Tongue picture feature intelligent identification method and system

Abstract

The invention provides an intelligent tongue picture feature identification method and system, relates to the technical field of biomedical detection, and aims to avoid confusion of subsequent signal processing caused by environmental noise by accurately identifying and eliminating ambient light interference with fixed frequency through fast Fourier transform analysis after a pre-processed original composite light signal is obtained. When the target lingual surface area shifts to cause dynamic change of the background fluorescence signal, the preset adaptive filtering algorithm can accurately identify and reject the changed background fluorescence digital signal values of the tongue picture in real time. The step-by-step and self-adaptive processing strategy ensures that the removal of ambient light interference and the correction of the background fluorescence signal are mutually independent and do not interfere with each other, thereby avoiding the generation of internal processing conflicts and calculation artifacts described in the background technology, and obtaining highly pure and accurate tongue picture fluorescence digital signals. The accuracy and reliability of tongue picture feature identification are remarkably improved, and a more reliable data basis is provided for oral health detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biomedical detection, in particular to the field of oral health detection and signal processing, and more particularly to a tongue feature intelligent identification method and system. BACKGROUND

[0002] In the field of biomedical detection, especially in the aspect of oral health assessment, the detection of tongue surface bacterial flora metabolite characteristics has important application value. Such detection usually utilizes fluorescence imaging technology to obtain relevant information by exciting specific substances to emit their own fluorescence. However, in the actual operating environment, there are many challenges in accurately obtaining and analyzing the target fluorescence signal.

[0003] In the prior art, when collecting the tongue surface fluorescence signal, the sensor often receives a composite light signal. This composite light signal not only contains the effective fluorescence signal from the target bacterial flora metabolite, but also inevitably mixes the background light signal from the detection environment and the background fluorescence signal produced by the subject's own tongue surface tissue.

[0004] In addition, during the detection process, due to the routine operation of the operator (such as fine-tuning the probe position to review the detection area), the target tongue surface area may be slightly displaced. The tongue surface tissue is not homogeneous, and the thickness of the tongue coating, the state of the epithelial cells and the distribution of the blood in different areas are different, which causes the background fluorescence intensity and spectral composition of the tongue surface tissue itself to change at different positions. When the probe position is displaced, the tongue surface tissue area collected by the sensor changes, causing the background fluorescence signal part in the composite light signal to change.

[0005] Further problems are that the processing module for correcting the changing background fluorescence signal in the prior art may have an unexpected internal correlation with the processing module for identifying and filtering out the fixed frequency noise signal. For example, when the module for correcting the background fluorescence signal adjusts the parameters, it may accidentally affect the threshold or gain parameters of the noise signal identification module, causing the noise identification ability to decrease. This conflict or coupling of internal processing logic makes it difficult for the system to effectively process environmental noise and changing background fluorescence signals at the same time, and ultimately may output a continuously erroneous detection result that combines environmental noise, changing background signals and calculation artifacts generated by internal processing conflicts.

[0006] Therefore, it is an urgent problem for those skilled in the art to provide a tongue feature intelligent identification method and system for solving the above problems. SUMMARY

[0007] To solve the problems in the prior art, the tongue feature intelligent recognition method and system have the advantages that the ambient light interference can be effectively filtered out, and the tongue background fluorescence change caused by displacement can be adaptively corrected, thereby improving the accuracy of tongue feature recognition.

[0008] To achieve the above object, the technical scheme provided by the present application is as follows: The tongue feature intelligent recognition method is applied to an oral health detection scene and includes the following steps: In a first preset time period, a preset standard light source collects an original composite light signal of a target tongue surface area, and the original composite light signal is preprocessed to obtain an original composite light digital signal; The original composite light digital signal is subjected to fast Fourier transform analysis to obtain an ambient light digital signal; In a second preset time period, the ambient light digital signal in the original composite light digital signal is removed to obtain an effective light digital signal; When the target tongue surface area is displaced to cause the effective light digital signal to change, the change trend of the effective light digital signal is analyzed according to a preset adaptive filtering algorithm, the tongue background fluorescence digital signal value corresponding to the change trend is removed, and a tongue fluorescence digital signal is obtained for tongue feature recognition.

