Receiver for data decompression with autoencoder enhancement

Abstract

The invention relates to a receiver (120) comprising:-a memory (502) storing a first autoencoder convolutional neural network designed to obtain enhanced data from compressed data; -a first interface (503) designed to receive compressed data via a communication channel (130); -a processor (501) configured to decompress the data into decompressed data and to apply the first autoencoder convolutional neural network to the compressed data in order to obtain enhanced data; and-a second interface (504) designed to transmit the enhancement data.

Description

[0001] The present invention relates to the field of reception and processing of data, in particular image data. The invention is particularly but not exclusively applicable to data exchanged in or by a motor vehicle.

[0002] Data compression, in particular image data compression, is known for limiting the amount of data transmitted over a communication channel.

[0003] Some communication channels have limited bandwidth. This is particularly the case for the CAN bus, although this type of communication channel is widely used because it is secure and inexpensive. This communication channel is particularly used in motor vehicles, for example to transmit photometric images between a central control module of the vehicle and the lighting devices of the vehicle.

[0004] The image data transmitted, including in automobiles, now has high resolution. The same limitation on the throughput of the communication channel results in a high compression ratio, which can cause defects in the decompressed image at the receiver, such as noise, artifacts, or low PSNR. PSNR refers to "peak signal-to-noise ratio".

[0005] This is particularly the case for lossy compression algorithms, such as linearization-based algorithms, gradient-based algorithms, JPG algorithms, PCA algorithms, or other algorithms. When these images are photometric images capable of controlling the lighting devices of a motor vehicle, the defects or artifacts affect the projected beam, especially when the resolution of the pixelated beam lighting module is also high. Such defects can cause safety problems and / or make the beam non-compliant with regulations.

[0006] Therefore, there is a need to receive and process high-resolution image data via a communication channel with limited throughput without causing substantial defects or artifacts in the image finally processed at the receiver.

[0007] For this purpose, a first aspect of the present invention relates to a data processing method, which includes the following operations:

[0008] · In a preliminary stage, supervised learning is performed on a first autoencoder convolutional neural network based on a first set of training data, which consists of image pairs including an image of a given quality and a compressed image obtained by compressing the image of the given quality, wherein the supervised learning can minimize the difference between the enhanced image obtained by processing the compressed image by the first autoencoder convolutional neural network and the image of the given quality associated with the compressed image on a pairwise basis;

[0009] · Store the first autoencoder convolutional neural network in the receiver.

[0010] · During the current stage implemented by the receiver, the method further includes:

[0011] · Receive compressed data from a communication channel;

[0012] · Decompress the compressed data into decompressed data;

[0013] · Apply a first autoencoder convolutional neural network to the compressed data to obtain enhanced data;

[0014] · Transmit the enhanced data.

[0015] This enhancement by artificial intelligence eliminates at least some of the defects in the decompressed data. Thus, compression with a high compression ratio can be provided, and thus high-resolution data can be transmitted via a communication channel with limited throughput.

[0016] Depending on the embodiment, the decompression of the compressed data can use a linearization-based decompression algorithm, a gradient-based decompression algorithm, a JPG decompression algorithm, or a PCA decompression algorithm.

[0017] Thus, an image compressed by a compression algorithm can be enhanced. For this purpose, the first set of training data can advantageously include image pairs, where the images are compressed according to different compression algorithms. In fact, a high level of enhancement of images compressed by such algorithms can be achieved.

[0018] Depending on the embodiment, the decompression of the compressed data can be implemented by processing by neuron layers, which include an output layer of a second autoencoder convolutional neural network and at least one convolutional hidden layer, the output layer including a first number of dimensions and the at least one convolutional hidden layer including a second number of dimensions, which is less than the first number of dimensions.

[0019] Thus, the compression itself can originate from an autoencoder (referred to as the second autoencoder). Thus, the compression level can be controlled by setting the number of dimensions of the latent vector. Even at a high compression level, the enhancement provides high-quality enhanced data with few or no defects.

[0020] Additionally, in a preliminary stage, the method can further include unsupervised learning of the second autoencoder convolutional neural network based on a second set of training data.

[0021] Thus, the learning step is simplified for the second autoencoder. For example, the unsupervised learning can include optimizing the second autoencoder to minimize the difference between the input data and the output data of the autoencoder obtained after compressing and decompressing the input data.

