Method and device for classifying targets
A method for classifying targets is proposed, which comprises the extraction of features from measurement data of one or several receiving elements of a sensor by means of a neuronal network or by means of a Gaussian Mixture Model, wherein the respective measurement data of the at least one receiving element of the sensor involve at least one section of a photon histogram, and wherein the neuronal network involves a fully connected neuronal network or a convolutional neuronal network.
1 . A method for classifying targets, comprising:
obtaining measurement data from at least one receiving element of a sensor;
generating a photon histogram as a one-dimensional array from the measurement data;
converting the photon histogram into a two-dimensional image representation;
extracting features from the two-dimensional image representation of the photon histogram by means of a convolutional neuronal network or a Gaussian Mixture Model to obtain extracted features; and
classifying the targets based at least in part on the extracted features, wherein the extracted features comprise a shape of a peak in the two-dimensional image representation of the photon histogram, and wherein the classifying comprises distinguishing between object reflections and environmental condition reflections based on the extracted features.
2 . The method according to claim 1 , wherein the extracted features further comprise a height of the peak in the two-dimensional image representation of the photon histogram.
3 . The method according to claim 1 ,
wherein the sensor is a solid-state Lidar sensor, wherein the method comprises a performance of light transit time measurements for at least one defined measuring interval by means of the sensor, and
wherein the generating of the photon histogram comprises generating the photon histogram per measuring interval based upon the light transit time measurements.
4 . The method according to claim 1 ,
wherein the method comprises a provision of the measurement data for the neuronal network or the Gaussian Mixture Model,
wherein only at least one section of the photon histogram is provided, and wherein the section comprises precisely one peak.
5 . The method according to claim 1 , further comprising determining and allocating a distance to each of the targets based on a position of the peak in the two-dimensional image representation of the photon histogram.
6 . The method according to claim 1 , wherein the photon histogram is generated for each measuring interval and for each receiving element of the sensor.
7 . The method according to claim 1 ,
wherein the classifying takes place by means of an additional neuronal network, wherein the additional neuronal network is designed as a fully connected neuronal network.
8 . The method according to claim 1 , wherein the method comprises a determination of a statement about an environmental condition.
9 . The method according to claim 8 ,
wherein the statement relates to a presence of the environmental condition and/or a type of environmental condition and/or a strength of the environmental condition.
10 . The method according to claim 1 wherein the method comprises a determination and allocation of a distance to each detected target, primarily to each detected object and/or each detected environmental condition.
11 . A device for classifying targets, comprising:
a sensor with several receiving elements;
an evaluation unit with a neuronal network, wherein the neuronal network involves a fully connected neuronal network or a convolutional neuronal network,
wherein the neuronal network is configured to extract features from measurement data of one or several receiving elements of the sensor,
wherein the measurement data of the at least one receiving element of the sensor each involve at least one section of a two-dimensional image representation of a photon histogram, wherein the neuronal network is configured to classify the targets based at least in part on the features, wherein the features comprise a shape of a peak in the two-dimensional image representation of the photon histogram, and wherein the targets are classified as either object reflections or environmental condition reflections based on the features.
12 . The device according to claim 11 , wherein the features further comprise a height of the peak in the two-dimensional image representation of the photon histogram.
13 . The device according to claim 11 , wherein the features further comprise a width of the peak in the two-dimensional image representation of the photon histogram.
14 . The device according to claim 11 , wherein the features further comprise a smoothness of the peak in the two-dimensional image representation of the photon histogram.
15 . A non-transitory computer-readable storage medium on which a program is stored that allows a computer, once it has been loaded into a memory of the computer, to implement a method, wherein the method comprises:
extracting features from measurement data of one or several receiving elements of a sensor by means of a neuronal network or by means of a Gaussian Mixture Model, wherein the respective measurement data of the at least one receiving element of the sensor involve at least one section of a two-dimensional image representation of a photon histogram,
wherein the neuronal network involves a fully connected neuronal network or a convolutional neuronal network; and
classifying targets based at least in part on the features comprise a shape of a peak in the two-dimensional image representation of the photon histogram, and wherein the classifying comprises distinguishing between object reflections and environmental condition reflections based on the features.
16 . The non-transitory computer-readable storage medium according to claim 15 , wherein the classifying takes place by means of an additional neuronal network, wherein the neuronal network is designed as a fully connected neuronal network.
17 . The non-transitory computer-readable storage medium according to claim 15 , wherein the method comprises a determination of a statement about an environmental condition.
18 . The non-transitory computer-readable storage medium according to claim 17 , wherein the statement relates to a presence of the environmental condition and/or a type of environmental condition and/or a strength of the environmental condition.
19 . The non-transitory computer-readable storage medium according to claim 15 , wherein the features further comprise a smoothness of the peak in the two-dimensional image representation of the photon histogram.
20 . The non-transitory computer-readable storage medium according to claim 15 , wherein the features further comprise a width of the peak in the two-dimensional image representation of the photon histogram.