Infrared spectrum detection method, device and system for wearable equipment

Abstract

The invention discloses an infrared spectrum detection method, device and system for wearable equipment. The device can be implemented as an integrated circuit chip, a module or a chip-module combination. The device comprises a multi-discrete SWIR light source driving and interface unit, an InGaAs detection reading unit, a time division scanning control unit and a synchronous demodulation processing unit. The control unit sequentially drives the multiple discrete SWIR light sources to emit according to a preset time division scanning sequence, and modulates emitted light in each time slot; the InGaAs detector collects skin reflection or back scattering signals, the synchronous demodulation unit executes related demodulation or phase-locked demodulation, amplitude features of all discrete wavelengths are extracted, and discrete spectrum feature vectors are formed and output. The scheme is suitable for wearable equipment, and has the advantages of resisting ambient light interference, being capable of expanding multi-emission position or multi-distance measurement and facilitating chip and modular integration.

Description

Technical Field

[0001] This invention relates to the field of wearable optoelectronic detection and integrated circuit application technology, specifically to an infrared spectral detection method, device, and system for wearable devices. Background Technology

[0002] Wearable devices (such as smartwatches, wristbands, and patches) typically use visible light or near-infrared solutions in the field of health monitoring. However, the short-wave infrared (SWIR) band is more sensitive to the absorption of molecular vibrations such as water and lipids, and has shown potential in applications such as detecting changes in skin hydration and analyzing tissue state trends.

[0003] However, introducing SWIR measurement into wearable scenarios faces multiple limitations and technical challenges: On the one hand, wearable products are highly sensitive to size and power consumption, and the traditional spectral architecture of "broadband light source + spectrometer + multi-channel detection" is often difficult to implement in a limited space, and the system power consumption and number of devices are not conducive to long-term wear; on the other hand, the background light composition in open environments is complex and may be superimposed with periodic interference such as power frequency, which can easily introduce additional noise and lead to a decrease in signal-to-noise ratio; at the same time, movement, tightness changes and fit fluctuations during wearing can cause instability in reflection / backscattering intensity, making the contrast relationship between different wavelengths more likely to be distorted, thus affecting the reliability of feature extraction; in addition, the multi-detection channel structure also brings challenges to cross-channel gain consistency and calibration maintenance, further increasing system complexity and mass production difficulty.

[0004] Based on the above problems, there is an urgent need for a miniaturized, low-power SWIR discrete multi-wavelength detection solution for wearable devices, and it is hoped that it can be implemented in the form of chips or modules to meet the needs of large-scale integration and mass production applications.

[0005] In view of this, this application proposes an infrared spectral detection method, apparatus and system for wearable devices. Summary of the Invention

[0006] To achieve the above objectives, this application provides an infrared spectral detection method, apparatus, and system for wearable devices, the specific technical solution of which is as follows:

[0007] An infrared spectral detection device for wearable devices, comprising:

[0008] A multi-discrete SWIR light source interface / drive unit is used to connect and drive at least two discrete SWIR light source units to emit short-wave infrared light of different discrete wavelengths;

[0009] A single InGaAs detector readout unit is used to connect to a single InGaAs detector and read out its output to obtain a detection signal;

[0010] The time-division scanning control unit is used to sequentially drive the discrete SWIR light source unit to emit according to a preset time-division scanning sequence in different time slots;

[0011] The synchronous demodulation processing unit performs synchronous demodulation on the detection signal based on the modulation reference signal provided by the time-division scanning control unit to extract the amplitude features corresponding to each discrete wavelength;

[0012] The feature output interface unit is used to summarize the amplitude features of each discrete wavelength to form a discrete spectral feature vector and output it.

[0013] Preferably, the device is implemented using an integrated circuit chip, which integrates at least the multi-discrete SWIR light source interface / driving unit, the time-division scanning control unit, the synchronous demodulation processing unit, and the feature output interface unit;

[0014] The integrated circuit chip also integrates the transimpedance amplification and / or analog-to-digital conversion circuits in the single InGaAs probe readout unit.

[0015] Preferably, the device is implemented by a module, which includes: an integrated circuit chip, a discrete SWIR light source unit, a single InGaAs detector, and an optical window package; the optical window package is used to form a skin-like reflection or backscattering optical path and to shield and position the discrete SWIR light source unit and the InGaAs detector.

[0016] The discrete SWIR light source unit is a laser or a narrowband light-emitting device. The center wavelength of the discrete SWIR light source unit is in the range of 900nm to 2500nm, and the number of discrete wavelengths is 3 to 10.

[0017] Preferably, the time-division scanning control unit is further configured to modulate the emitted light of the corresponding discrete SWIR light source unit in each time slot, wherein the modulation includes at least one of square wave modulation, pseudo-random code modulation and multi-frequency modulation.

[0018] The synchronous demodulation processing unit extracts the amplitude features using at least one of correlation demodulation, phase-locked demodulation, or synchronous integration.

[0019] The discrete SWIR light source unit includes at least two different emission positions, enabling the device to form at least two source detection distances, and the time-division scanning control unit performs time-division multiplexing measurements between the different emission positions.

[0020] An infrared spectral detection system for wearable devices includes: a discrete SWIR light source unit, a single InGaAs detector, and the aforementioned infrared spectral detection device for wearable devices.

[0021] The device controls the emission of the discrete SWIR light source unit through time-division scanning and synchronously demodulates the detection signal collected by the single InGaAs detector to output a discrete spectral feature vector.

[0022] An infrared spectral detection method for wearable devices, applied to the aforementioned infrared spectral detection device for wearable devices, includes:

[0023] Set a time-division scanning sequence containing multiple discrete wavelengths;

[0024] According to the time-division scanning sequence, the discrete SWIR light source unit is driven to emit in different time slots, and the emitted light is modulated in each time slot;

[0025] The detection signal is obtained by acquiring the reflected or backscattered light from the object under test using a single InGaAs detector and reading it out.

[0026] The detection signal is synchronously demodulated based on the modulation reference signal to obtain the amplitude characteristics corresponding to each discrete wavelength.

[0027] The amplitude characteristics corresponding to each discrete wavelength are summarized to form a discrete spectral feature vector and then output.

[0028] Preferably, the measurement results of different emission positions or different source detection distances are acquired by time-division multiplexing and output as a discrete spectral feature tensor containing the source detection distance dimension; the source detection distance is defined as the measurement distance along the skin interface between the geometric center of the light-emitting surface of the discrete SWIR light source unit and the geometric center of the light-incident surface of the InGaAs detector.

[0029] The system collects motion, temperature, or contact state parameters and outputs these parameters along with discrete spectral feature vectors for quality control or compensation. The contact state parameters are acquired by a skin contact state sensor, which includes at least one of a capacitive proximity sensor, a piezoresistive pressure sensor, and a photoelectric coupling strength detector.

[0030] Preferably, differential absorption feature extraction based on wavelength pairs specifically includes: grouping the multiple discrete wavelengths into multiple wavelength pairs, each wavelength pair containing an absorption-sensitive wavelength and an absorption reference wavelength; performing logarithmic difference operations on the two amplitude features within each wavelength pair to obtain differential absorption features; summing the differential absorption features of each wavelength pair to form a differential absorption feature vector and outputting it.

