The present invention relates to a method for improving the reliability and the stream reception quality video to a digital wireless network among others, governed by communication protocols for example of WiFi. Flows involved are sent in multicast, or multicast, that is to say that they are derived from at least one transmitter to the broadcasting receiver of a subscriber group to said diffusion. The available-sitifs receivers or clients, forming the target group are able to view the sent stream, and are provided for this purpose at least one application for processing and display 1 information received.

More particularly, the present invention relates to a trai-knitted performed at the receiving video streams, allowing soul improve the quality of decoded signals in the presence of losses of data blocks.

One of the possible applications of the invention is the dissem-sion in near real time of sporting events like meetings or shows, within the perimeter of the enclosure where appropriate the said event. The client application software allows one hand the fluid display of video images proposed among others by a service provider linked to the organizer of the event, and also to enhance viewing by treatments that could add value the received video stream, for example a repeat and therefore Reviewing certain sequences, if any idle.

Such treatments involve the video streams, classic-ing transmitted as packets of information containing es sentially of to groups of images, be handed over in sufficient quality first for their viewing or sim-plement comfortable and also for any subsequent trai-apparel signals can be based on sufficient information base for their realization.

Now on Wireless LANs, whatever: the communication protocols used and the method of dissemination - it s' therefore ap-plique also to WiFi networks in multicast distribution that will be used in the text as prime example - the rate transmission error is between 0% and 15%, sometimes more. This error rate is in practice in a loss of information blocks transmitted by the network. In the event of trans-mission video stream, the result of these losses is a poor quality of the resulting video, images that can present exogenous macro blocks that degrade and therefore degrade the quality of the stream when they are not purely and simply lost. As appropriate, these losses can alter the video stream until

In practice, video data streams are encoded as G packets that are actually mainly consist of groups of pictures of different categories, these packets being en-suite divided into blocks of data. According to a classic pattern that results elsewhere in the implementation in advance of lost data recovery conditions, these G packages contain cha cun K first data blocks which essentially encode the images of the video stream and N said blocks redundancy used where appropriate for forward error correction (FEC by an algorithm).

A stream therefore comprises a series of group of pictures or G packets each having K + N blocks transmitted by the transmitting device, which sends multicast to visuali-sation client devices, such as smartphones or tablets.

To ensure reliable reception and quality of the video stream despite still possible losses, the method includes the invention and principally the following steps implemented by each client device for each data packet received or G emitting devices:

a / control the correct reception of the first K blocks of data, decoding and display of the flow vi- deo in case of positive result;

b / in the absence of reception of all K pre- miers data blocks, and if the total number X of K + N blocks received is at least equal to K, decoding by means of a correction algorithm Forward Error (FEC) to recover missing data for K blocks, decoding and display of the video stream in case of positive result tif;

c / if the total number X of K + N blocks received is less than K or absence of a positive result to the decoding using the forward error correction algorithm (CEF), en- voi by the client device message unicast Don-ing to the information transmitter or at least on the lost blocks;

d / if a statistical algorithm emitting device organizes the return of missing data, control of the correct receipt of data returned by the transmitting device according to a / b /, decoding and illumi- nating the video stream in case of a positive result .

The idea at the base 1 of the invention is to attempt to correct errors in successive stages, the implementation of each of the dependent steps of at least one test performed in the previous step. The implementation of the steps depends especially on the rate and distribution of detected errors in the data packets. In a favorable case, only the first step, that is to say, the first test is carried out. If all of the first K blocks was received at the client device, it means that there are no errors, no loss of information, and Deco-dage video can be made to visualization.

In a worst case, it is necessary to appeal to the FEC algorithm to retrieve missing this step to actually correct the loss of blocks distributed in time, when the error rate is not too large (it then qualified low to medium). The implementation condition of this step is that the total number of received blocks is at least equal to K.

According to the typology of errors, this step may well be either a failure or not feasible for lack of data, and therefore does not allow to correct errors: it is for example the case when it occurs errors eg by whole packet, i.e. of the combined errors representing a qualified error rate of medium to high. Another routine is then to be implemented according to the invention, which involves an exchange with the sending device. The customer who is in this case sends a message in unicast to the transmitter, which eventually returns after missing a statistical algorithm. This obviously assumes that these have been identified by the client device.