[0009] Through the above scheme, the ambient light interference can be effectively filtered out, and the tongue background fluorescence change caused by displacement can be adaptively corrected, thereby improving the accuracy of tongue feature recognition.

[0010] Preferably, the collection of the original composite light signal of the target tongue surface area in the first preset time period and the preprocessing of the original composite light signal to obtain the original composite light digital signal include the following steps: In the first preset time period, the original composite light signal of the target tongue surface area is collected; The original composite light signal is subjected to photoelectric conversion to obtain an original composite electric signal; The original composite electric signal is subjected to analog-digital conversion to obtain the original composite light digital signal.

[0011] Through the above scheme, the acquisition process of the original composite light digital signal is refined, and the accuracy and reliability of signal collection are ensured.

[0012] Preferably, the fast Fourier transform analysis of the original composite light digital signal to obtain the ambient light digital signal includes the following steps: In a preset data window, the original composite light digital signal is subjected to fast Fourier transform analysis to obtain the original composite light frequency of a plurality of sampling points; fitting the original composite light frequency of the plurality of sampling points to obtain an original composite light frequency curve; determining whether a specific frequency in the original composite light frequency curve has a peak value; If so, define the specific frequency peak value as the ambient light digital signal.

[0013] Through the above scheme, a specific method for identifying the ambient light digital signal through fast Fourier transform analysis is provided, and the accuracy of ambient light interference identification is improved.

[0014] Preferably, the removing the ambient light digital signal in the original composite light digital signal to obtain an effective light digital signal within the second preset time period comprises the following steps: Based on a fixed frequency equal to the ambient light digital signal, a first digital notch filter is set; In the second preset period, the original composite light digital signal is input into the first digital notch filter; The effective light digital signal is output after being filtered by the first digital notch filter.

[0015] Through the above scheme, the scheme of removing the ambient light digital signal by using the digital notch filter is clarified, and the purity of the effective light digital signal is ensured.

[0016] Preferably, when the target tongue surface area is displaced, causing the effective light digital signal to change, the change trend of the effective light digital signal is analyzed according to a preset adaptive filtering algorithm, comprising the following steps: When the target tongue surface area is displaced, causing the first effective light digital signal to change to a second effective light digital signal; According to the sliding average algorithm or the low-pass filtering algorithm, the absolute value of the difference between the second effective light digital signal and the first effective light digital signal is calculated as the effective light digital signal change value.

[0017] Through the above scheme, a specific algorithm for calculating the effective light digital signal change value is provided, which provides a quantitative basis for subsequent adaptive filtering.

[0018] Preferably, after obtaining the effective light digital signal change value, the tongue image background fluorescence digital signal value corresponding to the change trend is removed to obtain a tongue image fluorescence digital signal, comprising the following steps: Based on the corresponding relationship between the target tongue surface area and the tongue image background fluorescence digital signal, the effective light digital signal change value is taken as the tongue image background fluorescence digital signal change value; According to the first tongue background fluorescence digital signal in the first effective light digital signal and the tongue background fluorescence digital signal change amount, a second tongue background fluorescence digital signal is obtained as the tongue background fluorescence digital signal value. A difference between the second effective light digital signal and the tongue background fluorescence digital signal value is calculated as the tongue fluorescence digital signal.

[0019] Through the above scheme, the specific steps of eliminating the tongue background fluorescence digital signal value according to the effective light digital signal change amount are elaborated, and the accuracy of background fluorescence correction is further improved.

[0020] Preferably, after obtaining the original composite light frequency curve by fitting the original composite light frequency of a plurality of sampling points, the method further comprises the following steps: Record all fixed frequencies and corresponding periodic characteristics and waveform stability; Respectively, the similarity calculation is performed between the fixed frequency and the corresponding periodic characteristic and the waveform stability and the preset biological criterion, to obtain a confusion risk index of each fixed frequency; According to the confusion risk index, each fixed frequency is prioritized from high to low; Based on the fixed frequency with the highest priority of the confusion risk index, a second digital notch filter is set; In a second preset period, the original composite light digital signal is input into the second digital notch filter; After filtering by the second digital notch filter, the effective light digital signal is output.

