FLIGHT TIME SENSOR AND DISTANCE MEASUREMENT METHOD

Description

[0001] Embodiments and implementations relate to time-of-flight sensors.

[0002] A time-of-flight sensor is a sensor used to measure distances between this sensor and elements of a scene in which this sensor is placed. The time-of-flight sensor is configured to emit light rays into the scene. These light rays can be reflected by the different elements of the scene. Furthermore, the time-of-flight sensor can include a matrix of photosensitive pixels. Each photosensitive pixel of the matrix is ​​configured to detect light signals incident on this pixel. Each pixel thus makes it possible to detect the light rays emitted by the sensor and then reflected by the different elements. The detection of rays reflected by a pixel of the matrix allows the time-of-flight sensor to calculate a distance between the sensor and the elements of the scene having reflected the light rays.The time-of-flight sensor then generates an image of the scene defining, for each pixel of the image, a distance between the sensor and an element of the scene.

[0003] US patent applications published under numbers US 2021 / 209778 A1 and US 2019 / 369247 illustrate state-of-the-art time-of-flight sensors.

[0004] Some time-of-flight sensors use a point projector. The image generated by the time-of-flight sensor comprises a plurality of points, each point covering several pixels. These points correspond to the pixels in the matrix detecting the rays emitted by the sensor that are reflected by elements in the scene. Such time-of-flight sensors reduce disturbances related to ambient light in the scene. The local concentration of light at each point improves distance determination performance.

[0005] It is known to calculate an average value of the distances associated with the pixels of the same point. This average value of the distances is then associated with this point of the image.

[0006] It is possible for a point in the image to include pixels providing distance values ​​for different elements in the scene arranged at different distances from the sensor. In this case, the sensor associates with the point an average value of the calculated distances which may correspond to a distance for which it is possible that no element is actually present in the scene. The value associated with this point is therefore erroneous. Such a time-of-flight sensor is therefore relatively unreliable under certain conditions and does not allow multiple elements to be detected at the same point in the image.

[0007] Therefore, there is a need to provide a reliable time-of-flight sensor under the above-mentioned conditions.

[0008] According to one aspect, there is provided a time-of-flight sensor comprising: an emitting device configured to emit light rays, into a scene in which the time-of-flight sensor is placed, a matrix of photosensitive pixels configured to receive light signals from the scene and to generate an image illustrating points associated with the rays emitted by the emitting device, reflected by elements of the scene and received by the matrix of pixels, each point covering a plurality of pixels of the image, each pixel of the image being associated with a distance value measured between the time-of-flight sensor and an element of the scene, the time-of-flight sensor being characterized in that it comprises a processing unit configured to, for each point of the image: o partitioning the pixels of this point into at least one group of pixels, o calculating, for each group of pixels, a value representative of the distance values ​​of the pixels of this group, this representative distance value then being applied for each pixel of the group of pixels.

[0009] Such a time-of-flight sensor makes it possible to isolate image points from others, and then define groups of pixels for each image point based on their distance values. The groups of an image point can then be associated with the presence of different elements in the scene.

[0010] Thus, such a time-of-flight sensor makes it possible to detect several elements of the scene located at the same point of the image, and to return, for this point of the image, the same number of distance values ​​as elements at this point.

[0011] Such a time-of-flight sensor also makes it possible to limit errors in distance values ​​associated with certain pixels of a point in the image. Indeed, grouping pixels with distance values ​​close to each other can make it possible to extract pixels with aberrant distance values. By not taking into account pixels with aberrant distance values, the signal-to-noise ratio of such a sensor is improved.

[0012] Such a time-of-flight sensor is therefore reliable and precise.

[0013] In an advantageous embodiment, the processing unit is also configured to merge groups of pixels of the same point if the difference in the representative values ​​calculated for these groups of pixels is less than a given threshold, then to calculate a new representative distance value for the new group of pixels formed by the merging of these groups. In this way, the processing unit makes it possible to avoid having several groups associated with the presence of the same element of the scene.

[0014] Preferably, the partitioning of the pixels of each point comprises an implementation of a segmentation algorithm, in particular a k-means partitioning algorithm.