[0031] Interference source identification and suppression based on modulation frequency diversity specifically includes: alternating modulation of the same discrete wavelength using at least two different modulation frequencies in a single time slot; synchronously demodulating the detection signals corresponding to the at least two different modulation frequencies to obtain multiple demodulation results; comparing the consistency of the multiple demodulation results, determining the presence of frequency interference when the consistency is lower than a preset threshold, and selecting the demodulation result with the smallest deviation from the demodulation result from the multiple demodulation results as the amplitude feature of the discrete wavelength.

[0032] Preferably, the estimation of tissue optical parameters based on the amplitude attenuation law at multiple distances specifically includes: obtaining the amplitude characteristics of the same discrete wavelength at at least three different source detection distances; performing curve fitting based on the attenuation law of the at least three amplitude characteristics as the source detection distance changes; and extracting attenuation characteristic parameters related to tissue absorption and scattering from the fitting results and outputting them as additional features.

[0033] The closed-loop stabilization control of the light source emission power specifically includes: acquiring the actual emission power indication value of each discrete SWIR light source unit through the built-in optical power monitoring channel in each time slot; comparing the actual emission power indication value with the preset target power value to obtain the power deviation; and adjusting the driving current of the corresponding discrete SWIR light source unit according to the power deviation to achieve closed-loop stabilization of the emission power.

[0034] Preferably, the identification of abnormal wavelengths based on wavelength correlation constraints specifically includes: determining the expected correlation range between the amplitude characteristics of each discrete wavelength according to a pre-established normal correlation model between discrete wavelengths; calculating the actual correlation between the amplitude characteristics of each discrete wavelength in the current measurement period; and marking the discrete wavelength as abnormal and outputting an abnormality flag when the correlation between the amplitude characteristics of a discrete wavelength and other wavelengths deviates from the expected correlation range.

[0035] Spectral acquisition trigger control based on motion parameters specifically includes: real-time acquisition of motion acceleration parameters of the wearable device; when the amplitude of the motion acceleration parameters is lower than a preset static threshold and the duration exceeds a preset stable duration, triggering the execution of the time-division scanning sequence for spectral acquisition; when the amplitude of the motion acceleration parameters is continuously higher than a preset motion threshold, spectral acquisition is paused and a low-power standby state is entered.

[0036] Randomizing the scanning sequence to suppress periodic interference specifically includes: before the start of each complete measurement cycle, pseudo-randomizing the scanning order of each discrete wavelength in the time-division scanning sequence; recording the scanning order used in the current measurement cycle for wavelength-time slot correspondence by the synchronous demodulation processing unit; and breaking the fixed phase relationship with external periodic interference sources by randomizing the scanning order.

[0037] Compared with the prior art, the present invention has at least the following beneficial effects:

[0038] This application utilizes a single InGaAs detector combined with time-division scanning to achieve multi-discrete wavelength measurement, reducing issues related to size, power consumption, and channel consistency. The modulation and synchronous demodulation techniques employed significantly suppress ambient light and power frequency / low-frequency drift, improving the signal-to-noise ratio. The short duty cycle modulation and on-demand scanning in this application reduce average optical power. This application integrates light source driving, timing control, synchronous demodulation, and interface output into a chip or module for easy mass production integration. This application can expand multi-location / multi-distance measurements by extending different emission positions or source-detector distance information through time-division multiplexing without increasing the number of detectors. Attached Figure Description

[0039] Figure 1 A structural block diagram of an infrared spectral detection system / device for wearable devices provided in this application;

[0040] Figure 2 The timing diagram for time-division scanning and synchronization adjustment provided in this application;

[0041] Figure 3 This is a schematic diagram of the skin-contact reflective optical structure provided in this application;

[0042] Figure 4 A schematic diagram of single-position wavelength operation provided in this application;

[0043] Figure 5 This is a schematic diagram of the wavelength dual-position / dual-distance operation provided in this application;

[0044] Figure 6 The signal processing flowchart provided for this application. Detailed Implementation

[0045] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0046] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0047] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of this application. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0048] Example 1

[0049] The first embodiment of this application provides a wearable SWIR detection system with three wavelengths and a single emission position. As shown in the figure, Figure 1 This is a block diagram of the overall system structure of the present invention, showing the overall hardware architecture consisting of a light source interface / drive unit, a detection readout unit, a time-division scanning control unit, a synchronous demodulation processing unit, and a feature output interface unit, and indicating the signal flow between each unit. As shown in the figure... Figure 2 This is a timing diagram illustrating time-division scanning, modulation, and synchronous demodulation. It demonstrates how the system allocates measurement time slots of different wavelengths along the time axis and extracts signals within these time slots through the synchronous coordination of the modulation signal and the demodulation reference signal. As shown in the figure... Figure 3 This is a schematic diagram of a skin-adhesive reflective optical structure, showing the spatial layout of multiple discrete SWIR light sources and a single InGaAs detector, as well as the optical path process after the light beam enters the skin tissue, is reflected or backscattered, and is received by the detector. Figure 3 It contains multiple discrete SWIR light sources and a single InGaAs detector, and specifically describes the optical path process of the light beam emitted by the light source entering the skin tissue, being reflected or backscattered, and then being received by the detector; the optical window encapsulation plays a role in shielding and isolating light, preventing direct crosstalk of light, and ensuring that only the signal after passing through the skin is collected.

[0050] In terms of light source configuration, this embodiment uses a discrete SWIR light source unit comprising three narrowband light-emitting devices with center wavelengths set to λ1, λ2, and λ3, respectively. All three center wavelengths fall within the short-wave infrared band of 900nm to 2500nm. The three discrete SWIR light source units are centrally arranged at the same emission position within an optical window package, forming a fixed source-detection distance with the single InGaAs detector. The optical window package provides light-shielding isolation between the three discrete SWIR light source units and the InGaAs detector, preventing emitted light from directly entering the detector without the intervention of skin tissue, thus preventing crosstalk.

[0051] In terms of time-division scanning control, the time-division scanning control unit cyclically drives the three discrete SWIR light source units to emit sequentially in the order λ1→λ2→λ3. A single complete measurement cycle is divided into three time slots. In the first time slot, only the discrete SWIR light source unit with a center wavelength of λ1 is driven to emit; in the second time slot, only the discrete SWIR light source unit with a center wavelength of λ2 is driven to emit; and in the third time slot, only the discrete SWIR light source unit with a center wavelength of λ3 is driven to emit. A guard interval is set between adjacent time slots to avoid signal overlap during the light source switching process. In each time slot, the time-division scanning control unit performs square wave modulation on the currently emitting discrete SWIR light source unit. The modulation frequency of the square wave modulation is selected to avoid power frequency interference and its harmonic frequencies.

[0052] In terms of signal acquisition and processing, after the short-wave infrared light emitted by the three discrete SWIR light source units illuminates the skin surface of the wearer's wrist, part of the light is reflected from the skin surface, while the rest enters the skin tissue and undergoes scattering and absorption by the tissue components before returning to the skin surface as backscattered light. A single InGaAs detector collects the reflected and backscattered light signals returning from the skin surface, and a single InGaAs detector readout unit performs transimpedance amplification and analog-to-digital conversion on the photocurrent signal output by the InGaAs detector to obtain a digitized detection signal.