[0021] Through the above scheme, the confusion risk index and the second digital notch filter are introduced, which can more intelligently identify and filter out the fixed frequency interference that may be confused with biological characteristics, and further improve the signal purity.

[0022] Preferably, before the effective light digital signal is obtained by eliminating the ambient light digital signal in the original composite light digital signal in the second preset time period, the method further comprises the following steps: A preset first light source linearly scans the tongue tip to the tongue root at a preset fixed frequency to obtain a third tongue background fluorescence signal, and obtains a third tongue background fluorescence digital signal after preprocessing; The third tongue background fluorescence digital signal is fitted to generate a third tongue background fluorescence intensity curve as a tongue background fluorescence position reference map.

[0023] Through the above scheme, the method for pre-generating the tongue background fluorescence position reference map is provided, which provides basic data for subsequent background fluorescence compensation.

[0024] Preferably, after the tongue image background fluorescence position reference map is acquired, the following steps are further included: The standard light source is preset to linearly scan from the tongue tip to the tongue root at a preset fixed frequency to acquire the effective light digital signal; According to the current position in the linear scanning process, the tongue image background fluorescence reference value is retrieved from the tongue image background fluorescence position reference map; The difference between the effective light digital signal and the tongue image background fluorescence reference value is calculated as the tongue image fluorescence digital signal after compensation.

[0025] Through the above scheme, the background fluorescence compensation based on the position reference map is realized, and the accuracy of the tongue image fluorescence digital signal is further improved.

[0026] A tongue image feature intelligent recognition system comprises: A collection and preprocessing module is configured to collect an original composite light signal of a target tongue surface area by a preset standard light source within a first preset time period, and preprocess the original composite light signal to obtain an original composite light digital signal; A signal processing module is configured to perform fast Fourier transform analysis on the original composite light digital signal to obtain an ambient light digital signal; A signal noise reduction module is configured to eliminate the ambient light digital signal in the original composite light digital signal within a second preset time period to obtain an effective light digital signal; A signal screening module is configured to, when the target tongue surface area is displaced to cause a change in the effective light digital signal, analyze the change trend of the effective light digital signal according to a preset adaptive filtering algorithm, eliminate a tongue image background fluorescence digital signal value corresponding to the change trend, and obtain a tongue image fluorescence digital signal for tongue image feature recognition.

[0027] The present application also provides a tongue image feature intelligent recognition system. Since the system and the method belong to the same technical concept, solve the same technical problem, and should have the same beneficial effects, no further description is given here. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed in the embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0029] Figure 1 A flowchart of a tongue image feature intelligent recognition method provided by an embodiment of the present application is shown in the figure; Figure 2 The flowchart of step S2 provided for the embodiments of the present application; Figure 3 The flowchart of acquiring the tongue image fluorescence digital signal in S4 provided for the embodiments of the present application; Figure 4 The flowchart after step A2 provided for the embodiments of the present application; Figure 5 The flowchart before step S3 provided for the embodiments of the present application; Figure 6 The structural schematic diagram of a tongue image feature intelligent recognition system provided for the embodiments of the present application. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work belong to the scope of protection of the present application.

[0031] The embodiments of the present application are written in a progressive manner.

[0032] When the traditional existing tongue image feature intelligent recognition method is used for oral health detection, if there is a fixed frequency illumination noise signal in the detection environment, and the target tongue surface area is displaced to cause the background fluorescence signal to change, the existing processing mechanism cannot effectively avoid interference with the processing module used to identify and filter out the fixed frequency noise signal when correcting the changed background fluorescence signal, thereby causing the system to output a detection result that fuses environmental noise, a changed background signal, and calculation artifacts generated by internal processing conflicts. The occurrence of such composite error signals seriously affects the accuracy and reliability of tongue image feature recognition, limiting its application value in clinical diagnosis and health assessment.

[0033] Due to the unreliability of the detection result, repeated detection or additional verification may be required, increasing the complexity and time cost of the detection process, and reducing the overall efficiency and user experience of the system.

