Method for classifying moving vehicles
A method for classifying vehicles in which an angle-resolving radar device yields measurement signals which have frequencies corresponding to a Doppler shift and which originate from measured vehicles and from which radial distances, object angles and radial velocities can be derived. The frequencies of the acquired measurement signals are stored as functions over the measurement time period, and a spectrogram is formed for every vehicle therefrom. Subsequently, the spectrograms are checked for assessment regions with maximum bandwidth of the frequency. These assessment regions are compared with assessment regions of stored spectrograms for different vehicle classes and associated with the most similar such that the measured vehicles are classified.
1. Method for classifying vehicles in which vehicles passing through a radar cone formed by radar radiation are measured, said vehicles having surfaces on which radar radiation impinges thus reflecting said radar radiation, and wherein measurement signals with frequencies corresponding to a Doppler shift are generated, said method comprising:
a) arranging, at a vertical installation height relative to a surface of a roadway and at a horizontal acute installation angle relative to an edge of the roadway, an angle-resolving radar device for yielding measurement signals which have frequencies corresponding to the Doppler shift and from which radial distances, object angles and radial velocities can be derived,
b) emitting a radar beam from said radar device forming a radar cone, and acquiring measurement signals from at least one vehicle traveling through the radar cone at a plurality of measuring times during a measurement time period,
c) storing the frequencies of the acquired measurement signals of measured vehicles as functions over the measurement time period, and forming a spectrogram for every measured vehicle,
d) checking the spectrograms for assessment regions with maximum bandwidth of the frequency, and at least one measuring time is determined in the assessment region as an assessment time,
e) comparing said assessment regions with assessment regions of stored spectrograms for different vehicle classes, and assigning the most similar spectrogram in order to classify the vehicles traveling through said cone, and
f) deriving object angles from the measurement signals at the at least one assessment time within the assessment regions with maximum bandwidth, wherein the respective derived object angles for the respective assessment time are stored in each instance as a function over the frequency, and the curve of the function is compared with comparison models stored for different vehicle classes in order to verify the classification of vehicles.
2. Method according to claim 1 , further comprising modeling of a size of the at least one vehicle from the comparison of all formed functions with stored comparison models.
3. Method according to claim 1 , wherein a probability distribution is determined for different vehicle sizes when forming each function, and these probabilities are accumulated in a weighted manner over the entire time curve.
4. Method for classifying vehicles in which vehicles passing through a radar cone formed by radar radiation are measured, said vehicles having surfaces on which radar radiation impinges and reflects the radar radiation, and wherein measurement signals with frequencies corresponding to a Doppler shift are generated, said method comprising the following steps:
a) arranging, at a vertical installation height relative to a surface of a roadway and at a horizontal acute installation angle relative to an edge of the roadway, an angle-resolving radar device producing measurement signals which have frequencies corresponding to the Doppler shift and from which radial distances, object angles and radial velocities can be derived,
b) emitting a radar beam from said radar device forming a radar cone, and acquiring measurement signals from at least one vehicle traveling through the radar cone at a plurality of measuring times during a measurement time period,
c) storing the frequencies of the acquired measurement signals of measured vehicles as functions over the measurement time period, and forming a spectrogram for every measured vehicle,
d) checking the spectrograms for assessment regions with maximum bandwidth of the frequency, and at least one measuring time is determined in the assessment region as an assessment time,
e) comparing said assessment regions with assessment regions of stored spectrograms for different vehicle classes, and an assignment is made to the most similar spectrogram in order to classify the vehicles traveling through said cone, and
f) deriving radial distances from the measurement signals at the at least one assessment time within the assessment regions with maximum bandwidth, wherein the respective derived radial distances for the respective assessment time are stored in each instance as a function over the frequency, and the curve of the functions is compared with comparison models stored for different vehicle classes in order to verify the classification of vehicles.
5. Method according to claim 4 , further comprising modeling of a size of the at least one vehicle from the comparison of all formed functions with stored comparison models.
6. Method for classifying vehicles in which vehicles passing through a radar cone formed by radar radiation are measured, said vehicles having surfaces on which radar radiation impinges and reflects the radar radiation, and wherein measurement signals with frequencies corresponding to the Doppler shift are generated, said method comprising:
a) arranging, at a vertical installation height relative to a surface of a roadway and at a horizontal acute installation angle relative to an edge of the roadway, an angle-resolving radar device yielding measurement signals which have frequencies corresponding to the Doppler shift and from which radial distances, object angles and radial velocities can be derived,
b) emitting a radar beam from said radar device forming a radar cone, and acquiring measurement signals from at least one vehicle traveling through the radar cone at a plurality of measuring times during a measurement time period,
c) storing the frequencies of the acquired measurement signals of measured vehicles are stored as functions over the measurement time period, and forming a spectrogram for every measured vehicle,
d) checking the spectrograms for assessment regions with maximum bandwidth of the frequency, and at least one measuring time is determined in the assessment region as an assessment time,
e) comparing said assessment regions with assessment regions of stored spectrograms for different vehicle classes, and an assignment is made to the most similar spectrogram in order to classify the vehicles traveling through said cone, and
f) deriving signal amplitudes of the measurement signals from the measurement signals at the at least one assessment time within the assessment regions with maximum bandwidth, wherein the signal amplitudes of the measurement signals for the respective assessment time are stored in each instance as a function over the frequency, and the curve of the functions is compared with comparison models stored for different vehicle classes in order to verify the classification of vehicles.
7. Method according to claim 6 , further comprising modeling of a size of the at least one vehicle from the comparison of all formed functions with stored comparison models.
8. Method according to claim 6 , wherein a probability distribution is determined for different vehicle sizes when forming each function, and these probabilities are accumulated in a weighted manner over the entire time curve.