[0053] In terms of synchronous demodulation, the synchronous demodulation processing unit performs synchronous integration or correlation demodulation processing on the detection signals of each time slot based on the square wave modulation reference signal provided by the time-division scanning control unit. Synchronous demodulation processing extracts the signal component synchronized with the modulation frequency by performing correlation operations between the detection signal and the modulation reference signal, while suppressing the ambient light DC component and power frequency interference components. The synchronous demodulation processing unit obtains amplitude features A(λ1), A(λ2), and A(λ3) corresponding to three discrete wavelengths. Amplitude feature A(λ1) characterizes the intensity information of short-wave infrared light with a center wavelength of λ1 after reflection or backscattering by skin tissue; amplitude feature A(λ2) characterizes the intensity information of short-wave infrared light with a center wavelength of λ2 after reflection or backscattering by skin tissue; and amplitude feature A(λ3) characterizes the intensity information of short-wave infrared light with a center wavelength of λ3 after reflection or backscattering by skin tissue.

[0054] As shown in the figure Figure 4 The diagram below illustrates the operation of Example 1, demonstrating the specific implementation logic of the system sequentially driving three wavelength channels and outputting corresponding feature vectors in a three-wavelength, single-emission-position configuration. The wearable SWIR detection system with three wavelengths and a single emission position divides the measurement period into three time slots, sequentially driving three light source channels of different wavelengths. Within each time slot, modulation emission and synchronous demodulation are completed, and finally, the amplitude characteristics of the three wavelengths are summarized, outputting a one-dimensional discrete spectral feature vector.

[0055] Regarding feature output, the feature output interface unit summarizes and organizes the amplitude features corresponding to the three discrete wavelengths in wavelength order to form a discrete spectral feature vector A=[A(λ1),A(λ2),A(λ3)]. The feature output interface unit transmits the discrete spectral feature vector to the wearable main control / application processor through a digital interface, so that the upper-layer algorithm can perform skin hydration trend analysis or tissue state identification.

[0056] This implementation achieves measurement of three discrete wavelengths by combining a single InGaAs detector with time-division scanning, avoiding the channel gain consistency calibration problem caused by multi-detector schemes; it improves the measurement signal-to-noise ratio by suppressing ambient light and power frequency interference through square wave modulation and synchronous demodulation; the overall architecture is small in size and low in power consumption, making it suitable for integration into wearable devices such as smartwatches.

[0057] Example 2

[0058] A second embodiment of this application provides a wearable SWIR detection system with four wavelengths and dual emission positions / dual distances. See also... Figure 4 The diagram below illustrates the working principle of Example 2, which shows how, in a four-wavelength, dual-emission-position configuration, the time slot is further divided into position sub-time slots to acquire signals and output feature tensors at different source-detection distances. Figure 5 This study demonstrates the time-division multiplexing principle under a four-wavelength, dual-emission-position configuration. The main wavelength time slot is further divided into position sub-time slots, each driving an emission source at a different distance. The detector acquires signals from shallow and deep tissues in different sub-time slots, ultimately outputting a discrete spectral feature tensor containing both wavelength and distance dimensions.

[0059] This embodiment provides a four-wavelength dual-range shortwave infrared spectroscopy detection system for smart bracelet applications.

[0060] In terms of light source configuration, the discrete SWIR light source unit in this embodiment includes four wavelength channels, with center wavelengths of λ1, λ2, λ3, and λ4, all within the short-wave infrared band of 900nm to 2500nm. At least some of the four wavelength channels are equipped with two sets of light-emitting elements, arranged at two different emission positions within the optical window package. These two different emission positions and the single InGaAs detector form two different source-detection distances, d1 and d2, where the source-detection distance d1 is less than the source-detection distance d2. The source-detection distance is defined as the measured distance along the skin-contact interface from the geometric center of the emission surface of the discrete SWIR light source unit to the geometric center of the incident surface of the InGaAs detector. A smaller source-detection distance d1 corresponds to a shallower tissue detection depth, while a larger source-detection distance d2 corresponds to a deeper tissue detection depth.

[0061] In terms of time-division scanning control, the time-division scanning control unit divides a single complete measurement cycle into multiple time slots. For a wavelength channel configured with two emission positions, the time-division scanning control unit further divides the main time slot of the corresponding wavelength into two position sub-time slots. In the first position sub-time slot, the time-division scanning control unit drives only the light-emitting element at the first emission position to emit and modulate; in the second position sub-time slot, the time-division scanning control unit drives only the light-emitting element at the second emission position to emit and modulate. The time-division scanning control unit achieves time-division multiplexing measurement between different wavelength channels and different emission positions by sending a light source selection signal containing wavelength selection and position selection fields to the multi-discrete SWIR light source interface / drive unit.

[0062] In terms of signal acquisition and processing, short-wave infrared light emitted by discrete SWIR light source units at each emission position illuminates the skin surface of the wearer's wrist and is reflected or backscattered back by the skin tissue. Because the source-detection distances between the two emission positions and the InGaAs detector are different, light emitted from the first emission position mainly propagates through the superficial skin tissue before returning to the detector, while light emitted from the second emission position propagates through deeper skin tissue before returning to the detector. A single InGaAs detector acquires reflected or backscattered light signals from different emission positions within each sub-time slot. A single InGaAs detector readout unit performs transimpedance amplification and analog-to-digital conversion on the photocurrent signal to obtain the detection signal.

[0063] In terms of synchronous demodulation, the synchronous demodulation processing unit performs synchronous demodulation processing on the detection signals of each sub-time slot according to the division of the position sub-time slots. For a wavelength channel λi configured with two transmission positions, the synchronous demodulation processing unit obtains the amplitude feature A(λi,d1) at the source detection distance d1 and the amplitude feature A(λi,d2) at the source detection distance d2. The amplitude feature A(λi,d1) characterizes the intensity information of the short-wave infrared light with center wavelength λi after reflection or backscattering by the superficial skin tissue corresponding to the source detection distance d1, and the amplitude feature A(λi,d2) characterizes the intensity information of the short-wave infrared light with center wavelength λi after reflection or backscattering by the deep skin tissue corresponding to the source detection distance d2.

[0064] Regarding feature output, the feature output interface unit organizes the amplitude features obtained at different source detection distances for the same discrete wavelength into a distance dimension vector for that discrete wavelength. Then, it stacks these distance dimension vectors along the wavelength dimension to form a discrete spectral feature tensor containing the distance dimension. The row dimension of the feature tensor corresponds to the discrete wavelength indices λ1 to λ4, and the column dimension corresponds to the source detection distance indices d1 and d2. The feature output interface unit transmits the discrete spectral feature tensor to the wearable main controller / application processor via a digital interface.

[0065] This implementation uses dual emission positions to create two source detection distances, obtaining multidimensional measurement features containing depth information without increasing the number of detectors. The amplitude feature ratio of the same wavelength at the two source detection distances can be used to eliminate common-mode interference caused by changes in skin surface adhesion, improving robustness to variations in wearing tightness. The discrete spectral feature tensor containing the distance dimension provides richer depth-sensitive features for upper-layer algorithms, which can be used to distinguish optical information from different depth tissue layers of the skin, supporting more refined skin hydration state layering analysis or tissue state trend recognition applications.