[0034] As shown in Figure 1 Based on this, a tongue image feature intelligent recognition method is provided, which is applied to an oral health detection scene and includes the following steps: S1. In a first preset time period, a preset standard light source acquires an original composite light signal of a target tongue surface area, and pre-processes the original composite light signal to obtain an original composite light digital signal; S2. The original composite light digital signal is subjected to fast Fourier transform analysis to obtain an environmental light digital signal; S3. In the second preset time period, the ambient light digital signal in the original composite light digital signal is eliminated to obtain an effective light digital signal; S4. When the target tongue surface area is displaced, causing the effective light digital signal to change, a preset adaptive filtering algorithm is used to analyze the change trend of the effective light digital signal, and the tongue image background fluorescence digital signal value corresponding to the change trend is eliminated to obtain a tongue image fluorescence digital signal for realizing tongue image feature recognition.

[0035] The preset standard light source in step S1 refers to a light source that provides stable, known spectrum and intensity of light during detection, which can be realized by using a specific wavelength LED array, a halogen lamp with a filter, or a laser diode. The original composite light signal refers to the raw light signal collected by the sensor from the target tongue surface area in the first preset time period without processing, which specifically includes the superposition of target fluorescence signal, ambient light signal, and tongue tissue background fluorescence signal. The preprocessing refers to the preliminary signal adjustment and conversion of the original composite light signal, which can be realized by using photoelectric conversion, analog-to-digital conversion, signal amplification, or noise suppression technology; The fast Fourier transform in step S2 refers to a mathematical algorithm for converting time domain signals to frequency domain signals, which can be realized by using the FFT library function integrated in the digital signal processor or general microcontroller. The ambient light digital signal refers to the digital signal component representing external environmental light interference identified and extracted from the original composite light digital signal by fast Fourier transform analysis; The preset adaptive filtering algorithm in step S4 refers to an algorithm that can dynamically adjust its filtering parameters according to the change trend of the input signal, which can be realized by using the sliding average algorithm, low-pass filtering algorithm, Kalman filtering, or least mean square algorithm. The tongue image background fluorescence digital signal value refers to the digital signal component of the background interference generated by the tongue tissue itself, which is not the fluorescence of the target bacterial group metabolites. Specifically, it presents different intensity and spectral characteristics with the change of the position of the target tongue surface area.

[0036] Steps S1 to S4 are to accurately identify and eliminate the fixed frequency of environmental light interference by fast Fourier transform analysis after obtaining the pre-processed original composite light signal, avoiding the confusion of environmental noise to subsequent signal processing. Secondly, when the target tongue surface area is displaced, causing the background fluorescence signal to change dynamically, the pre-set adaptive filtering algorithm can accurately identify and eliminate these changing tongue image background fluorescence digital signal values in real time. This step-by-step and adaptive processing strategy ensures that the removal of environmental light interference and the correction of background fluorescence signals are independent and do not interfere with each other, thereby avoiding the generation of internal processing conflicts and calculation artifacts described in the background technology. Finally, the present scheme can obtain highly pure and accurate tongue image fluorescence digital signals, significantly improving the accuracy and reliability of tongue feature recognition, and providing a more reliable data basis for oral health detection.

[0037] In some preferred embodiments, the application is implemented as follows. In the first preset time period, an array of multiple specific wavelength LEDs is used as the preset standard light source, for example, a LED array with a wavelength of 405 nm is used to excite the tongue surface, and a photodiode array or a CCD / CMOS image sensor is used to collect the original composite light signal of the target tongue surface area. The collected analog light signal is then amplified by a low-noise amplifier and input to a high-precision analog-to-digital converter for analog-to-digital conversion, thereby obtaining the original composite light digital signal. Then, the obtained original composite light digital signal is input into a digital signal processor, and the DSP runs a fast Fourier transform algorithm library. Through FFT analysis, specific frequency peaks present in the signal can be identified, for example, if the ambient lighting is 50 Hz alternating current, there will be obvious frequency peaks at 50 Hz and its harmonics, which are defined as ambient light digital signals. In the second preset time period, in order to eliminate these ambient light digital signals, a digital notch filter can be implemented in the DSP. The center frequency of the notch filter is accurately set to the frequency of the ambient light digital signal identified by the FFT analysis. The original composite light digital signal is filtered through the digital notch filter, thereby effectively removing the ambient light interference and outputting the effective light digital signal. When the target tongue surface area is displaced, for example, by detecting the change in tongue position through image recognition algorithm, or by judging the displacement by the significant change in the amplitude or spectral characteristics of the effective light digital signal, the system will start the preset adaptive filtering algorithm. This algorithm can specifically use a sliding average algorithm or a low-pass filtering algorithm to analyze the trend of the effective light digital signal. For example, by calculating the absolute value of the difference between the current effective light digital signal and the previous signal, and combining sliding average to smooth these changes, the amount of change in the background fluorescence signal caused by tongue displacement is identified. Based on this amount of change, the system can dynamically adjust the background fluorescence reference value, and subtract the corresponding tongue background fluorescence digital signal value from the current effective light digital signal, to finally obtain the pure tongue fluorescence digital signal for subsequent tongue feature recognition.

