Identity recognition system based on compressed signals and method thereof
An identity recognition system based on compressed signals and a method thereof are provided. When a sensing end is in an identification mode, it continuously measures a physiological identification signal of a user having at least one first predetermined length of time, and performs a compression process on the physiological identification signal having at least one first predetermined length of time to generate a first compressed signal. An identification end receives the first compressed signal, obtains first identification information of the first compressed signal in a discrimination subspace according to a principal eigenvector, and recognizes the first identification signal according to a classification model from a learning end to determine that the user is one of multiple subjects. Therefore, the sensing end uses compressive technology to reduce the energy required to transmit signals, and the identification end recognizes identity without reconstructing the first compressed signal transmitted by the sensing end.
1. An identity recognition system based on compressed signals, comprising:
a sensing wearable device, having an identification mode and a learning mode and including a measurement module, a compression module and a first transmission module, wherein when the sensing wearable device is in the identification mode, the measurement module is configured to continuously measure a physiological identification signal of a user having at least one first predetermined length of time, and the compression module is configured to perform a compression process on the physiological identification signal having at least one first predetermined length of time to generate a first compressed signal, and the first transmission module is configured to transmit the first compressed signal; and
an identification computer device, including a first signal separation module and an identification module, wherein the first signal separation module is configured to receive the first compressed signal, and obtain first identification information of the first compressed signal in a discrimination subspace according to a principal eigenvector, and the identification module is configured to receive the first identification information, and recognize the first identification information according to a classification model from a learning computer device, to determine that the user is one of multiple subjects without reconstruction of the first compressed signal transmitted by the measurement module, when the sensing wearable device is in the learning mode, the measurement module is configured to continuously measure a physiological identification signal of each subject having at least one second predetermined length of time, and the compression module is configured to perform the compression process on the physiological identification signal of each subject having at least one second predetermined length of time to generate a second compressed signal, and the first transmission module is configured to transmit the second compressed signal corresponding to each subject, and the learning computer device comprises a second signal separation module, a machine learning module and a second transmission module, wherein the second signal separation module is configured to receive the second compressed signal corresponding to each subject, and obtain second identification information of the second compressed signal in the discrimination subspace according to the principal eigenvector, and the machine learning module is configured to train the classification module according to the second identification information through a machine learning algorithm, and the second transmission module is configured to transmit the classification model to the identification computer device, wherein the compression process on the physiological identification signal having at least one first predetermined length of time is performed by using discrete cosine transform (DCT), discrete wavelet transformation (DWT), principal component analysis (PCA) or compressive sensing (CS) technology.
2. The identity recognition system according to claim 1 , wherein the sensing wearable device further includes a setting module for receiving setting information corresponding to each subject when the sensing wearable device is in the learning mode, to make the first transmission module transmit the second compressed signal and the setting information corresponding to each subject to the learning computer device.
3. The identity recognition system according to claim 1 , wherein the sensing wearable device further includes a noise processing module for removing noise from the physiological identification signal having at least one first predetermined length of time before the compression module performs the compression process on the physiological identification signal having at least one first predetermined length of time.
4. An identity recognition method based on compressed signals, comprising:
when being in a identification mode, a sensing wearable device continuously measuring a physiological identification signal of a user having at least one first predetermined length of time, and performing a compression process on the physiological identification signal of the user having at least one first predetermined length of time to generate a first compressed signal, and transmitting the first compressed signal; and an identification computer device receiving the first compressed signal, and obtaining first identification information of the first compressed signal in a discrimination subspace according to a principal eigenvector, and receiving the first identification information, and recognizing the first identification information according to a classification model from a learning computer device, to determine that the user is one of multiple subjects without reconstruction of the first compressed signal transmitted by he measurement module, the sensing wearable device, when being in a learning mode, continuously measuring a physiological identification signal of each subject having at least one second predetermined length of time, and performing the compression process on the physiological identification signal of each subject having at least one second predetermined length of time to generate a second compressed signal, and transmitting the second compressed signal corresponding to each subject, and
the learning computer device receiving the second compressed signal corresponding to each subject, and obtaining second identification information of the second compressed signal in the discrimination subspace according to the principal eigenvector, and train the classification model according to the second identification information through a machine learning algorithm, and transmitting the classification model to the identification computer device, wherein the compression process on the physiological identification signal having at least one first predetermined length of time is performed by using DCT, DWT, PCA or CS technology.
5. The identity recognition method according to claim 4 , wherein further comprising:
the sensing wearable device receiving setting information corresponding to each subject when the sensing wearable device is in the learning mode, to transmit the second compressed signal and the setting information corresponding to each subject to the learning computer device.
6. The identity recognition method according to claim 4 , wherein before the step of performing the compression process on the physiological identification signal having at least one first predetermined length of time, further comprising: removing noise from the physiological identification signal having at least one first predetermined length of time.