[0022] Depending on the embodiment, the compressed data and the enhanced data can represent the photometry of a motor vehicle lighting device.

[0023] For regulatory and safety reasons, it is important to minimize defects in this photometric data. Therefore, it is particularly advantageous to use the enhancement according to the present invention.

[0024] Additionally, the enhanced data can be transmitted to the control module of the motor vehicle lighting device.

[0025] Accordingly, the receiver can advantageously be implemented in the lighting device in order to control at least one of the lighting modules of the lighting device.

[0026] Additionally or alternatively, compressed data can be received from a central control module of the motor vehicle, and the communication channel can be a CAN bus.

[0027] The advantage offered by this type of communication channel is that it is secure and inexpensive. However, it has a limited throughput and may require a high compression ratio, making it particularly advantageous to use the enhancement according to the present invention.

[0028] A second aspect of the present invention relates to a receiver comprising:

[0029] · a memory storing a first autoencoder convolutional neural network capable of obtaining enhanced data from compressed data;

[0030] · a first interface capable of receiving compressed data via a communication channel;

[0031] · at least one processor configured to decompress the data into decompressed data and apply the first autoencoder convolutional neural network to the compressed data in order to obtain enhanced data;

[0032] · a second interface capable of transmitting the enhanced data.

[0033] A third aspect of the present invention relates to a system comprising: a receiver according to the second aspect of the present invention; an encoder capable of receiving input data, compressing the input data into compressed data and transmitting the compressed data to the receiver via a communication channel.

[0034] Depending on the embodiment, the encoder can be integrated into the central control module of the motor vehicle, the receiver can be integrated into the lighting device of the motor vehicle, the communication channel can be a CAN bus, and the input data can represent lighting photometry.

[0035] Other features and advantages of the present invention will become apparent by reference to the following detailed description and the drawings, in which:

[0036] Figure 1 shows a data transmission system according to an embodiment of the present invention; ​

[0037] Figure 2 shows the structure of a first autoencoder convolutional neural network according to an embodiment of the present invention;

[0038] Figure 3 is a schematic diagram showing the steps of a data processing method according to an embodiment of the present invention;

[0039] Figure 4 shows a data compression system using a second autoencoder convolutional neural network according to an embodiment of the present invention;

[0040] Figure 5 shows the structure of a receiver according to an embodiment of the present invention.

[0041] This description focuses on the features that distinguish the methods, systems, encoders, and decoders from those known in the prior art.

[0042] Figure 1 shows a system 100 for transmitting data, in particular image data.

[0043] System 100 includes an encoder 110 and a receiver 120 connected by a communication channel 130.

[0044] The encoder 110 can be integrated into a motor vehicle device, such as a control module responsible for controlling the lighting of a motor vehicle. For example, such a control module can be a PCM (Powertrain Control Module) or an ECU (Electronic Control Unit).

[0045] The receiver or decoder 120 can be integrated into a motor vehicle device, such as a lighting device including a lighting module capable of performing a lighting function based on data transmitted by a control module including the encoder 110. At least one lighting module of the lighting device is preferably a pixelated module, such as a matrix with electroluminescent elements (such as LEDs), a matrix with micromirrors (such as DMD (Digital Micromirror Device)), a monolithic source of electroluminescent elements on the same substrate, or any other technology capable of implementing a pixelated lighting beam. The monolithic source includes a plurality of submillimeter-sized electroluminescent semiconductor elements directly epitaxially grown on a common substrate, which is typically made of silicon. Different from a conventional LED matrix where each basic light source is a separately produced electronic component mounted on a substrate such as a printed circuit (PCB), the monolithic source should be regarded as a single electronic component in which several ranges of semiconductor electroluminescent junctions are formed in a matrix form on a common substrate.

[0046] ​​​​​Thus, communication channel 130 can be a wired link, such as a CAN bus or an Ethernet link. An example of a CAN bus is considered by way of illustration below. It offers the advantage of a secure and inexpensive link. However, the CAN bus has a limited throughput and requires data to be transmitted in the form of 8-bit coded integers.

[0047] Alternatively, encoder 110 is integrated into the control module of a motor vehicle, and receiver 120 is integrated into a remote server of the motor vehicle. In this case, communication channel 130 includes a wireless communication channel that allows the encoder to access an IP network (in which the remote server including decoder 120 is located). Such a wireless communication channel can be a 3G, 4G, 5G or any next-generation cellular link.

