WO2008029038A1 - Device and method for estimating the dimensions of a parking space, motor vehicle comprising such a device - Google Patents

Abstract

This device for estimating the dimensions of a parking space (12), intended to be carried on board a motor vehicle (10), comprises: a synthetic aperture radar sensor (20) for generating a complex image of SAR type of this parking space; a module (22, 24) for processing this complex image of SAR type in such a way as to obtain a real-valued image; and a computer (26) for extracting a value of length and a value of depth of the parking space from this real-valued image.

Description

 Device and method for estimating the dimensions of a parking space, a motor vehicle comprising such a device

The present invention relates to a device for estimating the dimensions of a parking space. It also relates to a corresponding method and a motor vehicle comprising such an estimation device. This device for estimating the dimensions of a parking space is intended to be loaded into a motor vehicle.

Such devices are known, in particular that described in patent application GB 2 319 420.

In this document, a device forming a transmitter and radar sensor is disposed on a lateral portion of a motor vehicle. This device operates by remission and reception of radar pulses. It is thus possible to calculate the available length of a parking space between two vehicles already parked. However, given the technical limitations of the radar device, the precise dimensions of this parking space can not be calculated.

Other technologies may be used, such as laser telemetry, the use of ultrasonic sensors, but do not provide a more accurate result for estimating the dimensions of a parking space. The invention therefore aims to provide a device for accurate estimation of the dimensions of a parking space, including its length and depth.

The subject of the invention is therefore a device for estimating the dimensions of a parking space intended to be loaded into a motor vehicle, characterized in that it comprises: a synthetic aperture radar sensor for the generation a complex SAR-type image (of the "Synthetic Aperture Radar") of this parking space, - a processing module of this SAR-type complex image so as to obtain an image with real values, and - a computer for extracting a length value and a depth value of the parking space from this real-valued image.

A synthetic aperture radar sensor is a lateral aiming system provided with a special device for processing backscattered signals, making it possible to improve the geometric resolution of the image along the axis parallel to the vector path. The successive echoes of the same point are superimposed and identified by their Doppler effect. Thus, the use of a synthetic aperture radar sensor in an estimation device mounted on a motor vehicle makes it possible, when the motor vehicle moves along the parking space, to generate a SAR-type image. which after processing to obtain a real-valued image can be a good basis for extracting the length and depth values of the parking space.

The invention also relates to a motor vehicle comprising an estimating device according to the invention.

Advantageously, the synthetic aperture radar sensor is fixed on a lateral outer portion of the motor vehicle.

The invention also relates to a method of estimating the dimensions of a parking space using a device embedded in a motor vehicle, comprising the following steps:

generation of a SAR-type complex image of this parking space by a synthetic aperture radar sensor,

processing this SAR complex image so as to obtain a real-valued image, and

extraction of a length value and a depth value of the parking space from this real-valued image. Preferably, the step of processing the complex image of the SAR type comprises generating the real-valued image from the module of the points of the SAR complex image and a contrast enhancement of the pixels of the SAR image. real-valued image.

Indeed, the complex SAR-type image is not directly exploitable, it is advisable to generate an image with real values from the module of the points of the complex image and to enhance the contrast, for example from minimum and maximum values of the pixels of the real-valued image.

The processing step of the SAR-type image may also comprise a filtering of the real-valued image, this filter performing a two-dimensional smoothing of its pixels. This removes the real-valued image of some of its artifacts.

Advantageously, the SAR image processing step comprises associating with each pixel of the real-valued image of a weighting parameter as a function of a measured variation between adjacent pixel values. This weighting parameter then makes it possible to refine the estimation of the dimensions of the parking space.

Finally, it is possible to provide a predetermined threshold value on the basis of which, during the processing step of the SAR-type image, a zero intensity is assigned to all the pixels of the real-valued image whose intensity is below this predetermined threshold value.

This treatment makes it possible to get rid of pixels of low intensity which could possibly distort the extraction of the values of length and depth of the parking space. The invention will be better understood with the aid of the description which follows, given solely by way of example and with reference to the appended drawings in which:

- Figure 1 shows schematically the general structure of a motor vehicle comprising an estimating device according to the invention;

FIG. 2 represents the successive steps of an estimation method implemented by the device of FIG. 1; and

FIGS. 3 and 4 represent images obtained using the estimation device of FIG. 1, before and after treatment according to the invention. FIG. 1 illustrates a motor vehicle 10 moving in the direction indicated by the arrow D along a parking space 12 delimited in length by two vehicles 14 and 16, one 14 being located at the rear and the other 16 at the front of the parking place.

The parking space 12 is delimited in width, for example by a sidewalk 18.

The motor vehicle 10 comprises a device for estimating the dimensions of the parking space according to the invention. The estimation device comprises a synthetic aperture radar sensor 20 for generating a SAR-type complex image of the parking space. This radar sensor 20 is fixed on a lateral outer part of the motor vehicle 10.

The acquisition system of the radar sensor 20 samples and digitizes the received signals. The number of quantization stages of the digitized signals is generally a few bits. The width of the image obtained is characterized by its swath. On the other hand, the azimuth size is limited only by the maximum total acquisition time, that is to say by memory constraints for embedded systems and through the bandwidth of the data transmission channel.