[0066] Example 3

[0067] A third embodiment of this application provides a multi-discrete shortwave infrared spectroscopy detection method for wearable devices. As shown in the figure, Figure 6 The signal processing flowchart provided for this application illustrates the complete steps from setting the scan sequence to outputting the feature vector.

[0068] Step 1: Set up a time-division scanning sequence containing multiple discrete wavelengths.

[0069] The process of setting up a time-division scanning sequence first requires determining the number of discrete wavelengths involved in the measurement and the center wavelength value of each discrete wavelength. In wearable device multi-discrete short-wave infrared spectroscopy applications, the selection of discrete wavelengths is determined based on the absorption spectral characteristics of target molecules in the tested biological tissue. For skin hydration status monitoring applications, the selection of discrete wavelengths needs to cover the characteristic absorption peaks of water molecules in the short-wave infrared band and their adjacent reference bands.

[0070] In the discrete wavelength selection stage, multiple discrete wavelengths are grouped into several wavelength pairs based on the absorption spectral characteristics of the target molecule. Each wavelength pair includes an absorption-sensitive wavelength and an absorption reference wavelength. The center wavelength of the absorption-sensitive wavelength is located in the peak region of the target molecule's absorption peak or in the region of steep change in absorption edge. Light at the absorption-sensitive wavelength experiences strong absorption attenuation when passing through tissue containing the target molecule. The center wavelength of the absorption reference wavelength is located in the low absorption region or absorption plateau region of the target molecule's absorption spectrum. Light at the absorption reference wavelength experiences relatively weak absorption attenuation when passing through the same tissue.

[0071] During the time-division scanning sequence setup phase, measurement tasks at different emission positions or different source detection distances must be incorporated into the time slot allocation plan. The discrete SWIR light source unit comprises multiple sets of light-emitting elements distributed at different spatial locations, with each set of light-emitting elements forming a different source detection distance with a single InGaAs detector. The source detection distance is defined as the measurement distance along the skin-contact interface from the geometric center of the emission surface of the discrete SWIR light source unit to the geometric center of the incident surface of the InGaAs detector. Different source detection distances correspond to different propagation path depths of light in skin tissue, with larger source detection distances corresponding to deeper light detection depths. In the spatial layout design of the discrete SWIR light source unit, at least three different emission positions are configured for the same discrete wavelength, forming at least three source detection distances. The values ​​of the different source detection distances should cover a continuous range from near-distance shallow detection to far-distance deep detection.

[0072] During the time-division scanning sequence setup phase, at least two different modulation frequencies are configured for each discrete wavelength. The values ​​of the different modulation frequencies must be selected such that there is no integer multiple relationship between the modulation frequencies, and each modulation frequency should avoid power frequency interference and its harmonic frequencies. The modulation frequency configuration information is stored in the register group of the time-division scanning control unit.

[0073] The time-division scanning sequence is organized using a fixed-period time slot allocation structure. A single complete measurement cycle is divided into time slots equal to the number of discrete wavelengths, with each time slot dedicated to the emission and detection of one discrete wavelength. The time-division scanning control unit further divides the main time slot corresponding to each discrete wavelength into position sub-time slots. The number of position sub-time slots is equal to the number of emission positions configured for that discrete wavelength. Only one light-emitting element corresponding to each emission position is activated for emission and modulation within each position sub-time slot. The duration of the time slot is determined based on the minimum number of integration cycles required for synchronous demodulation, which must meet the requirements for signal-to-noise ratio improvement and ambient light suppression ratio. Guard intervals are set between time slots to avoid overlapping interference between the light source emission signals of adjacent time slots. The duration of the guard interval must be greater than the time required for the discrete SWIR light source unit to decay from shutdown to the background noise level.

[0074] The repetition frequency of the time-division scanning sequence is determined by balancing the rate of change of the measured physiological signal with the power consumption budget of the wearable device. For slowly changing physiological parameters such as skin hydration status, the repetition frequency of the time-division scanning sequence can be set to a lower value to reduce average power consumption. The configuration parameters of the time-division scanning sequence are stored in the register group of the time-division scanning control unit, which generates timing control signals based on the stored configuration parameters.

[0075] During the system calibration phase, a large number of discrete spectral feature vector samples under normal measurement conditions are collected. Statistical analysis is performed on the discrete spectral feature vector sample set to calculate the correlation coefficient matrix between the amplitude characteristics of each discrete wavelength pair. The elements of the correlation coefficient matrix characterize the linear correlation strength between the amplitude characteristics of each discrete wavelength pair. Based on the correlation coefficient matrix, the normal correlation range between each discrete wavelength pair is determined, expressed as an interval between multiples of the mean and standard deviation of the correlation coefficients. The parameters of the normal correlation model are stored in the non-volatile memory of the synchronous demodulation processing unit or the feature output interface unit.

[0076] A pseudo-random number generator is built into the time-division scanning control unit. This generator employs a linear feedback shift register structure or other deterministic pseudo-random sequence generation algorithm, using the system clock or partial data from the previous measurement as a random seed. Before the start of each complete measurement cycle, the time-division scanning control unit calls the pseudo-random number generator to produce a set of pseudo-random numbers. The time-division scanning control unit then rearranges the scanning order of each discrete wavelength in the time-division scanning sequence within the current measurement cycle based on these pseudo-random numbers. The rearrangement algorithm uses random permutation to ensure that each discrete wavelength is scanned once within the current measurement cycle, but in a different order than the previous measurement cycle. The time-division scanning control unit records the scanning order used in the current measurement cycle in its internal register.

[0077] By pre-setting a time-division scanning sequence containing multiple discrete wavelengths, ordered measurements of multiple discrete wavelengths can be achieved using a single InGaAs detector, avoiding the channel gain consistency calibration problem and volume and power consumption overhead associated with multi-detector schemes. Randomizing the scanning order allows the time slot positions of each discrete wavelength in the time-division scanning sequence to vary randomly between different measurement cycles. If a periodic interference source with a period close to that of the time-division scanning sequence exists in the measurement environment, a fixed scanning order would cause specific discrete wavelengths to always be at the interference peak position and be significantly affected. After randomizing the scanning order, the interference effect is randomly distributed among the discrete wavelengths, and after averaging over multiple measurement cycles, the interference effect on each discrete wavelength tends to be equalized, preventing the continuous deterioration of the measurement quality of specific discrete wavelengths. Randomizing and rearranging the scanning sequence breaks the fixed phase relationship between the time-division scanning sequence and external periodic interference sources, reducing the systematic bias of periodic interference on the measurement results of specific discrete wavelengths and improving the robustness of discrete spectral feature vector measurements in complex interference environments.

[0078] Step 2: Drive the discrete SWIR light source unit to emit in different time slots according to the time-division scanning sequence, and modulate the emitted light in each time slot.