[0038] Preferably, in the first preset time period, the original composite light signal of the target tongue surface area is collected, and the original composite light signal is preprocessed to obtain the original composite light digital signal, including the following steps: In the first preset time period, the original composite light signal of the target tongue surface area is collected; The original composite light signal is photoelectrically converted to obtain an original composite electrical signal; The original composite electrical signal is analog-to-digital converted to obtain the original composite light digital signal.

[0039] Among them, the original composite optical signal refers to the original optical wave signal containing multiple optical components collected from the target tongue surface area, which can be the superposition of multiple optical signals such as excitation light source, ambient light, and tongue surface self-fluorescence; photoelectric conversion refers to the process of converting optical signals into electrical signals, which can be done by using photosensitive elements, such as photodiodes, photomultiplier tubes, or charge-coupled devices (CCDs), to convert the received photon energy into charge or current; the original composite electrical signal refers to the electrical signal corresponding to the original composite optical signal obtained after photoelectric conversion, which can be a voltage signal or a current signal; analog-to-digital conversion refers to the process of converting analog electrical signals into digital signals, which can be done by using an analog-to-digital converter (ADC) to sample and quantize continuously changing analog electrical signals at specific time points to generate discrete digital sequences.

[0040] The above steps are the specific implementation details of step S1, which involves converting the acquired raw composite optical signal from analog form into a raw composite optical digital signal suitable for subsequent digital processing and analysis. This conversion process overcomes the limitations of directly processing analog signals, providing a high-quality, standardized digital data foundation for subsequent signal denoising, feature extraction, and identification. This enables the effective implementation of complex digital signal processing algorithms, such as Fast Fourier Transform and adaptive filtering, thereby improving the accuracy and reliability of tongue image feature identification.

[0041] In some preferred embodiments, this application is implemented as follows: Within a first preset time period, a high-sensitivity optical sensor array, such as a CMOS image sensor or a photodiode array, can be used to acquire optical signals from the target tongue area, thereby obtaining raw composite optical signals. Subsequently, in order to convert these optical signals into electrical signals, a photoelectric conversion unit directly integrated with the optical sensor can be used, or the optical signals can be guided through an optical fiber to a separate photodiode to convert the optical signals into corresponding raw composite electrical signals, such as a voltage or current signal. Further, in order to enable these analog electrical signals to be recognized and analyzed by a digital processing unit, a high-resolution analog-to-digital converter (ADC) chip, such as a 16-bit or 24-bit ADC, can be used to sample and quantize the raw composite electrical signals, converting them into a series of discrete digital values, thereby obtaining raw composite optical digital signals. These digital signals can then be stored in a memory or directly transmitted to a microcontroller or digital signal processor (DSP) for subsequent complex algorithm processing.

[0042] like Figure 2 As shown, preferably, the ambient light digital signal is obtained by performing a fast Fourier transform analysis on the original composite optical digital signal, including the following steps: A1. In a preset data window, the original composite light digital signal is subjected to fast Fourier transform analysis to obtain original composite light frequencies of multiple sampling points; A2. The original composite light frequencies of the multiple sampling points are curve fitted to obtain an original composite light frequency curve; A3. It is judged whether a specific frequency in the original composite light frequency curve has a peak value; A4. If so, the specific frequency peak value is defined as the ambient light digital signal.

