would allow the extraction on recorded videos of face thumbnails of 90 x 60 pixels suitable.
It is certain that the diversity of situations will inevitably lead to litigious or ambiguous cases for which the bidder's proposal for a broad/short plan classification may appear questionable. In order to determine in practice the minimum characteristics of the stored images, the table of examples proposed in annex I must most often allow to assimilate these situations to an approaching case of use already treated.
"Other stored video streams have an image format greater than or equal to 352 x 288 pixels".
All other video streams from video surveillance systems covered by the law of January 21, 1995, as amended by the law of January 23, 2006, must be stored with a resolution of 352 x 288 pixels, also called CIF format. This is especially the case of images from a traffic control device.
"A minimum frequency of twelve images per second is required for the recording of video streams from cameras installed for one of the purposes mentioned in II of Article 10 of the Law of January 21, 1995 referred to above, excluding those of traffic regulation, and which, in view of their positioning and orientation, operate mainly in narrow plan and filtering mainly of the flow of individuals on rapid displacement. »
The number of images per second is also an important parameter when looking for precise elements in a moving video scene. However, the requirements should be modulated according to actual operational requirements to avoid oversizing the video surveillance system unnecessaryly. This is why the requirement to have 12 images recorded per second applies only to cameras operating mainly in a narrow plane (see article 2, paragraph 5), and only to cameras intended to monitor the flow of people in "fast displacement".
This concept explicitly refers to situations where filmed individuals are, except in particular circumstances, walking without encountering obstacles when they cross the shooting area. In particular, there is a question of rapid displacement for cameras intended to film a transit space in public places (subway corridor, airport hall, urban sidewalk...). On the other hand, images of individuals are not considered to be fast travel by crossing a door or metro tourniquet, or parking in a hall intended for waiting or collecting baggage.
Cases of the most typical or problematic figure are referred to in annex 2.
"For recording other video streams, a minimum frequency of six images per second is required. »
All other images covered by the law of January 21, 1995 must at least be recorded at a real rate of 6 images per second from a camera whose frequency of acquisition of images will be at least 6 images per second. Thus, it would not be a question of artificially reconstructing a flow to 6 images per second from an initial sequence to 3 images per second. The same applies to a 12-picture recording per second.
"The storage system used is associated with a newspaper that keeps track of all the actions carried out on video streams. »
The traceability of the actions carried out on the system is essential to verify that no abuse and that no malicious action has been committed. In the case of analog recording systems or digital video surveillance systems of less than eight cameras, a newspaper that keeps track of all the actions carried out on the streams (export, modification, suppression...) can be held by hand.
"For digital systems, this journal is generated automatically in electronic form. »
To simplify the logging operation, which can be tedious for large systems, it must be automated for digital systems. It is therefore necessary to ensure that the proposed system integrates this function and that the operator anticipates in its operating plan to implement it.
4. Interoperability constraints
The purpose of the Order of September 26, 2006 is to ensure that video surveillance techniques can effectively implement the provisions that the amended Act of January 21, 1995 enacted.
The provisions of Article 3 of the above-mentioned Order are intended to facilitate the operation of the systems by the police and gendarmerie.
"Video streams are exported without quality degradation. »
The transmission of video films to the police and gendarmerie forces requires an "export" operation. It is necessary that the quality of the images exported is maximum, which implies that the system must be able to export its data without loss of quality.
If, during the export operation, it is necessary to modify the format or type of compression of video streams, then it will be necessary to ensure that the compression of exported videos does not degrade their quality.
It is therefore important to know the method of exporting video streams and, in case it is not a simple copy of the data, the characteristics of the compression used for storage and export.
"For video surveillance systems using analog technology, a device determines the list of exported streams, indicating the date and time of filmed images, their duration, identifying the cameras concerned, the date and time of export, the identity of the person who made the export. »
It is important to maintain traceability of exports to ensure that no abuse is committed. The difficulty of this measure for an analog video surveillance system, and to a lesser extent for digital systems of less than eight cameras, is sometimes the lack of system automation. It is then necessary to integrate into the video flow export procedure the manual constitution of a newspaper of the various operations carried out on the system. This paper's constitution must in particular allow for the identification of the person(s) who exported the video streams.