[0079] The wearable device's built-in inertial measurement unit continuously outputs triaxial acceleration data. A time-division scanning control unit or the wearable main controller / application processor acquires the triaxial acceleration data in real time and calculates the amplitude of the motion acceleration parameters. The amplitude calculation of the motion acceleration parameters uses the vector magnitude of the triaxial acceleration components. After high-pass filtering to remove the gravitational acceleration component, the dynamic acceleration amplitude is obtained.

[0080] The time-division scanning control unit compares the dynamic acceleration amplitude with a preset static threshold. When the dynamic acceleration amplitude remains below the preset static threshold for a period exceeding a preset stabilization time, the time-division scanning control unit determines that the wearable device is currently in a relatively static wearing state, and triggers the execution of a time-division scanning sequence for spectral acquisition. The preset static threshold and preset stabilization time are set according to the wearing stability requirements of the target application scenario. The time-division scanning control unit also compares the dynamic acceleration amplitude with a preset motion threshold. The preset motion threshold is greater than the preset static threshold. When the dynamic acceleration amplitude remains above the preset motion threshold, the time-division scanning control unit determines that the wearer is currently in motion, pauses spectral acquisition, and controls the multi-discrete SWIR light source interface / drive unit, single InGaAs detector readout unit, and synchronous demodulation processing unit to enter a low-power standby state. In the low-power standby state, the power supply to the main functional circuits of each unit is turned off, retaining only the minimum power supply for motion parameter monitoring. When the motion state ends, the dynamic acceleration amplitude falls back below the preset static threshold, and the preset stabilization time condition is met, the time-division scanning control unit wakes up each unit, exits the low-power standby state, and resumes spectral acquisition. By using motion parameter-based spectral acquisition trigger control, it is possible to avoid acquiring low-quality spectral data during periods of drastic changes in fit caused by motion. At the same time, during motion, the average power consumption of the system is reduced through a low-power standby state, thus extending the battery life of the wearable device.

[0081] During the execution of the time-division scanning sequence, the wearable device synchronously acquires auxiliary state parameters related to measurement quality. These auxiliary state parameters include three categories: motion parameters, temperature parameters, and contact state parameters. Motion parameters are acquired through the wearable device's built-in inertial measurement unit (IMU). The triaxial acceleration data output by the IMU is low-pass filtered before the acceleration vector amplitude is calculated, which characterizes the current intensity of the wearable device's motion. Motion parameters are labeled with the same timestamp as the discrete spectral feature vector to facilitate time alignment during data fusion by the upper-layer algorithm. Temperature parameters are acquired through a temperature sensor on the skin-contact side. The temperature sensor measures the interface temperature between the optical window package and the skin contact area, reflecting the surface temperature state of the measured skin area. Temperature parameters are used by the upper-layer algorithm to compensate for temperature-sensitive optical characteristics. Contact state parameters are acquired by a skin-contact state sensor. This sensor includes at least one of a capacitive proximity sensor, a piezoresistive pressure sensor, and a photoelectric coupling strength detector. The capacitive proximity sensor determines the degree of contact by measuring the capacitance change between the optical window package surface and the skin. Piezoresistive pressure sensors determine whether the wearing pressure is within an appropriate range by measuring the pressure exerted on the skin by the optical window enclosure. Photoelectric coupling strength detectors determine the state of optical contact with the skin by measuring changes in the coupling efficiency of emitted light at the skin interface.

[0082] According to the time slot allocation of the time-division scanning sequence, the time-division scanning control unit sends a light source selection signal and a drive enable signal to the multi-discrete SWIR light source interface / drive unit at the beginning of each time slot. The light source selection signal sent by the time-division scanning control unit to the multi-discrete SWIR light source interface / drive unit includes both wavelength selection and position selection fields. The multi-discrete SWIR light source interface / drive unit determines the specific light-emitting element to be activated based on the combination of the wavelength selection and position selection fields. The multi-discrete SWIR light source interface / drive unit activates the discrete SWIR light source unit corresponding to the current slot according to the light source selection signal, and controls the current injection of the discrete SWIR light source unit according to the drive enable signal. Under the action of the drive current, the discrete SWIR light source unit emits short-wave infrared light of the corresponding center wavelength, and the emitted light is encapsulated through an optical window and irradiates the surface of the skin tissue being tested.

[0083] During the effective transmission period of each time slot, the time-division scanning control unit modulates the emitted light of the discrete SWIR light source unit. Modulation is achieved by periodically changing the on / off state of the drive current, which switches according to a preset modulation waveform. The modulation waveform uses square wave modulation, and the modulation frequency must be selected to avoid power frequency interference and its harmonic frequencies, while ensuring a sufficient number of complete modulation cycles within a single time slot to guarantee the integration accuracy of synchronous demodulation. Within a time slot corresponding to a single discrete wavelength, the time-division scanning control unit divides the time slot into sub-intervals equal to the number of modulation frequencies. Each sub-interval sequentially modulates the emitted light of the discrete SWIR light source unit using a different modulation frequency. The time-division scanning control unit outputs a modulation reference signal corresponding to the modulation frequency of the current sub-interval within each sub-interval.

[0084] The multi-discrete SWIR light source interface / drive unit generates a modulation drive waveform based on the modulation reference signal output by the time-division scanning control unit. The high-level period of the modulation drive waveform corresponds to the lit state of the discrete SWIR light source unit, and the low-level period corresponds to the extinguished state of the discrete SWIR light source unit. The duty cycle of the modulation drive waveform is selected based on the thermal management requirements and average power consumption constraints of the discrete SWIR light source unit.

[0085] The multi-discrete SWIR light source interface / drive unit incorporates a built-in optical power monitoring channel. This channel acquires the actual emission power indication value of the discrete SWIR light source unit via a back-facing photodiode or a monitoring photodiode integrated within the light-emitting element package. The back-facing photodiode receives leakage light emitted from the back of the discrete SWIR light source unit; the leakage light power is proportional to the forward emission power. The optical power monitoring channel amplifies and samples the photocurrent output by the monitoring photodiode to obtain a digital representation of the actual emission power indication value. The time-division scanning control unit sends sampling trigger signals to the optical power monitoring channel in each time slot, and the channel acquires the actual emission power indication value during the high-level period of the modulation signal. The multi-discrete SWIR light source interface / drive unit compares the actual emission power indication value with a preset target power value to obtain a power deviation value. The preset target power value is set based on the factory calibration data of the discrete SWIR light source unit and the optical power requirements of the system design. When the power deviation value is positive, it indicates that the actual emission power is higher than the target power, and the multi-discrete SWIR light source interface / drive unit reduces the drive current setting value of the corresponding discrete SWIR light source unit. When the power deviation is negative, it indicates that the actual transmitted power is lower than the target power. The multi-discrete SWIR light source interface / drive unit increases the drive current setting of the corresponding discrete SWIR light source unit. The drive current adjustment is proportional to the power deviation, and the proportionality coefficient is set based on the drive current-optical power response characteristics. The update cycle of the closed-loop stabilization control can be set to update once per time-division scan sequence cycle, or once per multiple time-division scan sequences to reduce the computational overhead of the control loop. Through closed-loop stabilization control of the light source's transmitted power, the transmit power drift of the discrete SWIR light source unit caused by factors such as temperature changes and aging can be compensated, ensuring that the transmit power of each discrete wavelength remains stable during long-term use, thereby improving the long-term measurement consistency of the discrete spectral eigenvectors.