[0043] In step A1, the preset data window refers to a finite length data sequence intercepted from the continuous original composite light digital signal stream before fast Fourier transform analysis; the original composite light frequencies of multiple sampling points refer to the signal intensity or amplitude information obtained at different frequency points after the original composite light digital signal in the time domain is converted to the frequency domain by fast Fourier transform. These sampling points represent the energy distribution of the original signal at different frequencies; In step A2, curve fitting refers to smoothing the original composite light frequency data of multiple sampling points by mathematical methods such as least squares, spline interpolation or polynomial fitting, thereby generating a continuous frequency curve that can reflect the overall trend; the original composite light frequency curve refers to a graph or function that can continuously represent the intensity distribution of the original composite light signal at different frequencies after curve fitting of the original composite light frequencies of multiple sampling points; In step A3, the specific frequency refers to a frequency point or frequency range in the original composite light frequency curve where ambient light interference is expected or known to exist; the peak value refers to a signal intensity or amplitude at a specific frequency point in the original composite light frequency curve that is significantly higher than its adjacent frequency points, forming a local maximum value to indicate that there is a significant, concentrated energy component at the specific frequency, which is typically a typical feature of ambient light interference.

[0044] Steps A1 to A4 are specific implementation details of step S2, which is a fine frequency analysis of the original composite light digital signal to accurately identify and extract the ambient light digital signal. This solves the problem in the prior art that ambient light is not a single frequency, and simple filtering may lose effective signals and cannot accurately identify the ambient light frequency that needs to be filtered. Thus, it can avoid misjudging non-ambient light frequency as ambient light, reduce the loss of effective signals, and provide a clear and accurate filtering target for subsequent signal noise reduction, thereby improving the accuracy of tongue image feature recognition.

[0045] In some preferred embodiments, the application is implemented as follows. After the original composite light digital signal is collected, a preset data window can be set, for example, taking every 2048 consecutive digital sampling points as a data window. Fast Fourier transform (FFT) analysis is performed on the original composite light digital signal in the data window, for example, the Cooley-Tukey algorithm can be used to achieve it, thereby obtaining the original composite light frequency of 2048 sampling points. These sampling points represent the signal energy distribution in the range from direct current to the Nyquist frequency. In order to smooth these discrete frequency sampling points and identify the overall trend, a cubic spline interpolation method can be used to curve fit these sampling points, thereby generating a continuous original composite light frequency curve. After obtaining the frequency curve, it can be judged whether there is a peak value of a specific frequency. For example, a range of ambient light frequency can be preset, such as 48Hz to 52Hz (corresponding to 50Hz of power supply) or 58Hz to 62Hz (corresponding to 60Hz of power supply), and it is searched whether there is a significant local maximum value in this range. Whether the peak value exists can be determined by setting a threshold, for example, if the amplitude of a certain frequency point exceeds a certain multiple of the average value of the amplitudes of the surrounding frequency points, or exceeds a preset absolute amplitude, it is considered that there is a peak value. If a peak value meeting the conditions is detected near 50Hz or 60Hz, the frequency and amplitude corresponding to the peak value are defined as the ambient light digital signal. For example, if there is a significant peak value at 50Hz, 50Hz and its corresponding amplitude are determined as the ambient light digital signal for subsequent filtering operation. In this way, the main periodic light interference in the environment can be accurately identified, providing accurate basis for subsequent signal processing.

[0046] Preferably, in the second preset time period, the ambient light digital signal in the original composite light digital signal is removed to obtain the effective light digital signal, including the following steps: A first digital notch filter is set based on a fixed frequency equal to the ambient light digital signal; In the second preset period, the original composite light digital signal is input into the first digital notch filter; After being filtered by the first digital notch filter, the effective light digital signal is output.

[0047] The first digital notch filter refers to a digital signal processing device or algorithm module designed to produce a narrow attenuation band at a specific frequency point, thereby significantly reducing or eliminating signal components in this frequency range while minimizing the impact on signals at other frequencies. It can be implemented using an infinite impulse response (IIR) filter or a finite impulse response (FIR) filter structure, such as a Butterworth notch filter, Chebyshev notch filter, etc., with the purpose of accurately removing specific frequency interference signals; the fixed frequency refers to a pre-determined and unchanged frequency value during signal processing, which matches the frequency of the ambient light digital signal that needs to be filtered out. It can be the dominant frequency determined by spectral analysis of the ambient light digital signal; the second preset period refers to a pre-set time length, during which the original composite light digital signal is continuously input to the first digital notch filter for processing. It can be a fixed sampling time period, such as seconds or tens of seconds.