"For video surveillance systems using digital technology, an electronic export log, including the information cited in the previous paragraph, is generated automatically. »
As with analog systems, export traceability is essential for digital systems. The advantage of a digital system is the ability to automate actions. Thus, to ensure the accuracy of the information contained in the list of exported flows, simply create an electronic "log" consisting automatically by the system.
"The registration system remains in operation during these export operations. »
The export of data must in no way diminish the capabilities of a video surveillance system. Indeed, it would be highly damaging that, in the export of video images, a serious event occurs and that it is impossible to record the related video streams. The fact that the registration system remains in operation during export operations in particular aims to prohibit the extraction of the storage units of the system during the investigative phases if this action prohibits the continued normal operation of the system. It is therefore important to verify that the export procedure complies with this requirement. A simple method is to provide additional storage media to replace those temporarily extracted from the system.
"The physical export support is a non-reinscriptible digital support and direct access, compatible with the volume of data to be exported. In the case of large volumes of data to be exported, hard disks using a standard connector may be used. For digital video surveillance systems, software for the operation of images is provided on a digital basis, disjointed with data support. »
The video recording storage system must be equipped with the ability to export movies and photos to a non-reinscriptible medium, which in the current state will most often be CD or DVD burner. All systems must therefore have this feature. This implies that USB drives (which constitute a reinscriptible support) cannot be the only export support on such a system.
The support must also be at direct access, i.e. that the information must be accessible without having to browse the entire support sequentially. In particular, DAT cassettes cannot be a valid export support.
However, it is sometimes necessary to export a large amount of data. In this case exclusively, it is authorized to use hard disks, which allow for greater storage capacity. This option is added to the export capacity on non-reinscriptible media, which in all cases constitute the default means of transmitting data to the security forces.
"The software allows:
« 1° Playing video streams without degradation of image quality;
« 2° Playing video streams in accelerated, backward, idle;
« 3° Image-by-image playback of video streams, stop on an image, save an image and sequence, in a standard format without loss of information;
« 4° The display on the screen of the camera ID, the date and time of the recording;
« 5° Camera search, date and time. »
Video streams are exported to be processed by police or gendarmerie. The characteristics mentioned must therefore be integrated into the reading software, provided on a separate digital medium distinct from that of the images, by the operator to the investigator services at the same time as the images.
A N N E X E T E C H N I Q U E 2
Specific examples, illustrating the notions of "small-planned operation" and "fast-moving flow of individuals":
You can see the table in the OJ
n° 196 of 25/08/2007 text number 7
These examples cover a large number of cases of camera implantations. This table is intended to allow departmental commissions to decide more easily on the broad, narrow plan classification.
However, it is of course that there may be some specific cases where this table is not applicable:
- cameras whose purpose is to facilitate flow control are cameras that usually operate in a broad plane. However, as soon as it is specified that this control must allow to know which people are on the videos, these cameras will be considered to be operating in a narrow plane;
- Likewise, a camera monitoring a parking entrance whose purpose is to control which person and/or vehicle accesses the parking lot will have to work in 4 CIFs and in 6 images per second or 12 images per second (flux on rapid displacement) depending on the regularized entrance of the vehicles and people;
- any camera whose purpose is to analyze information on individuals or objects in the scene should be considered to be operating in a narrow plane, regardless of its location and location;
- any camera whose purpose is to analyze information about individuals or objects in rapid displacement present in the scene will have to work in 12 images per second (persons on treadmill, entry into a store without filtering device...).
A N N E X E T E C H N I Q U E 3
Glossary
Definition of some technical terms
frequently used in video surveillance
Direct access (stocking to): This concept refers to the ability of a storage system to be able to access directly to recorded information without browsing the recording. The most common direct access storage system is the hard drive. These systems are to oppose sequential access storage systems.