The integrity or readability of the data returned contralateral EAA according to steps a / and b / the reception by the client device, which may then proceed to decode the stream for its visualization.

The method of the invention goes further, and also considers the hypothesis that the data returned by the transmitter device are still not réceptionnées réceptionnées or not fast enough in conditions for displaying the video stream properly .

In this case, the following steps are implemented according to the invention:

E / in the absence of correct reception Mynx REN- data, control of receiving an intra coded reference picture in the packet G received data, then

"F / if such a reference image is received completely implement an error concealment algorithm applied to the compressed data, and

'G / decoding and viewing the video stream.

The reference image to which reference is made is the one with which the package typically begins G consisting essentially of a group of images whose succession repeated periodi-cally to the end of encoding, is in fine the encoded video stream. This is an internal coded reference image whose decoding is autonomous and does not depend on previous or subsequent pictures. This feature is used in conjunction with images of other categories, e.g. predictive coding, in a sequence that defines a specific group. The vi- sible images are then generated (decoded) from the coded pictures in this predefined group.

This type of image, independent of other image types consti-killing group, is regarded as a reference image, and testing to determine the relevance of the implementation of an error concealment algorithm is preferably according to the invention, based on the full existence of such reference image.

When a client device could not properly rebuild-ing the video stream sent by the sending device, it implements a step of concealment of errors. This n 'is Cepen-ing done on the condition that a reference image is received in the G processed data packet. This new stage based on concealment can hide errors and provide, at the time of decoding the encoded samples and cal-culés replacing the missing samples of the signal from the data of one reference image and any other images received in the package G. the estimation of man-quantes parts of the images on the encoded data allows to visually reduce the data reception errors in compressed video streams, and s' carried out by exploiting the spatial and temporal correlations between images, either within the same image (spatial correlation) in the past or future frames (temporal correlation). The idea at the basis of these routines is that there is rarely abrupt changes between adjacent pixels spa-tially of an image, and there is a tem-Porelle continuity in the successive images of a video sequence .

The process of the invention then envisages the possibility of non-receipt of such an image intra coded reference in the stream data packet being processed. In this case, according to a proper possibility to the invention, the following steps are implemented:

- h / in case of non-receipt of an internal coded reference picture in the packet G data received, VE- VERIFICATION of receiving at least one reference image of the previous packets 6-1 and following G + l

- i / implementation of a method for interpolation on the uncompressed data or linear for Reconstruction of images, then

"J / decoding and viewing of the video stream.

This new stage of the process of the invention uses this time opportunities for algorithmic reconstruction rather than error concealment techniques used above, in a slightly different context. These methods interpola-ing that are also considered temporally or spa-tial, and this time working with non-com pressed video stream. In a temporal interpolation, they are used to generate missing images in a stream using the previous and next images.

If it happens that none of the above steps will put per-lead, particularly under the last step implementation, because there is no reception of at least a reference image Gl previous packages and following the G +, the package G is not displayed.

The process of the invention is actually a chain of successive solutions to adapt the system response to faulty transmission of the video signal whose nature and severity are detected progressively.

The invention will now be described in more detail with refe-rence to the accompanying figures, showing a possible implementation of the method of the invention, and for which

1 shows a flowchart diagramming the overall implementation of the method 1 of the invention; and

2 shows the same data packet, establishing a possible encoding of a group of pictures and, by repetition tion, a video stream, various types of possible data losses being shown in these data packets. Referring to FIG 1, the first test conducted under the method consists in verifying whether the first K blocks have been received. At each step, after each test, in a favorable hypothesis, that is to say if there is no data loss problem, or if they have been resolved, the consti-killing data packets the video stream are processed by the video decoder to the dissemination of the transmitted stream.

If the result at the first test is positive, this means that no-MENT the data packet corresponds to the case No. l of Figure 2, wherein the first K blocks of the data packet have been received. In Figure 2 a visual code defines the various blocks, appearing in the legend at the bottom of the figure, and including the first blocks K, K to N blocks, the reference images X and lost blocks.

In a negative case, the next test relates to the X number of blocks received from the set of K + N transmitted blocks, that is to say including the N redundancy blocks, placed one after the K blocks in the representations of Figure 2. If the total number of received blocks is at least equal to K (even in case of errors, lost blocks), that is to say if X ≥ K, corresponding to the case # 2 Figure 2, a forward error correction algorithm (FEC) is used, then a new verification test is impious-mented. Since this is a type of correction which is adapted to losses distributed in time, over the entire data packet, that is to say on the set of blocks N + K, this strate error correction -gie does not necessarily work.