[0015] According to another aspect, there is provided a method of measuring distances between a time-of-flight sensor and elements of a scene in which the time-of-flight sensor is placed, the method comprising: an emission of light points in the scene by an emitting device of the time-of-flight sensor, a reception by a pixel matrix of the time-of-flight sensor of the light signals coming from the scene and a generation by the pixel matrix of an image illustrating points associated with the rays emitted by the emitting device, reflected by elements of the scene and received by the pixel matrix, each point covering a plurality of pixels of the image, each pixel of the image being associated with a distance value measured between the time-of-flight sensor and an element of the scene, the method being characterized in that it comprises, for each point of the image: a partitioning of the pixels of this point into at least one group of pixels, a calculation for each group of pixels, of a value representative of the distance values ​​of the pixels of this group, this representative distance value then being applied for each pixel of the group of pixels.

[0016] Advantageously, the method further comprises a merging of the groups of the same point if the difference in the representative values ​​calculated for these groups of pixels is less than a given threshold, then a calculation of a new representative distance value for the new group formed by the merging of these groups.

[0017] Preferably, the partitioning of the pixels of each point comprises an implementation of a segmentation algorithm, in particular a k-means partitioning algorithm.

[0018] Other advantages and characteristics of the invention will appear on examining the detailed description of embodiments, which are in no way limiting, and the appended drawings in which: [ Fig 1 ] [ Fig 2 ] [ Fig 3 ] [ Fig 4 ] [ Fig 5 ] [ Fig 6 ] illustrate embodiments and implementations of the invention.

[0019] There figure 1 illustrates a TOF time-of-flight sensor. This TOF time-of-flight sensor may be an indirect time-of-flight sensor (also known by the acronym "iTOF"). The TOF time-of-flight sensor comprises an EMT transmitter device and a PXM pixel array. The TOF time-of-flight sensor also comprises a UT control and processing unit configured to control the EMT transmitter and the PXM pixel array.

[0020] The EMT emitter device is configured to emit light rays into a scene in which the TOF sensor is placed. These light rays are then reflected by the various elements of the scene towards the PXM pixel matrix.

[0021] The PXM pixel array is configured to detect the light signals it receives.

[0022] The PXM pixel matrix is ​​thus configured to generate IMG images of the scene from the light signals it detects. figure 2 illustrates an example of an IMG image that can be generated by the PXM pixel matrix. The image is composed of PX pixels associated with the different pixels of the PXM pixel matrix.

[0023] Each pixel PX of the IMG image is associated with a distance value measured between the time-of-flight sensor CPT and an element of the scene.

[0024] Each IMG image has a set of DT points corresponding to the different rays reflected by the elements of the scene and captured by the PXM pixel matrix. Each DT point of the IMG image covers a set of PX pixels of the image. For example, a DT point of the IMG image can cover a set of 10x10 pixels.

[0025] There figure 3 illustrates the pixel distance values ​​of a DT point in the image.

[0026] The processing unit UT is configured to process each image IMG generated by the pixel matrix PXM. In particular, the processing unit UT is configured to perform a partitioning of the pixels PX for each point DT of the image IMG. Thus, for each point DT of the image IMG, the processing unit UT is configured to define at least one group of pixels PX associated with this point DT.

[0027] In particular, for each point DT of the image IMG, the processing unit UT is configured to group the pixels PX of the point DT having distance values ​​that are close to each other. This makes it possible to identify different elements of the scene represented on the same point of the image. For example, pixels PX can be considered as close enough to be grouped in the same group if the difference between their distance values ​​is less than 2% of the average of the two distances (this value of 2% being a parameter of the algorithm that can be adjusted according to the performance of the time-of-flight sensor).

[0028] Of course, different partitioning algorithms can be implemented to define the groups of pixels PX of a point DT of the image IMG. For example, the groups can be defined by a k-means partitioning algorithm. Alternatively, it is possible to define the groups by an analysis of histograms of the distance values ​​of the pixels PX of the point DT of the image so as to detect local maxima to identify the different elements of the scene represented on the point DT.

[0029] There figure 4 illustrates the DT point of the figure 3 in which pixels with close distance values ​​were grouped into groups GP1, GP2, GP3, GP4.

[0030] For each point DT of the image, the processing unit is configured to calculate for each group a value representative of the pixel distance values ​​PX of this group. This value representative of a group can be an average value of the pixel distance values ​​PX of this group.

[0031] If two groups of the same DT point have representative values ​​close to each other, the processing unit can be configured to merge these two groups. For example, representative values ​​can be considered close when their difference is less than 2% of the average of the two distances (this 2% value being a parameter of the algorithm that can be adjusted according to the performance of the time-of-flight sensor). The processing unit is then configured to calculate a new representative value for the new group formed resulting from the merger of the two groups. For example, the figure 5 illustrates the point of the figure 3 in which the GP3 and GP4 groups of pixels have been merged to form a new NGP3 group of pixels.

