Traffic monitoring apparatus, system, traffic monitoring method and non-transitory computer readable medium
A traffic monitoring apparatus includes: at least one memory storing instructions; and at least one processor. The processor is configured to execute the instructions to; acquire waterfall data from a distributed acoustic sensor (DAS), wherein the waterfall data includes a generation position of a vibration on a roadway adjacent to the DAS, a generation time of the vibration and an amplitude of the vibration; preprocess the waterfall data; estimate at least one enhancement of the processed waterfall data, wherein an enhancement corresponds to a traffic flow property; and estimate at least one traffic flow property of the roadway from the enhancements of the processed waterfall data.
1. A traffic monitoring apparatus comprising:
at least one memory storing instructions; and
at least one processor configured to execute the instructions to:
acquire waterfall data from a distributed acoustic sensor (DAS), wherein the waterfall data includes a generation position of a vibration on a roadway adjacent to the DAS, a generation time of the vibration and an amplitude of the vibration;
preprocess the waterfall data;
iteratively estimate at least one enhancement of the preprocessed waterfall data; and
estimate at least one traffic flow property of the roadway from the enhancements of the preprocessed waterfall data;
wherein iteratively estimate at least one enhancement of the preprocessed waterfall data includes performing a masking operation on a portion of the preprocessed waterfall data in an iteration using an enhancement estimated in a preceding iteration until a satisfactory enhancement is achieved,
wherein the masking operation includes
removing a portion of the preprocessed waterfall data corresponding to a portion of an estimated enhancement and amplifying the remaining preprocessed waterfall data,
dampening the preprocessed waterfall data corresponding to an estimated enhancement, or
replacing a portion of the preprocessed waterfall data with a corresponding portion of an estimated enhancement and amplifying the remaining preprocessed waterfall data.
2. The traffic monitoring apparatus of claim 1 , wherein the processor is further configured to execute the instructions to estimate an enhancement using a pre-trained model.
3. The traffic monitoring apparatus of claim 1 , wherein each enhancement is estimated within a predetermined time range and within a predetermined position range.
4. The traffic monitoring apparatus of claim 1 , wherein the traffic flow property corresponding to an enhancement is the trajectory of each vehicle on the roadway.
5. The traffic monitoring apparatus of claim 1 , wherein the at least one traffic flow property is the size of each vehicle on the roadway and/or the occupancy of the roadway and/or the damage to the roadway.
6. The traffic monitoring apparatus of claim 2 , wherein the pre-trained model is generated by training a portion of the preprocessed waterfall data and at least one enhancement of the preprocessed waterfall data used as labels, wherein an enhanced waterfall data is obtained from a secondary acquisition means.
7. The traffic monitoring apparatus of claim 2 , wherein the pre-trained model is generated by training synthetically emulated waterfall data and at least one synthetically emulated enhancement of emulated waterfall data used as labels.
8. The traffic monitoring apparatus of claim 6 , wherein the secondary acquisition means comprises a video and/or an inductive loop based traffic monitoring system.
9. The traffic monitoring apparatus of claim 6 , wherein the secondary acquisition means comprises an electronic toll collection system.
10. A system comprising:
a distributed acoustic sensor (DAS);
an optical fiber connected to the DAS and provided along the roadway; and
the traffic monitoring apparatus of claim 1 .
11. A traffic monitoring method comprising:
acquiring the waterfall data from a distributed acoustic sensor (DAS), wherein the waterfall data includes a generation position of a vibration on a roadway adjacent to the DAS, a generation time of the vibration and an amplitude of the vibration;
preprocessing the waterfall data;
iteratively estimating at least one enhancement of the preprocessed waterfall data; and
estimating at least one traffic flow property of the roadway from the enhancements of the preprocessed waterfall data;
wherein iteratively estimating at least one enhancement of the preprocessed waterfall data includes performing a masking operation on a portion of the preprocessed waterfall data in an iteration using an enhancement estimated in a preceding iteration until a satisfactory enhancement is achieved,
wherein the masking operation includes
removing a portion of the preprocessed waterfall data corresponding to a portion of an estimated enhancement and amplifying the remaining preprocessed waterfall data,
dampening the preprocessed waterfall data corresponding to an estimated enhancement, or
replacing a portion of the preprocessed waterfall data with a corresponding portion of an estimated enhancement and amplifying the remaining preprocessed waterfall data.
12. A non-transitory computer readable medium storing a program for causing a computer to execute a traffic monitoring method, the method comprising:
acquiring the waterfall data from a distributed acoustic sensor (DAS), wherein the waterfall data includes a generation position of a vibration on a roadway adjacent to the DAS, a generation time of the vibration and an amplitude of the vibration;
preprocessing the waterfall data;
iteratively estimating at least one enhancement of the preprocessed waterfall data; and
estimating at least one traffic flow property of the roadway from the enhancements of the preprocessed waterfall data;
wherein iteratively estimating at least one enhancement of the preprocessed waterfall data includes performing a masking operation on a portion of the preprocessed waterfall data in an iteration using an enhancement estimated in a preceding iteration until a satisfactory enhancement is achieved,
wherein the masking operation includes
removing a portion of the preprocessed waterfall data corresponding to a portion of an estimated enhancement and amplifying the remaining preprocessed waterfall data,
dampening the preprocessed waterfall data corresponding to an estimated enhancement, or
replacing a portion of the preprocessed waterfall data with a corresponding portion of an estimated enhancement and amplifying the remaining preprocessed waterfall data.