[0048] There is no limitation on communication channel 130, which can be a wired or wireless link. As will be better understood from reading the description below, most communication channels have a limited throughput for the data they transmit.

[0049] Encoder 110 includes data compression module 111, which is capable of receiving input data, such as data for encoding an image (such as a photometric image), and capable of obtaining compressed data from the input data and from a data compression algorithm (such as a linearization-based algorithm, a gradient-based algorithm, a JPG algorithm, a PCA algorithm or other algorithms). Alternatively, the compressed data corresponds to the latent vector of an autoencoder convolutional neural network, as described in detail below with reference to Figure 4 as detailed.

[0050] Decoder 120 or receiver 120 includes decompression module 121, which corresponds to decompression module 111 and is capable of decompressing the compressed data received via communication channel 130 in order to obtain decompressed data.

[0051] As previously mentioned, when the compression ratio is high (e.g., greater than 80%), defects may appear in the image corresponding to the decompressed data. Thus, the present invention provides adding a software or hardware module 122 for enhancing the compressed data in order to obtain enhanced data with fewer defects than the decompressed data at the output of decompression module 121.

[0052] For this purpose, enhancement module 122 can include the first autoencoder convolutional neural network 200 described in Figure 2 as described.

[0053] The autoencoder convolutional neural network (hereinafter also referred to as the autoencoder) includes multiple neuron layers, which include an input layer 200.1, an output layer 201.1, at least one convolutional hidden layer on the input layer 200.1, and at least one convolutional hidden deconvolution layer on the output layer 201.1.

[0054] In Figure 2 the example shown, the first autoencoder 200 includes a first convolutional hidden layer 200.2 and a second convolutional hidden layer 200.3. The first autoencoder 200 includes a first deconvolutional hidden layer 201.2 and a second deconvolutional hidden layer 201.3 arranged symmetrically.

[0055] The first convolutional hidden layer 200.2 has the same number of dimensions as the first deconvolutional hidden layer 201.2. Similarly, the second convolutional hidden layer 200.3 includes the same number of dimensions as the second deconvolution layer 201.3.

[0056] Therefore, the autoencoder includes a set of symmetric neuron layers.

[0057] The hidden layer or the central layer with the minimum number of dimensions (in this case, layer 200.3 and layer 201.3) can exchange data called "code" or "latent vector" 210. The latent vector is a compressed version of the data received by the input layer 200.1.

[0058] The autoencoder 200 according to the present invention can enhance the compressed data received at the input end into enhanced data with fewer defects and closer to the input image received and compressed by the encoder 110.

[0059] For this purpose, the first autoencoder 200 is the result of supervised learning based on a first set of training data. The first set of training data includes image pairs, and each image pair consists of:

[0060] · An image of a given quality, especially an image of the best quality, that is, an uncompressed image. There is no limit to the resolution of this type of image. The given or best-quality image has no defects;

[0061] · A compressed image, which is obtained by compressing an image of a given quality by a given compression algorithm (for example, one of the compression algorithms mentioned above). Such a compressed image may have the defects as described above.

[0062] The training data set preferably includes image pairs that vary in the following aspects:

[0063] · In the compression algorithm applied to obtain the compressed image;

[0064] · In the compression ratio applied to obtain the compressed image; and / or

[0065] · In terms of the types of defects that the compressed image has, these include artifacts, PSNR below a given threshold (e.g., less than 25), missing parts of the image, or other compression quality metrics such as the maximum error or the mean squared error (MSE).

[0066] The first set of training data includes more than a hundred image pairs, preferably thousands or tens of thousands of image pairs. Then, the supervised learning consists of submitting the compressed image at the input of the first autoencoder 200 for each image pair. The image obtained at the output of the output layer 201.1 is compared with the best quality image associated with the compressed image in order to estimate the difference, such as the mean squared error between the image at the output of the first autoencoder and the best image quality. Then, the first autoencoder 200 is modified according to the determined deviation, for example by changing the eigenvalue of one or more neuron layers, so as to reduce the calculated deviation.

[0067] In the example considered here, where the encoder is integrated into the PCM or ECU of a motor vehicle and the decoder is integrated into the lighting device, the data from the first set of training data are image pairs representing the light beam to be implemented by the motor vehicle lighting device. Such an image is also called a photometric image.

[0068] However, there is no limitation on the data from the first set of training data, which can be any type of image. In the example where the decoder 120 is implemented in a remote server, the data from this set of training data can be, for example, images acquired by a motor vehicle camera.