[0086] By modulating the emitted light within a time slot, the reflected or backscattered signal of the tested skin tissue carries modulation frequency characteristics, providing the basis for frequency-selective filtering in subsequent synchronous demodulation processing, thereby enabling the modulation signal to be separated and extracted from the ambient light background.

[0087] Step 3: Acquire the reflected or backscattered light from the object under test using a single InGaAs detector and read out the detection signal.

[0088] When short-wave infrared light emitted by the discrete SWIR light source unit illuminates the skin tissue being tested, part of the light undergoes specular reflection at the skin surface, while the remaining light enters the skin tissue and is scattered and absorbed by the tissue components before returning to the skin surface as backscattered light. A single InGaAs detector receives the mixed light signal of the reflected and backscattered light returning from the skin surface through an optical window enclosure. The optical window enclosure provides spatial light-shielding isolation between the discrete SWIR light source unit and the InGaAs detector, preventing the emitted light from the discrete SWIR light source unit from directly entering the InGaAs detector without being processed by the skin tissue, thus preventing crosstalk.

[0089] The InGaAs detector converts the received optical signal into a photocurrent signal, the amplitude of which is proportional to the incident light power. A single InGaAs detector readout unit performs transimpedance amplification on the photocurrent signal output from the InGaAs detector. The transimpedance amplifier converts the photocurrent signal into a voltage signal and provides the first stage of signal amplification. The feedback resistor value of the transimpedance amplifier is selected to match the photocurrent range of the InGaAs detector with the input voltage range of the subsequent analog-to-digital converter.

[0090] The voltage signal, after transimpedance amplification, enters the analog-to-digital converter (ADC) circuit for sampling and quantization. The sampling frequency of the ADC circuit must meet the Nyquist sampling theorem's requirement for the modulation signal bandwidth; the sampling frequency should be more than twice the modulation frequency to ensure complete reconstruction of the modulation signal. The digital detection signal output by the ADC circuit includes the modulated optical signal component, the ambient light DC component, and the circuit noise component. By acquiring the reflected or backscattered signals of all discrete wavelengths using a single InGaAs detector, the inter-wavelength measurement error caused by the difference in responsivity between different detectors in multi-detector schemes is eliminated, simplifying the system calibration process.

[0091] Step 4: Perform synchronous demodulation on the detection signal based on the modulation reference signal to obtain the amplitude characteristics corresponding to each discrete wavelength.

[0092] The synchronous demodulation processing unit obtains a modulation reference signal synchronized with the transmitted light modulation in the current time slot from the time-division scanning control unit. The frequency and phase of the modulation reference signal maintain strict synchronization with the modulation drive signal of the discrete SWIR light source unit, and the modulation reference signal is used as the local oscillator signal for the synchronous demodulation operation. The synchronous demodulation processing unit determines the discrete wavelength index corresponding to each time slot based on the current scanning sequence information provided by the time-division scanning control unit, thereby correctly establishing the mapping relationship between the demodulation result and the discrete wavelength. The synchronous demodulation processing unit processes the digital detection signal using correlation demodulation, which involves multiplying the detection signal and the modulation reference signal point by point and then integrating the results.

[0093] The integration time of the correlation demodulation operation covers the entire complete modulation cycle within the current time slot. During the integration time, signal components with the same frequency and phase as the modulation reference signal are multiplied and accumulated to produce positive DC accumulation, while ambient light signal components with different frequencies than the modulation reference signal are multiplied and accumulated to produce alternating positive and negative cancellation. The longer the integration time, the stronger the suppression capability of synchronous demodulation for non-co-frequency signals, but the integration time is constrained by the duration of a single time slot.

[0094] The synchronous demodulation processing unit performs synchronous demodulation operations on the detected signals in each sub-interval, obtaining multiple demodulation results corresponding to each modulation frequency. Since the reflected or backscattered signals from real skin maintain the same optical response characteristics in each sub-interval, theoretically, the demodulation results corresponding to each modulation frequency should have high consistency. When a periodic interference source close to a certain modulation frequency exists in the measurement environment, the demodulation result corresponding to that modulation frequency will be superimposed with the interference component, deviating from the true value. The synchronous demodulation processing unit calculates the relative deviation between multiple demodulation results and compares the relative deviation with a preset consistency threshold. When the relative deviation exceeds the preset consistency threshold, the synchronous demodulation processing unit determines that there is specific frequency interference. By comparing the noise level or stability index of each demodulation result, the synchronous demodulation processing unit selects the demodulation result with less interference as the amplitude characteristic output of the current discrete wavelength. Through interference source identification and suppression based on modulation frequency diversity, it can adaptively avoid interference effects in complex electromagnetic environments or scenarios with periodic optical interference, improving the reliability of discrete wavelength amplitude characteristic measurement.

[0095] After completing the relevant demodulation integration, the synchronous demodulation processing unit performs a normalization operation, dividing the accumulated integration result by the number of integration points to obtain the amplitude feature of the discrete wavelength corresponding to the current time slot. The amplitude feature characterizes the intensity information of the current discrete wavelength shortwave infrared light after reflection or backscattering by skin tissue. The synchronous demodulation processing unit processes the detection signals of each time slot sequentially according to the time-division scanning sequence, extracting the amplitude features corresponding to each discrete wavelength. The synchronous demodulation processing unit performs synchronous demodulation processing on the detection signals of each sub-time slot according to the division of the position sub-time slots. The synchronous demodulation processing unit outputs an amplitude feature value for each combination of discrete wavelength and source detection distance. Through synchronous demodulation processing, the influence of ambient light DC component and power frequency interference and their harmonic components on the measurement results can be effectively suppressed, significantly improving the measurement signal-to-noise ratio and stability of discrete wavelength amplitude features.

[0096] During real-time measurement, the synchronous demodulation processing unit or the feature output interface unit acquires the amplitude characteristics of each discrete wavelength in the current measurement period and calculates the actual correlation between the amplitude characteristics of each discrete wavelength pair in the current measurement period. The calculation of the actual correlation uses the amplitude characteristic data of the most recent few measurement periods within a sliding window to estimate the short-term correlation coefficient. The synchronous demodulation processing unit or the feature output interface unit compares the actual correlation coefficient of each discrete wavelength pair with the pre-stored normal correlation range. When the actual correlation coefficient between a discrete wavelength and multiple other discrete wavelengths deviates from the corresponding normal correlation range, the discrete wavelength is determined to be abnormal. The abnormality may be caused by factors such as abnormal emission power of the corresponding discrete SWIR light source unit, local contamination of the optical window, or nonlinearity of the detector response. The feature output interface unit adds an abnormality flag to the discrete wavelengths identified as abnormal. The abnormality flag is output along with the discrete spectral feature vector for the upper-layer algorithm to remove or reduce the weight of abnormal wavelength data in subsequent processing. By identifying abnormal wavelengths based on wavelength correlation constraints, the abnormal state of a single wavelength channel can be detected in real time during the measurement process, avoiding the contamination of the overall spectral feature analysis results by abnormal wavelength data and improving the system's fault detection and self-diagnosis capabilities.