In azimuth, the pitch between pixels is obviously inherent to the frequency of the radar pulses and the speed of progression of the main lobe of illumination in the azimuthal direction.

If V is the standard of the relative speed of the motor vehicle with respect to the ground, that is to say its speed in the direction of the azimuth, then the pitch is given by the following relation:

V azimuth p J? F ^' In distance, it is the sampling frequency F _e which is binding. A factor 2 intervenes in the step of resolution, since the distance separating the radar sensor 20 from the target is traveled twice. In addition, it is necessary to take into account the local angle of incidence of the wave with respect to the vertical to obtain the projected resolution on the ground, given by the following relation:

Δ distance to the ground 2F _e smi

Thanks to the known principle of pulse compression, a pulse linearly modulated in frequency is sent then its correlation is made to the returned signal.

The resulting effect is that a point target produces a cardinal sinus when the transmitted pulse has a substantially constant spectral density. In addition, it is all the more narrow as the bandwidth is large. The time τ of the compressed pulse is then given by the following relation:

0.885 τ =

B

The estimating device further comprises a module for processing the complex image of the SAR type. This processing module comprises a generator 22 of a real-valued image from the SAR complex image. The real-valued image is generated simply from the point module of the SAR complex image.

In addition, the processing module implements means 24 for processing the real-valued image obtained. These processing means 24 comprise, for example, a contrast enhancer of the pixels of the real-valued image, a filter performing two-dimensional smoothing of the pixels of the real-valued image, means for associating with each pixel of the real-valued image of a weighting parameter as a function of a measured variation between values of adjacent pixels, and means for assigning a zero intensity to all the pixels of the image to actual values whose intensity is below a predetermined threshold value.

Le capteur 20 est par exemple implanté derrière le pied milieu du véhicule automobile 10, c'est-à-dire le montant séparant la place avant du conducteur de la place arrière correspondante. Cette implantation permet d'avoir un lobe d'antenne du capteur radar 20 correctement orienté pour l'application envisagée.The real-valued image thus processed by the means 24 is supplied to a computer 26 for extracting a value of length and a depth value of the parking space 12 from this real-valued image.

The sensor 20 is for example implanted behind the center pillar of the motor vehicle 10, that is to say the amount separating the front seat of the driver from the corresponding rear seat. This implantation makes it possible to have an antenna lobe of the radar sensor 20 correctly oriented for the intended application.

The azimuth resolution of such a radar sensor is given by the following equation:

0.2213 c

V azimuth sin (Δ0) / ₀

where / o is the transmission frequency of the radar sensor, and c is the speed of light in the air. In addition, for a displacement of ten meters along the parking space

12: Δθ = 2 arctan (-) = 2.0608 rad.

This makes it possible to obtain a value of δ _azιmu less than 5 cm for a frequency / o greater than 1.5 GHz. In addition, the higher the radar emission band, the better the accuracy in distance. Preferably, a SAR-type complex image is obtained by means of a radar sensor comprising five different angles of analysis which possibly make it possible to detect the best angle for visualizing the parking space 12.

dynamique élevée permet de voir l'ensemble des points détectés alors que la dynamique basse a un seuil de saturation plus faible pour ne voir que les points nettement détectés. Chaque angle de vue est obtenu par filtrage passe-bande du lobe total. Les cinq angles de vues sont ensuite superposés, c'est-à-dire sommés dans un même repère en amplitude et en phase, ce qui permet d'obtenir une meilleure détection des objets isotropes. Cette opération permet également de baisser le niveau de bruit étant donné que celui-ci n'est pas isotrope par définition.

High dynamic range makes it possible to see all the points detected while the low dynamic range has a lower saturation threshold to see only the points clearly detected. Each angle of view is obtained by bandpass filtering of the total lobe. The five angles of view are then superimposed, that is to say, summed in the same reference frame in amplitude and in phase, which makes it possible to obtain a better detection of the isotropic objects. This operation also lowers the noise level since it is not isotropic by definition.

The method implemented by the estimation device described above will now be detailed with reference to FIG. 2.

During a first transmission step 100, the radar sensor 20 transmits a signal laterally when the vehicle approaches the parking space 12. At the next reception step 102, the radar sensor 20 receives an echo of the signal issued.

These two steps 100 and 102 are repeated as many times as necessary at a predetermined frequency, while the motor vehicle 10 runs along the parking space 12 in the direction D, so as to obtain complete information on the parking space 12.

Then, when the entire parking space 12 has been scanned by the radar sensor 20, a step 104 is taken, in which a SAR-type complex image is generated, according to a known method, from the data collected by the sensor. radar 20. This generation is performed by the sensor 20 which does not provide one but two images, an amplitude image and a phase image, and possibly another image if there is further polarization information.

The principle of the processing carried out to obtain the complex image of the SAR type is known and will not be detailed. In a next step 106, a real image is generated from the SAR complex image.