Sequential access (stocking to): storage where reading and recording are performed according to a predefined order. For example, VHS, K7, DV, DAT, where, to access the third minute of the recording, it is necessary to browse the first three minutes, are sequential storage systems.
Bandwidth (network): in the field of computing, the term bandwidth refers to a flow of information, specifically the amount of information that a network can transmit (computer system). This bandwidth is usually measured in bytes per second or in bits per second.
VHS cassettes: analogue recording support with sequential access using the VHS standard.
Field (optical): in optical, the field concept refers to the portion of visible space through the camera's objective.
Compression: reducing the space required for storage and data transmission (videos, images...). This compression can be done with or without loss of information on these data.
DAT: Digital Audio Tape is a digital audio recording medium. This support is also used today to store videos, audio or computer data. This type of storage is sequential access.
Denial of service: computer security, "denial of service attack" is an attempt to make an application, system or computer resource unavailable to its authorized users. If a computer system (server for example) is no longer able to process customer requests for reasons voluntarily caused by a third party, there is "denial of service". The most common type of attack is to make a server inoperative by addressing too many requests. The consequences of such an act can be translated in the case of a network of video surveillance system by:
- an unusually slowed network (difficulty to continuously communicate with a camera for example);
- impossibility of accessing a particular camera;
- impossible to access any camera;
- increase in the number of messages received via the network (mail, control message, error message...).
Hard disk: storage system with direct access and non-volatile memory based on the magnetic memory principle. Developed initially for computer use, it gradually replaced all other video and audio storage systems by the rapid evolution of its storage capacity and ease of access to saved data.
Export (data): operation of copying or extracting the targeted information storage system.
Flux: computer, basic data set from a computer system.
Focale (distance): the focal length of an optical system is one of the dimensions that fully defines an optical system. It can be assimilated in most cases at the distance between the lens and the camera sensor.
CIF format (4 CIF): Common Intermediate Format. The CIF format is a digital image format of 352 x 288 pixels. The 4 CIF format mentioned in this circular is the standard 704 x 576 pixels image format.
Image format: image size defined in terms of pixels or lines and columns.
Software link: connection provided by computer software automatically between several data or operators.
Angular opening (optical): this size represents the space portion in terms of visible angle through the camera's objective.
Pixel (Picture Element): basic structure of a digital image. This is the smallest discernable point on an image. The pixel can be any geometric form, even if the square is its most common structure. Each pixel contains color information (color image) or gray level (black and white)
Resolution: cf. image format.
Storage (analog/digital): storage, data backup (in this video case) on an analog (VHS cassette...) or digital (hard disk, DVD...).
Digital system: the notion of digital system, in the context of the order, applies exclusively to storage modules. Thus a system composed of analog cameras, but with a digital storage module, will be considered a digital video surveillance system.
Analogue system: the concept of analog system, in the context of the order, applies exclusively to storage modules. On a video surveillance installation, if the storage module is analog, then the video surveillance system will be considered analog. This category will be extended to video surveillance systems of less than eight cameras with digital storage modules, but whose features are limited to those of an analog storage module.
Visage: the distances between the bottom of the chin and the top of the hair or the skull will be heard by dimensions of the face and between the two ears. Depending on the requirements in the stop, the dimensions of a face on a camera of less than 4 CIF, operating in a narrow plane, must therefore be at least 60 pixels for the distance between the two ears and 90 pixels for the distance between the bottom of the chin and the top of the hair or the skull.
Watermarking/filigranage: technique to add information to secure an image, video or any other type of digital document, integrating it into the file without modifying or deteriorating it
Wi-Fi: wireless computer network technology that operates on a short distance (from about ten to a hundred metres in standard operating conditions).
Wi-Max: Standard family for wireless computer networks using microwave technologies.
Zoom (optical): objective on which the focal length is continuously editable. A camera equipped with a zoom allows to restrict or increase the field (optical) visible on the recorded video without changing the video resolution.