[0069] After training with the entire first set of training data, the mean squared error is minimized, and the first autoencoder 200 is able to reconstruct the best quality image from an image with one or more defects after compression. Thus, the autoencoder 200 can be implemented in the receiver 120 described above.

[0070] Advantageously, the first autoencoder 200 can provide the reuse of the features of the convolutional neuron layer by performing deconvolution on the neuron layer. At least one skip connection can allow this reuse, and the at least one skip connection connects two non - consecutive convolutional neuron layers of the first autoencoder 200.

[0071] For example, the first autoencoder 200 includes at least one skip connection 220.1 between a pair of layers having the same number of dimensions or features, each pair including a convolutional layer and a deconvolution layer. In Figure 2In the example shown, the skip connection 220.1 can thus include a connection between the input layer 200.1 and the output layer 201.1, a connection between the first convolutional hidden layer 200.2 and the first deconvolutional hidden layer 201.2, and a connection between the second convolutional hidden layer 200.3 and the second deconvolutional hidden layer 201.3.

[0072] Additionally or alternatively, the first autoencoder 200 includes at least one skip connection 220.2 between a pair of layers having different numbers of dimensions or features, each pair including a convolutional layer and a deconvolutional layer. In Figure 2 In the example shown, the skip connection 220.2 can thus include a connection between the first convolutional hidden layer 200.2 and the second deconvolutional hidden layer 201.3, and a connection between the second convolutional hidden layer 200.3 and the first deconvolutional hidden layer 201.2.

[0073] Advantageously, such skip connections enable complex data processing operations to be performed using a deep neural network.

[0074] Figure 3 shows a data transmission system according to an embodiment of the present invention.

[0075] The method includes a preliminary phase 300, which includes a step 301 of obtaining a first set of training data, the first set of training data including image pairs as described above. There is no limitation on the way the training data set is obtained. For example, the best quality images can be sourced from real-world situations or from simulations.

[0076] In step 302 of the preliminary phase 300, based on the image pairs of the first set of training data obtained in the previous step 301, the first autoencoder convolutional neural network 200 is trained by supervised learning. The first autoencoder 200 thus obtained is capable of enhancing the quality of the compressed data.

[0077] In step 303 of the preliminary phase 300, the first autoencoder convolutional neural network 200 is stored in a receiver, such as the receiver 120 described above. As described above, the receiver 120 can be integrated into a lighting device for a motor vehicle or a remote server of a motor vehicle.

[0078] The processing method further includes a current phase 310, which includes a step 311 of receiving compressed data by the receiver 120 via the communication channel 130 described above. In particular, the received data has been pre-compressed by the encoder 110 described above, and there is no limitation on the data compression technique.

[0079] In step 312, the decompression module 121 decompresses the compressed data received in the previous step 311, as described above.​

[0080] In step 313, the decompressed data is processed by the first autoencoder 200 so as to be enhanced. Thus, at the end of step 313, enhanced data is obtained. Given that the first autoencoder 120 is derived from machine learning, the enhanced data can achieve an optimal image quality, approaching the image initially compressed by the encoder 110.

[0081] In step 314, the receiver 120 can transmit the enhanced data. For example, the receiver 120 can transmit the enhanced data to a memory for storage. In an embodiment where the encoder 110 is integrated into the PCM and the receiver 120 is integrated into the signaling device, the enhanced data can advantageously be transmitted to the light source control module to implement photometry corresponding to the output data.

[0082] Figure 4 shows a data compression module 111 and a data decompression module 121 according to an embodiment of the present invention.

[0083] As described above, the compression module 111 and the decompression module 121 can be capable of implementing compression / decompression algorithms, such as linearization-based algorithms, gradient-based algorithms, JPG algorithms, PCA algorithms, or other algorithms.

[0084] According to Figure 4 a variant shown, the compression / decompression is implemented by means of a second autoencoder convolutional neural network (hereinafter also referred to as the second autoencoder), and the compression module 111 and the decompression module 121 each include a part of the second autoencoder. Thus, the compression module 111 includes a first part 400 of the second autoencoder, while the decompression module 121 includes a second part 410 of the second autoencoder.

[0085] The second autoencoder is capable of compressing input data, in particular images of optimal quality, such as photometry images for lighting devices. The second autoencoder can be constructed by unsupervised learning based on a second set of training data different from the first set of training data.