[0048] The above steps are the specific implementation details of step S3. Based on the fixed frequency of the ambient light digital signal identified in the above steps, a first digital notch filter is set. This notch filter is designed to form a deep attenuation notch at this specific frequency point, thereby selectively and accurately suppressing or eliminating interference components at this frequency. During the second preset period, the original composite light digital signal is input to the first digital notch filter. In a continuous period of time, all collected composite light signals will be processed by this specially designed filter. This process ensures that the persistent ambient light interference in the signal stream can be effectively captured and attenuated. The signal filtered by the first digital notch filter is output as an effective light digital signal. In this effective light digital signal, the original specific frequency ambient light interference component has been significantly weakened, while other frequency effective signal components have been preserved. In this way, the present scheme can separate the more pure effective light digital signal from the complex original composite light digital signal, providing high-quality input data for subsequent tongue image feature recognition, thereby improving the accuracy and reliability of recognition.

[0049] In some preferred embodiments, the application is implemented as follows: assuming that through fast Fourier transform analysis of the original composite light digital signal, it has been determined that the main interference frequency of the ambient light digital signal is 50 Hz. At this time, the system can set a first digital notch filter based on this fixed frequency of 50 Hz. This filter can be a digital IIR notch filter, for example, a second-order Butterworth notch filter design, with the center frequency accurately set to 50 Hz, and the bandwidth can be adjusted according to the actual spectrum width of the interference signal to ensure effective attenuation of the 50 Hz and its narrowband frequencies nearby. Subsequently, in the second preset period, for example, in the entire duration of each tongue image collection (for example, 5 seconds), the original composite light digital signal is input into the first digital notch filter in real time or quasi-real time. After the signal is processed by the filter, the 50 Hz ambient light interference component contained therein is greatly suppressed. Finally, the signal output from the first digital notch filter is the effective light digital signal. In this effective light digital signal, the influence of ambient light interference is minimized, so that the tongue fluorescence information contained therein is clearer, which is conducive to subsequent tongue feature analysis and recognition.

[0050] In the field of biomedical detection, especially in the aspect of oral health assessment, the detection of tongue flora metabolite characteristics has important application value. Such detection usually uses fluorescence imaging technology to obtain relevant information by exciting specific substances to emit their own fluorescence. However, in the actual operating environment, due to the existence of various interference factors, accurate acquisition and analysis of target fluorescence signals face many challenges. For example, the presence of background light interference in the detection environment, background fluorescence generated by the subject's own tissue, and unexpected associations that may occur in the internal processing logic of the device can all affect the accuracy and reliability of the detection results.

[0051] For example, in a detection workroom for oral health status assessment, a high-precision fluorescence imaging detection device is preparing to analyze the tongue flora metabolites of the subject. The working principle of the device is to emit light of a specific waveband through the built-in excitation light source, which irradiates the subject's tongue surface. Certain flora metabolites on the tongue surface, such as porphyrin substances, will emit their own characteristic fluorescence after excitation. The optical sensor array at the front end of the device is responsible for capturing these returned fluorescence signals and converting them into data streams, which are sent to the back-end processing unit for feature recognition and quantification.

[0052] At the start of the test, the subject opens his mouth as requested, and the operator places the device's probe over the tongue surface at a predetermined distance. The device projects a light beam onto the tongue surface, and the sensor array begins collecting data. At this point, the sensor receives not a pure signal, but a composite light signal. This signal contains three main components: first, the effective fluorescence signal from the target bacterial flora metabolites; second, the ambient light signal from the overhead lighting fixtures in the test room; and third, the background fluorescence signal from the subject's tongue tissue cells themselves. These three light signals are superimposed together as they enter the sensor.