Indeed, radiometry is naturally appreciable information, so that the image of the modules is interesting for further processing. In addition, the phase image can be used to extract additional information.

Then, during a step 108, the image is enhanced with real values taking into account V _mn , the minimum value taken by the pixels of this image, and V _max , the maximum value taken by the pixels of this image.

The enhancement is to update the initial value V _ιnrtιa ι _e of each pixel as follows: Initial V -V mm

V _n l, news -

V max -V min

Then, in a next step 1 10, this enhanced image is filtered.

The filtering consists of smoothing the values of this image in both directions, one of the directions corresponding to the depth of the parking space

12, the other direction corresponding to the length of the parking space 12. This smoothing can be done in several ways. One of the possibilities is to browse a line (or a column) and calculate the average value of two adjacent pixels and then recalculate the average of the adjacent averages which gives a smoothed value of the pixels of the row (or column) .

Furthermore, advantageously, a comparison of the variation between the values of the adjacent pixels can be performed to associate with each pixel a weighting parameter indicating the degree of reliability of the detection. Thus, for example, a large variation on three adjacent pixels indicates that the detection is not necessarily very good on this part of the image.

Optionally also, it is possible to assign a zero intensity to all the pixels of value less than a predetermined threshold, for example set at 0.1. Two separate areas appear on the resulting image. A black zone for the values equal to 0 and a gray zone for the values between 0.1 and 1. The gray zone advantageously also comprises a weighting for each pixel.

The filtering carried out is based on the following hypothesis: by traversing the image in the direction of the length of the parking space, if an area is clear at a distance less than 20 cm from another clear zone, then these two zones belong to the same obstacle. If two vehicles are parked 20cm from each other then anyway, it is impossible to park and therefore this case is not a problem for the desired application. In addition, the smoothing makes it possible to erase some irregularities, especially in the vacuum zone between the two parked vehicles.

Finally, in two steps 1 12 and 1 14, the length and the depth of the parking space 12 are calculated from the filtered image obtained in step 1 10.

The filtering made it possible to obtain an image comprising gray pixels and black pixels. This distinction makes it possible to calculate, during step 1 12, the distance separating the obstacles in the direction parallel to the movement of the vehicle. This distance is the length of the square obtained with very good precision. In the same way, the depth can be calculated during step 1 14 after having covered the different obstacles. Once the obstacles are encompassed, there remains the empty place which is either delimited in depth by an obstacle

(sidewalk type, ...), or not delimited in depth. Both cases allow to know the depth of the free place.

FIG. 3 represents a real-valued image, as obtained at the end of step 106.

FIG. 4 represents this same image after the filtering step 1, from which it is easy to extract the length and the depth of the parking space 12.

Note that for the invention to be feasible, it is assumed that the speed of the vehicle, when it runs along the parking space 12, is less than 20km / h.

Additional services can then be envisaged by the system: - steering indication so that the driver can park; indication of a step to achieve a niche; automatic parking of the vehicle, the engine, brakes and steering being controlled by the system; and semi-automatic parking, when the steering is piloted. It is clear that a device as described above makes it possible to accurately estimate the dimensions of a parking space next to which a motor vehicle moves.

Claims

1. Device for estimating the dimensions of a parking space (12) intended to be loaded into a motor vehicle (10), characterized in that it comprises: a synthetic aperture radar sensor (20) for the generation of a complex image of the SAR type of this parking space, a module (22, 24) for processing this SAR complex image so as to obtain a real-valued image, and - A calculator (26) for extracting a length value and a depth value of the parking space from this real-valued image. The parking space dimension estimator of claim 1, wherein the aperture synthesis radar sensor (20) has a plurality of different angles of analysis. 3. Motor vehicle comprising an estimation device according to claim 1 or 2. 4. Motor vehicle according to claim 3, wherein the synthetic aperture radar sensor (20) is fixed on a lateral outer portion of the motor vehicle (10). 5. A method for estimating the dimensions of a parking space (12) using a device embedded in a motor vehicle (10), comprising the following steps: generating (104) a complex SAR-type image of this parking space by a synthetic aperture radar sensor (20), - processing (106, 108, 1 10) of this complex SAR-type image so that to obtain an image with real values, and extraction of a length value and a depth value of the parking space from this real-valued image. 6. A method of estimating the dimensions of a parking space according to claim 22, wherein the step of processing the SAR complex image comprises generating (106) the real-valued image. from the point module of the SAR complex image and a contrast enhancement (108) of the pixels of the real-valued image. The method of estimating the dimensions of a parking space according to claim 5 or 6, wherein the step of processing the SAR-type image. comprises filtering (1 10) the real-valued image, this filter performing a two-dimensional smoothing of its pixels. The method for estimating the dimensions of a parking space according to any one of claims 5 to 7, wherein the step of processing the SAR-type image comprises the association with each pixel of the image. with real values of a weighting parameter as a function of a measured variation between values of adjacent pixels. 9. A method of estimating the dimensions of a parking space according to any one of claims 5 to 8, wherein during the step of processing the SAR-type image is assigned zero intensity to all pixels the real-valued image whose intensity is less than a predetermined threshold value.