[0086] The second autoencoder consists of an input layer 401 implemented in the encoder 110 and an output layer 411 implemented in the receiver 120, the input layer and the output layer having the same number of nodes or neurons, and thus the same number of dimensions.

[0087] The autoencoder system further includes one or more convolutional hidden layers, each convolutional hidden layer having fewer dimensions than the number of dimensions of the input layer 401 and the output layer 411.

[0088] ​The convolutional hidden layer or the central layer with the minimum number of dimensions is capable of exchanging "codes" or "latent vectors", and this latent vector is thus a compressed version of the input data.

[0089] Therefore, the central layer can be shared between the encoder 112 and the decoder 123 to exchange compressed data, thus reducing the throughput requirements and the amount of data exchanged between the encoder and the decoder while minimizing the loss. Thus, the first part 400 includes a central encoding layer 402, while the second part 410 includes a central decoding layer 412. The central encoding layer 402 and the central decoding layer 412 are capable of exchanging codes or latent vectors including a number of dimensions less than that of the input data.

[0090] The second autoencoder is trained by unsupervised learning in such a way that the mean squared error between the input data and the output data from the output layer is minimized, where the central layer has a given number of dimensions (i.e., a given compression level).

[0091] For this purpose, a second training dataset can be submitted to the second autoencoder. The training dataset can include a set of images, such as the photometric lighting images for vehicles in the example considered here. For each image in the second set of training data, the autoencoder evaluates the mean squared error between the image submitted to the input layer 401 and the image supplied by the output layer 411, and based on this mean squared error, it changes the characteristics of its neurons as well as the number of neurons and even the number of hidden layers while maintaining the constraints in order to obtain a latent vector with a given number of dimensions. Thus, the aim is to minimize the mean squared error through learning.

[0092] The latent vector is a compressed version of the input data, where the compression ratio CR is according to the following formula:

[0093] ·CR = (Nbits * Im_Size – NbitsLV * LVdim) / Nbits * Im_Size;

[0094] In the formula, Nbits is the number of bits used to encode each pixel of the input image, Im_Size is the size of the image in terms of the number of pixels, NbitsLV is the number of bits used to encode each dimension of the latent vector, which is defined and usually equal to 32 bits, and LVdim is the number of dimensions of the latent vector.

[0095] More generally, Nbits * Im_Size represents the size of the input data in terms of the number of bits.

[0096] For a given image size, the compression ratio CR can thus be changed by changing the number of dimensions LVdim of the latent vector.

[0097] In particular, the following compression ratios can be obtained:

[0098] · For LVdim = 516, CR = 92%;

[0099] · For LVdim = 1024, CR = 84%;

[0100] · For LVdim = 3072, CR = 52%.

[0101] Therefore, the lower the number of dimensions of the latent vector, the higher the compression ratio CR. The choice of the compression ratio can depend on the quality metric for comparing the output data with the input data. For example, such a metric can include the peak signal-to-noise ratio (PSNR) or the mean squared error (MSE).

[0102] For example, the number of dimensions of the latent vector that provides a PSNR greater than a given threshold (e.g., 30) can be defined.

[0103] Thus, the first part 400 and the second part 410 of the autoencoder convolutional neural network are obtained and can be implemented in the encoder 110 and the decoder 120 respectively during the preliminary stage 300 described above.

[0104] However, when the communication channel 130 has a limited throughput, especially in the case of the CAN bus commonly used between the PCM and the lighting device, a high compression ratio, and even data reformatting, are required to be able to transmit the latent vector over the communication channel 130. As a result, the decompressed data from the output layer 411 may be defective, so it is advantageous to use the enhancement module 122 described above.

[0105] In particular, compared with the decompressed data from the output layer 411 of the second autoencoder, the PSNR value of the enhanced data can be several points more, especially 5 points. Other metrics (such as the maximum error and the mean squared error) are also improved.

[0106] In the case where the compression / decompression is not performed by the second autoencoder but by one of the compression / decompression algorithms discussed above, the PSNR gain allowed by the enhancement module 122 can even reach 10 points. Other metrics (such as the maximum error and the mean squared error) are also improved.

[0107] Figure 5 shows the structure of the decoder or receiver 120 according to an embodiment of the present invention.