[0097] After the synchronous demodulation processing unit outputs the amplitude characteristics of each discrete wavelength, the feature output interface unit performs logarithmic difference operations on the two amplitude characteristics within each wavelength pair. The logarithmic difference operation first calculates the natural logarithmic values ​​of the amplitude characteristics of the absorption-sensitive wavelength and the absorption-reference wavelength, respectively. Then, it calculates the difference between the natural logarithmic values ​​of the amplitude characteristics of the absorption-sensitive wavelength and the absorption-reference wavelength. The resulting difference is the differential absorption characteristic of the current wavelength pair. The physical significance of the logarithmic difference operation lies in utilizing the logarithmic linear relationship of the Beer-Lambert law, where the differential absorption characteristic is proportional to the difference in the absorption coefficients of the target molecule at the two wavelengths and its concentration-optical path product. Since the absorption-sensitive wavelength and the absorption-reference wavelength undergo similar skin surface scattering and interface reflection processes, the logarithmic difference operation can eliminate the common optical path loss factors of the two wavelengths, highlighting the specific absorption contribution of the target molecule. The feature output interface unit performs logarithmic difference operations on each wavelength pair to obtain the corresponding differential absorption characteristic values, and then summarizes the differential absorption characteristic values ​​of each wavelength pair according to the wavelength pair index order to form a differential absorption characteristic vector. The number of elements in the differential absorption eigenvector is equal to the number of wavelength pairs. By extracting differential absorption features based on wavelength pairs, the influence of skin surface scattering variations and adhesion fluctuations on measurement results can be reduced, enhancing the sensitivity and specificity of measurement results to changes in target molecule concentration.

[0098] Using the numerical values ​​of each source detection distance as the independent variable and the natural logarithm of the corresponding amplitude feature as the dependent variable, an amplitude attenuation data point set is established. Based on a simplified model of diffuse light propagating in a semi-infinite homogeneous scattering medium, the natural logarithm of the amplitude feature and the source detection distance exhibit an approximately linear attenuation relationship. The slope of this linear attenuation relationship is related to the effective attenuation coefficient of the tissue, which comprehensively reflects the tissue's absorption coefficient and reduced scattering coefficient. The synchronous demodulation processing unit or feature output interface unit performs a linear least squares fitting operation on the amplitude attenuation data point set. The linear least squares fitting operation solves for the slope parameter and intercept parameter that minimize the sum of squared residuals between the fitted line and each data point. The slope parameter obtained from the fitting represents the effective attenuation feature parameter corresponding to the current discrete wavelength. The feature output interface unit summarizes the effective attenuation feature parameters corresponding to each discrete wavelength to form an attenuation feature vector. By estimating tissue optical parameters based on the multi-distance amplitude attenuation law, attenuation feature parameters directly related to tissue absorption and scattering characteristics can be extracted from multi-distance measurement data, providing a more physically meaningful tissue state characterization index for upper-level algorithms.

[0099] Step 5: Summarize the amplitude features corresponding to each discrete wavelength to form a discrete spectral feature vector and output it.

[0100] The feature output interface unit receives amplitude feature data corresponding to each discrete wavelength from the synchronous demodulation processing unit. The feature output interface unit summarizes and organizes the amplitude features according to a preset arrangement order of the discrete wavelengths, forming a discrete spectral feature vector data structure. The number of elements in the discrete spectral feature vector is equal to the number of discrete wavelengths involved in the measurement, and the position of each element in the discrete spectral feature vector has a fixed index mapping relationship with its corresponding discrete wavelength.

[0101] The feature output interface unit organizes the amplitude features of the same discrete wavelength at different source detection distances into a distance dimension vector for that discrete wavelength, and then stacks the distance dimension vectors of each discrete wavelength along the wavelength dimension to form a two-dimensional feature matrix structure. The row dimension of the two-dimensional feature matrix structure corresponds to the discrete wavelength index, and the column dimension corresponds to the source detection distance index. The feature output interface unit encapsulates the two-dimensional feature matrix structure as a discrete spectral feature tensor. The data structure of the discrete spectral feature tensor contains wavelength dimension size, distance dimension size, and mapping table information between each dimension index and physical parameters. By time-division multiplexing the measurement results from different emission positions or different source detection distances and outputting a discrete spectral feature tensor containing the source detection distance dimension, the upper-layer algorithm can use the amplitude variation information of the same wavelength at different detection depths to analyze the hierarchical characteristics of the organization. At the same time, the fusion of multi-distance measurement data helps to improve the robustness to the fluctuation of fit state caused by changes in wearing tightness.

[0102] The feature output interface unit outputs the differential absorption feature vector as an additional feature or replaces the original discrete spectral feature vector. The feature output interface unit also outputs the attenuation feature vector as an additional feature along with the discrete spectral feature tensor.

[0103] The feature output interface unit adds measurement timestamp and sequence number information to the discrete spectral feature vector to support data traceability and time-series alignment for upper-layer applications. While outputting the discrete spectral feature vector, the feature output interface unit also encapsulates motion parameters, temperature parameters, and contact state parameters as additional data fields. These additional data fields and the discrete spectral feature vector share the same data frame for transmission to ensure time synchronization. By collecting motion, temperature, or contact state parameters and outputting them along with the discrete spectral feature vector, the upper-layer algorithm can assess the quality of the current measurement conditions based on the auxiliary state parameters, eliminating low-quality measurement data during periods of intense motion or poor contact, or using temperature parameters to compensate for environmental factors in the measurement results, thereby improving the accuracy of physiological state analysis based on discrete spectral features.

[0104] The feature output interface unit transmits discrete spectral feature vectors, discrete spectral feature tensors, and various additional features to the wearable main controller / application processor via a standard digital interface. The digital interface uses a serial communication protocol to reduce the number of interface pins. The data frame format of the serial communication protocol includes a frame header identifier, feature vector data area, checksum, and frame tail identifier.

[0105] After receiving discrete spectral feature vectors, the wearable main controller / application processor can call upper-layer algorithms to perform skin hydration trend analysis or tissue condition identification. By summarizing the amplitude features of each discrete wavelength into a unified discrete spectral feature vector, the upper-layer algorithm can easily obtain complete multi-wavelength measurement information. At the same time, the compact data format of the discrete spectral feature vector helps reduce data transmission bandwidth and storage space requirements.

[0106] The multi-discrete shortwave infrared spectroscopy detection method in this embodiment achieves multi-wavelength measurement under single detector conditions through time-division scanning, suppresses ambient light interference and improves the signal-to-noise ratio through modulation and synchronous demodulation within the time slot, and provides structured measurement data for the upper-level algorithm through discrete spectral feature vector output. Thus, it achieves high-quality shortwave infrared discrete spectroscopy detection function while meeting the constraints of small size and low power consumption of wearable devices.

[0107] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings or direct couplings or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0108] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of this application without departing from the spirit and scope of protection of the claims. All of these variations are within the protection scope of this application.

