Feature extraction apparatus and method for biometric information detection, biometric information detection apparatus, and wearable device

Abstract

A feature extraction apparatus configured to perform biometric information detection includes a bio-signal obtainer configured to acquire a bio-signal; and a processor configured to decompose a waveform of the acquired bio-signal into component pulses and extract a feature for the biometric information detection based on characteristic points of the component pulses.

What is claimed is: 
     
         1 . A feature extraction apparatus configured to perform biometric information detection, the feature extraction apparatus comprising:
 a sensor configured to acquire a bio-signal; and   a processor configured to:
 decompose a waveform of the acquired bio-signal into component pulses; 
 obtain subsequent component pulses based on new bio-signals generated based on iteratively eliminating the component pulses; and 
 extract a feature for the biometric information detection based on characteristic points of the component pulses. 
   
     
     
         2 . The feature extraction apparatus of  claim 1 , wherein the processor is configured to:
 estimate a given component pulse to be decomposed, among the component pulses, based on the waveform of the bio-signal and thereby generate an estimation result; and   generate the given component pulse based on the estimation result.   
     
     
         3 . The feature extraction apparatus of  claim 2 , wherein the processor is configured to, when the given component pulse is generated, eliminate the generated component pulse from the bio-signal, and when a number of generated component pulses is smaller than a pre-set number, estimate another subsequent component pulse based on another new bio-signal generated by eliminating the generated component pulse. 
     
     
         4 . The feature extraction apparatus of  claim 2 , wherein the processor is configured to model a component pulse waveform function based on the waveform of the acquired bio-signal and estimate parameters of the component pulse waveform function. 
     
     
         5 . The feature extraction apparatus of  claim 4 , wherein the processor is configured to model the component pulse waveform function to be bilaterally asymmetric with respect to a time axis, based on a Gaussian waveform function and the asymmetry factor. 
     
     
         6 . The feature extraction apparatus of  claim 5 , wherein the processor is configured to, when the given component pulse is estimated based on a current bio-signal and the asymmetry factor, adjust the asymmetry factor based on a new bio-signal generated by eliminating the estimated component pulse from the current bio-signal, and re-estimate the given component pulse based on the current bio-signal and the adjusted asymmetry factor. 
     
     
         7 . The feature extraction apparatus of  claim 6 , wherein the processor is configured to adjust the asymmetry factor based on a determination that a mean time period of the eliminated component pulse is smaller than a time period at a peak point. 
     
     
         8 . The feature extraction apparatus of  claim 1 , wherein the processor is configured to extract the feature for the biometric information detection based on characteristic points which include one or more of time, an amplitude, a standard deviation, and an offset of the decomposed component pulses. 
     
     
         9 . The feature extraction apparatus of  claim 1 , wherein the sensor is configured to acquire the bio-signal by emitting light onto a user's skin and detecting light returning from the user's skin. 
     
     
         10 . The feature extraction apparatus of  claim 1 , further comprising a communication interface configured to receive the bio-signal from a biometric information detection apparatus, forward the bio-signal to the sensor and transmit a result of processing based on the received bio-signal to the biometric information detection apparatus. 
     
     
         11 . A feature extraction method for biometric information detection, the feature extraction method comprising:
 acquiring a bio-signal;   decomposing a waveform of the acquired bio-signal into component pulses;   obtaining subsequent component pulses based on new bio-signals generated based on iteratively eliminating the component pulses; and   extracting a feature for the biometric information detection based on characteristic points of each of the component pulses.   
     
     
         12 . The feature extraction method of  claim 11 , wherein the decomposing comprises:
 estimating a given component pulse to be decomposed, among the component pulses, based on the waveform of the bio-signal; and   generating the given component pulse based on the estimation result.   
     
     
         13 . The feature extraction method of  claim 12 , further comprising, when the given component pulse is generated, eliminating the generated component pulse from the bio-signal,
 wherein the estimating of the given component pulse comprises, when a number of generated component pulses is smaller than a pre-set number, estimating another subsequent component pulse based on another new bio-signal generated by eliminating the generated component pulse.   
     
     
         14 . The feature extraction method of  claim 12 , wherein the estimating of the given component pulse comprises modeling a component pulse waveform function based on the waveform of the bio-signal and estimating parameters of the component pulse waveform function. 
     
     
         15 . The feature extraction method of  claim 14 , wherein the modeling of the component pulse waveform function comprises modeling the component pulse waveform function to be bilaterally asymmetric with respect to a time axis, based on a Gaussian waveform function and the asymmetry factor. 
     
     
         16 . The feature extraction method of  claim 15 , further comprising adjusting the asymmetry factor for a waveform of the given component pulse to be estimated. 
     
     
         17 . The feature extraction method of  claim 16 , wherein the adjusting of the asymmetry factor comprises:
 estimating the given component pulse based on the bio-signal and the asymmetry factor;   eliminating the estimated component pulse from the bio-signal;   comparing a mean time period of the bio-signal from which the component pulse is eliminated with a time period at a peak point, and,   in response to determining that the mean time period is smaller than the time period at a peak point, adjusting the asymmetry factor and re-estimating the given component pulse.   
     
     
         18 . A biometric information detection apparatus comprising:
 a sensor configured to emit light onto a user's skin, detect the light reflecting from the user's skin, and measure a bio-signal based on the detected light; and   a processor configured to decompose a waveform of the bio-signal into component pulses, obtain subsequent component pulses based on new bio-signals generated based on iteratively eliminating the component pulses, and extract a feature of the bio-signal based on characteristic points of the component pulses, and detect biometric information based on the feature.   
     
     
         19 . The biometric information detection apparatus of  claim 18 , wherein the processor is further configured to model a component pulse waveform function which is asymmetric with respect to a time axis, based on a Gaussian waveform function and the asymmetry factor, and decompose the bio-signal into the component pulses based on the modeled component pulse waveform function. 
     
     
         20 . The biometric information detection apparatus of  claim 18 , wherein the processor is further configured to extract the feature based on characteristic points which include one or more of time, an amplitude, a standard deviation, and an offset of the component pulses.