[0108] ​The decoder 120 includes a processor 501 configured to communicate unidirectionally or bidirectionally with a memory 502 (such as a random access memory (RAM), a read-only memory (ROM), or any other type of memory (flash memory, EEPROM, etc.)) via one or more buses or via a wired connection. As a variant, the memory 502 includes several of the above-described types of memories. Preferably, the memory 502 is a non-volatile memory.

[0109] The memory 502 permanently or temporarily stores all data generated by executing steps 311 to 314 of the data processing method described above. In step 303 described above, the memory 502 further stores the first autoencoder convolutional neural network 200.

[0110] The memory 502 further stores the reference Figure 4 decompression algorithm or the second part 410 of the second convolutional neural network described.

[0111] The processor 501 is capable of executing instructions stored in the memory 502 to perform steps 312 and 313 of the method shown in the reference Figure 3 . Alternatively, the processor 501 can be replaced by a microcontroller designed and configured to perform steps 312 and 313 of the method according to Figure 3 .

[0112] Therefore, the decompression module 121 and the enhancement module 122 shown above can be implemented by the processor 501 or the microcontroller. As a further alternative, one processor or one microcontroller is dedicated to the decompression function, and another processor or microcontroller is dedicated to the enhancement function.

[0113] In step 311 described above, the receiver 120 may include an input interface 503 capable of receiving compressed data. There is no limitation on the first input interface 503, and the first input interface is functionally connected to the communication channel 130 described above.

[0114] The decoder 120 may further include a second interface 504 (i.e., an output interface) capable of transmitting output data in step 314 described above.

[0115] The present invention is not limited to the embodiments described as examples above, but extends to other alternatives.

Claims

1. A data processing method, comprising: the following operations: · In a preliminary stage (300), perform supervised learning (302) on a first autoencoder convolutional neural network based on a first set of training data, the first set of training data consisting of image pairs, the image pairs including an image of a given quality and a compressed image obtained by compressing the image of the given quality, wherein the supervised learning can minimize the difference between the enhanced image obtained by processing the compressed image by the first autoencoder convolutional neural network and the image of the given quality associated with the compressed image on a pairwise basis; · Store (303) the first autoencoder convolutional neural network in a receiver (120); wherein, during a current stage (310) performed by the receiver, the method further comprises: · Receive (311) compressed data from a communication channel (130); · Decompress (312) the compressed data into decompressed data; · Apply (313) the first autoencoder convolutional neural network to the compressed data to obtain enhanced data; · Transmit (314) the enhanced data.

2. The data processing method according to claim 1, wherein, the decompression of the compressed data uses a linearization-based decompression algorithm, a gradient-based decompression algorithm, a JPG decompression algorithm, or a PCA decompression algorithm.

3. The data processing method according to claim 1, wherein, the decompression of the compressed data is implemented by processing by a neuron layer, the neuron layer including an output layer (411) of a second autoencoder convolutional neural network and at least one convolutional hidden layer (412), the output layer including a first number of dimensions and the at least one convolutional layer including a second number of dimensions, the second number of dimensions being less than the first number of dimensions.

4. The method according to claim 3, further comprising: In the preliminary stage (300), perform unsupervised learning on the second autoencoder convolutional neural network based on a second set of training data.

5. The method according to any one of the preceding claims, wherein, the compressed data and the enhanced data represent the photometry of a lighting device.

6. The method according to claim 5, wherein, the enhanced data is transmitted to a control module for a lighting device of a motor vehicle.

7. The method according to claim 5 or 6, wherein, compressed data is received from an encoder (110) of a central control module of a motor vehicle, and wherein the communication channel (130) is a CAN bus.

8. A receiver (120), comprising: · A memory (502) that stores a first autoencoder convolutional neural network capable of obtaining enhanced data from compressed data; · A first interface (503) that can receive compressed data via a communication channel (130); · A processor (501) configured to decompress data into decompressed data and apply the first autoencoder convolutional neural network to the compressed data to obtain enhanced data; · A second interface (504) capable of transmitting the enhanced data.

9. A system, comprising: A receiver (120) as claimed in claim 8; an encoder (110) capable of receiving input data, compressing the input data into compressed data and transmitting the compressed data to the receiver via a communication channel (130).

10. The system as claimed in claim 9, wherein, the encoder (110) is integrated into a central control module of a motor vehicle, wherein the receiver (120) is integrated into a lighting device of the motor vehicle, wherein the communication channel (130) is a CAN bus, and wherein the input data represents lighting photometry.

Images