Claims

1. An infrared spectral detection device for wearable devices, characterized in that, include: A multi-discrete SWIR light source interface / driving unit is used to connect and drive at least two discrete SWIR light source units to emit short-wave infrared light of different discrete wavelengths; A single InGaAs detector readout unit is used to connect to a single InGaAs detector and read out its output to obtain a detection signal; The time-division scanning control unit is used to sequentially drive the discrete SWIR light source unit to emit according to a preset time-division scanning sequence in different time slots; The synchronous demodulation processing unit performs synchronous demodulation on the detection signal based on the modulation reference signal provided by the time-division scanning control unit to extract the amplitude features corresponding to each discrete wavelength; The feature output interface unit is used to summarize the amplitude features of each discrete wavelength to form a discrete spectral feature vector and output it.

2. The infrared spectral detection device for wearable devices according to claim 1, characterized in that, The device is implemented using an integrated circuit chip, which integrates at least the multi-discrete SWIR light source interface / driving unit, the time-division scanning control unit, the synchronous demodulation processing unit, and the feature output interface unit. The integrated circuit chip also integrates the transimpedance amplification and / or analog-to-digital conversion circuits in the single InGaAs probe readout unit.

3. The infrared spectral detection device for wearable devices according to claim 1, characterized in that, The device is implemented through a module, which includes: an integrated circuit chip, a discrete SWIR light source unit, a single InGaAs detector, and an optical window package; the optical window package is used to form a skin-like reflection or backscattering optical path and to shield and position the discrete SWIR light source unit and the InGaAs detector. The discrete SWIR light source unit is a laser or a narrowband light-emitting device. The center wavelength of the discrete SWIR light source unit is in the range of 900nm to 2500nm, and the number of discrete wavelengths is 3 to 10.

4. The infrared spectral detection device for wearable devices according to claim 1, characterized in that, The time-division scanning control unit is also used to modulate the emitted light of the corresponding discrete SWIR light source unit in each time slot, and the modulation includes at least one of square wave modulation, pseudo-random code modulation and multi-frequency modulation. The synchronous demodulation processing unit extracts the amplitude features using at least one of correlation demodulation, phase-locked demodulation, or synchronous integration. The discrete SWIR light source unit includes at least two different emission positions, enabling the device to form at least two source detection distances, and the time-division scanning control unit performs time-division multiplexing measurements between the different emission positions.

5. An infrared spectral detection system for wearable devices, characterized in that, include: A discrete SWIR light source unit, a single InGaAs detector, and an infrared spectral detection device for wearable devices as described in any one of claims 1 to 4; The device controls the emission of the discrete SWIR light source unit through time-division scanning and synchronously demodulates the detection signal collected by the single InGaAs detector to output a discrete spectral feature vector.

6. An infrared spectral detection method for wearable devices, applied to the infrared spectral detection device for wearable devices as described in any one of claims 1-4, characterized in that, include: Set a time-division scanning sequence containing multiple discrete wavelengths; According to the time-division scanning sequence, the discrete SWIR light source unit is driven to emit in different time slots, and the emitted light is modulated in each time slot; The detection signal is obtained by acquiring the reflected or backscattered light from the object under test using a single InGaAs detector and reading it out. The detection signal is synchronously demodulated based on the modulation reference signal to obtain the amplitude characteristics corresponding to each discrete wavelength. The amplitude characteristics corresponding to each discrete wavelength are summarized to form a discrete spectral feature vector and then output.

7. The infrared spectral detection method for wearable devices according to claim 6, characterized in that, The measurement results from different emission positions or different source detection distances are acquired by time-division multiplexing and output as a discrete spectral feature tensor containing the source detection distance dimension; the source detection distance is defined as the measurement distance along the skin interface between the geometric center of the light-emitting surface of the discrete SWIR light source unit and the geometric center of the light-incident surface of the InGaAs detector. The system collects motion, temperature, or contact state parameters and outputs these parameters along with discrete spectral feature vectors for quality control or compensation. The contact state parameters are acquired by a skin contact state sensor, which includes at least one of a capacitive proximity sensor, a piezoresistive pressure sensor, and a photoelectric coupling strength detector.

8. The infrared spectral detection method for wearable devices according to claim 6, characterized in that, Differential absorption feature extraction based on wavelength pairs specifically includes: grouping the multiple discrete wavelengths into multiple wavelength pairs, each wavelength pair containing an absorption-sensitive wavelength and an absorption reference wavelength; performing logarithmic difference operations on the two amplitude features within each wavelength pair to obtain differential absorption features; summing the differential absorption features of each wavelength pair to form a differential absorption feature vector and outputting it. Interference source identification and suppression based on modulation frequency diversity specifically includes: alternating modulation of the same discrete wavelength using at least two different modulation frequencies in a single time slot; synchronously demodulating the detection signals corresponding to the at least two different modulation frequencies to obtain multiple demodulation results; comparing the consistency of the multiple demodulation results, determining the presence of frequency interference when the consistency is lower than a preset threshold, and selecting the demodulation result with the smallest deviation from the demodulation result from the multiple demodulation results as the amplitude feature of the discrete wavelength.

9. The infrared spectral detection method for wearable devices according to claim 6, characterized in that, The estimation of tissue optical parameters based on the amplitude attenuation law at multiple distances specifically includes: obtaining the amplitude characteristics of the same discrete wavelength at at least three different source detection distances; performing curve fitting based on the attenuation law of the at least three amplitude characteristics as a function of source detection distance; and extracting attenuation characteristic parameters related to tissue absorption and scattering from the fitting results and outputting them as additional features. The closed-loop stabilization control of the light source emission power specifically includes: acquiring the actual emission power indication value of each discrete SWIR light source unit through the built-in optical power monitoring channel in each time slot; comparing the actual emission power indication value with the preset target power value to obtain the power deviation; and adjusting the driving current of the corresponding discrete SWIR light source unit according to the power deviation to achieve closed-loop stabilization of the emission power.

10. The infrared spectral detection method for wearable devices according to claim 9, characterized in that, Abnormal wavelength identification based on wavelength correlation constraints specifically includes: determining the expected correlation range between the amplitude characteristics of each discrete wavelength according to a pre-established normal correlation model between discrete wavelengths; calculating the actual correlation between the amplitude characteristics of each discrete wavelength in the current measurement period; and marking the discrete wavelength as abnormal and outputting an abnormality flag when the correlation between the amplitude characteristics of a discrete wavelength and other wavelengths deviates from the expected correlation range. Spectral acquisition trigger control based on motion parameters specifically includes: real-time acquisition of motion acceleration parameters of the wearable device; when the amplitude of the motion acceleration parameters is lower than a preset static threshold and the duration exceeds a preset stable duration, triggering the execution of the time-division scanning sequence for spectral acquisition; when the amplitude of the motion acceleration parameters is continuously higher than a preset motion threshold, spectral acquisition is paused and a low-power standby state is entered. Randomizing the scanning sequence to suppress periodic interference specifically includes: before the start of each complete measurement cycle, pseudo-randomizing the scanning order of each discrete wavelength in the time-division scanning sequence; recording the scanning order used in the current measurement cycle for wavelength-time slot correspondence by the synchronous demodulation processing unit; and breaking the fixed phase relationship with external periodic interference sources by randomizing the scanning order.

Images