If this test is negative, which may mean that data loss is rather concentrated, not spread, or the number of X blocks received from the N + K blocks is less than K -correspondant if No. 3 of the Figure 2 - a procedure (called NAK) for returning non-received data is started. This im-routine plique a unicast communication in the opposite direction, between one of the receiving client devices and a transmitting device, and requires an identification of the missing data.

This refers in unicast or multicast iden-tified by the client as data not received according to a sta-tical algorithm, and a new correct reception test is performed in the client device. The next test relates to the correct receipt of a reference image X internal encoding packet or image group i. In Figure 2, the reference I picture is placed at the beginning of the package, which is usually the place it occupies in such encodings. If the image I is com-pletely or partly received, that is to say in cases # 4 and # 5 illustrated in Figure 2, errors consta-Tees transmission and n 'have not been resolved in the previous steps are processed means a concealment algorithm. The errors in this case are hidden in the video decoder, but not corrected. The objective of such routines, which acts on the stream of compressed data, is to visually reduce reception errors in the video stream.

In the opposite case, that is to say if there is no re-receipt of an internal coded reference image I in the group of pictures or package G, the method 1 ' invention tests the reception of such Σ reference image in the packets or groups of images and G Gl + l. This corresponds to case No. 6 in Figure 2. If this is the case, a reconstruction of the deteriorated image can still be implemented on the basis of an interp-lation algorithm acting on uncompressed video stream and generating the missing frames or parts of missing picture respecti-vely by using the immediately preceding or sui-lowing images in the succession of images, in a perspective tempo-connects,

When none of these successive treatments succeeds satisfactorily my-Niere, the package or group of images is not affi-ket.

The invention does of course not limited to the examples described and explained with reference to figures, but covers the back and smiling versions that are within the scope of the claims.

WE CLAIM

1 - A method for improving the reliability and the quality of reception data stream video on a wireless network governed by WIFI type communication protocols, the data stream video being encoded as packets 6 each having K first data blocks and N redundancy blocks for the correction of forward error (FEC), said stream comprising a succession of packets G each having K + N blocks being transmitted by at least one transmitter device 1 'sending multicast to devices visualization client type smartphone, characterized by the following steps, implemented by the client device for each of the data packets received from the G or emitting devices:

a / control the correct reception of the first K blocks of data, decoding and display of the flow vi- deo in case of positive result;

b / in the absence of reception of all K pre- miers data blocks, and if the total number X of K + N blocks received is at least equal to K, decoding by means of a correction algorithm Forward Error (FEC) to recover missing data for K blocks, decoding and display of the video stream in case of positive result tif;

c / if the total number X of K + N blocks received is less than K or absence of a positive result to the decoding using the forward error correction algorithm (CEF), en- voi by the client device message unicast Don-ing to the information transmitter or at least on the lost blocks;

d / if a statistical algorithm emitting device organizes the return of missing data, control of the correct receipt of data returned by the transmitting device according to a / b /, decoding and illumi- nating the video stream in case of a positive result .

2- A method of improving the reliability and quality of video streams received on a wireless LAN according to the preceding claim, characterized by the following steps:

E / in the absence of correct reception of the data returned as d / or absence of returned data, control the reception of a reference image in internal co DAGE in the packet G received data, then

- f / if such a reference image is received com- pletely, implementation of a concealment algorithm error applied to the compressed data, and

- g / decoding and viewing the video stream.

3- A method for improving the reliability and the stream reception quality video data on a wireless LAN according to the preceding claim, characterized by the following steps

- h / in case of non-receipt of an internal coded reference picture in the packet G received data Veri fication of receipt of at least one previous packet reference image Gl and the following G + ,

- i / implementation of a method for interpolation on the uncompressed data or linear for Reconstruction of images, then

- j / decoding and viewing of the video stream.

4- A method for improving the reliability and the quality of reception data stream video on a wireless LAN according to the preceding claim, characterized in that, in the event of incompletion of reception of at least one image of re-ference Gl previous packages and following the G +, the package G is not displayed.