[0032] The processing unit thus generates a final image in which the DT points of the initial image have as distance values ​​the representative values ​​calculated for each group defined for each DT point.

[0033] Such a time-of-flight sensor makes it possible to detect several elements of the scene located on the same DT point of the image, and to return, for this DT point of the image, the same number of distance values ​​as elements on this DT point.

[0034] Such a time-of-flight sensor also makes it possible to limit the errors in distance values ​​associated with certain PX pixels of a DT point in the image. Indeed, grouping the PX pixels having distance values ​​close to each other can make it possible to extract the PX pixels having aberrant distance values. By not taking into account the PX pixels having aberrant distance values, the signal-to-noise ratio of such a sensor is improved.

[0035] There figure 6 illustrates a method of processing an image generated by a PXM pixel array of a time-of-flight sensor as described previously. As seen previously, the image generated by the PXM pixel array comprises several DT points.

[0036] The method comprises an implementation 40 of an algorithm for partitioning the pixels PX of each point DT. The implementation of such an algorithm makes it possible to define groups of pixels PX in each point DT of the image.

[0037] The method also comprises, for each group of each point DT, a calculation of a value representative of the distance values ​​of the pixels PX of this group.

[0038] The method may also comprise a merging 41 of the groups of the same point DT having representative values ​​close to each other. The method then comprises, following the merging of the groups, a calculation 42 of a new representative value for the new group formed resulting from the merging of these groups.

Claims

1. A time-of-flight sensor comprising: - an emitting device (EMT) configured to emit light rays, in a scene in which the time-of-flight sensor is placed, - a photosensitive pixel matrix (PXM) configured to receive light signals from the scene and to generate an image (IMG) illustrating dots (DT) associated with the rays emitted by the emitting device (EMT), reflected by elements of the scene and received by the pixel matrix, each dot (DT) covering a plurality of pixels (PX) of the image, each pixel of the image being associated with a distance value measured between the time-of-flight sensor and an element of the scene, the time-of-flight sensor being characterised in that it also comprises a processing unit (UT) configured to, for each dot of the image: o partition the pixels (PX) of this dot into at least one group (GP1, GP2, GP3, GP4) of pixels (PX), o calculate for each group (GP1, GP2, GP3, GP4) of pixels (PX), a value representative of the distance values of the pixels of this group, this representative distance value being then applied for each pixel of the group of pixels.

2. The time-of-flight sensor according to claim 1, wherein the processing unit (UT) is also configured to merge groups (GP3, GP4) of pixels (PX) of a same dot (DT) if the difference of the representative values calculated for these pixel groups (PX) is less than a given threshold, then to calculate a new representative distance value for the new group (NGP3) of pixels (PX) formed by merging these groups.

3. The time-of-flight sensor according to any of claims 1 or 2, wherein partitioning the pixels (PX) of each dot includes implementing a segmentation algorithm.

4. The time-of-flight sensor of claim 3, wherein the segmentation algorithm is a k-mean partitioning algorithm.

5. A method for measuring distances between a time-of-flight sensor and elements of a scene in which the time-of-flight (TOF) sensor is placed, the method comprising: - °emitting light rays in the scene by an emitting device (EMT) of the time-of-flight (TOF) sensor, - °receiving, by a pixel matrix (PXM) of the time-of-flight sensor (CPT), the light signals coming from the scene and - generating by the pixel matrix (PX) an image illustrating dots (DT) associated with the rays emitted by the emitting device (EMT), reflected by elements of the scene and received by the pixel matrix, each dot (DT) covering a plurality of pixels (PX) of the image, each pixel of the image being associated with a distance value measured between the time-of-flight sensor and an element of the scene, the method being characterised in that it comprises, for each dot (DT) of the image: - partitioning the pixels (PX) of this dot into at least one group (GP1, GP2, GP3, GP4) of pixels (PX), - calculating for each group (GP1, GP2, GP3, GP4) of pixels (PX), a value representative of the distance values of the pixels of this group, this representative distance value being then applied for each pixel of the group of pixels.

6. The method according to claim 5, further comprising merging the groups (GP3, GP4) of a same dot (DT) if the difference of the representative values calculated for these groups of pixels (PX) is less than a given threshold, then calculating a new representative distance value for the new group (NGP3) formed by merging these groups (GP3, GP4).

7. The method according to any of claims 5 or 6, wherein partitioning the pixels (PX) of each dot (DT) includes implementing a segmentation algorithm.

8. The method according to claim 7, wherein the segmentation algorithm is a k-mean partitioning algorithm.