[0053] The lighting system in the test room uses ordinary AC-powered fluorescent lamps. Due to slight voltage fluctuations in the power supply circuit, the fluorescent tubes produce a barely perceptible, fixed-frequency flicker of brightness. This flicker phenomenon causes the ambient light signal portion captured by the sensor to exhibit a high-frequency interference that fluctuates up and down at a fixed frequency. This interference signal enters the device's initial signal processing link along with the other light signals.

[0054] The device's back-end processing unit contains a module for filtering out background light interference. The design logic of this module is based on a premise that the background light intensity is constant or slowly varying. Therefore, its working mode is to calculate an average background light intensity value and subtract this constant value from the total signal. It is not designed to identify and process interference with a specific frequency pattern. As a result, the high-frequency interference signal from the lighting lamp flicker, due to its rapidly changing nature, is judged by this module as non-constant background light and thus passes through this processing link without being reduced, continuing to the next level of analysis module.

[0055] The signal enters the feature recognition module. The built-in rules of this module have a rule that interprets the rapid, periodic fluctuation of signal intensity as a biological indication of high activity of certain specific anaerobic bacterial flora metabolic activities. This interpretation model is derived from the observation under laboratory conditions of the rapid concentration changes of certain metabolites in the process of their generation and decomposition. Therefore, when the module receives that unfiltered high-frequency interference signal from the lamp flicker, its data pattern exactly matches this recognition rule. The processing logic mistakenly identifies this purely physical interference signal as the presence of a highly active metabolite and quantifies it, outputting an abnormally high reading on the display interface.

[0056] After the operator observes this abnormal reading that does not conform to routine experience, his professional judgment prompts him to perform a standard review procedure: fine-tune the distance and angle of the probe to the patient's tongue surface to confirm the accuracy of the test area and try to reproduce the result. This operational behavior causes a slight displacement of the probe's optical field of view on the tongue surface.

[0057] The tongue surface is not a homogeneous plane. The thickness of the tongue coating, the state of the epithelial cells and the distribution of the blood are different in different areas, which leads to the fact that the background fluorescence intensity and the spectral composition of the tongue surface itself are different at different positions. The fine adjustment of the probe position by the operator directly changes the tongue surface area collected by the sensor, thereby changing the intensity and spectral shape of the background fluorescence signal part of the composite light signal.

[0058] This change in background fluorescence triggers the response of the calibration module in the processing unit. The calibration module contains two independent correction programs: Program A, which is responsible for calculating and deducting the reference value of the background fluorescence in real time; and Program B, which is responsible for identifying and filtering high-frequency interference. When Program A starts to recalculate the reference value because it detects a change in the background fluorescence signal, an internal parameter adjustment process in Program A has an unexpected impact on the operation of Program B. Specifically, the reference value adjustment logic of Program A temporarily changes a gain parameter of the entire signal processing link, and this gain parameter is also the threshold basis used by Program B to judge the amplitude of high-frequency interference. This internal processing association causes the interference identification threshold of Program B to be accidentally raised, making it less sensitive to flicker interference from the light.

[0059] Finally, the data output by the device becomes unreadable. It is a result of multiple errors superimposed: it contains both the original high-frequency interference from the lighting device flicker, mixed with the background fluorescence of the tongue surface that changes due to the movement of the probe, and on top of that, the calculation artifact caused by the logical association of the two programs in the calibration module. The operator finds that no matter how the device position is adjusted again, this error cannot be eliminated, and only another reading with a different form but the same error can be obtained. The system enters a complex error detection state caused by the chain reaction of external environmental physical factors, the operator's regular behavior, and the internal data processing logic defects of the device, which cannot be decomposed and identified by the existing detection process.

[0060] Preferably, when the target tongue surface area is displaced, causing the effective light digital signal to change, the change trend of the effective light digital signal is analyzed according to a preset adaptive filtering algorithm, including the following steps: When the target tongue surface area is displaced, causing the first effective light digital signal to change to a second effective light digital signal; According to the sliding average algorithm or the low-pass filtering algorithm, the absolute value of the difference between the second effective light digital signal and the first effective light digital signal is calculated as the effective light digital signal change.

[0061] The moving average algorithm refers to a commonly used digital signal processing technique, which can smooth the data by calculating the average value of a fixed length sub-sequence in the data sequence, and the sub-sequence slides on the data sequence as new data arrives. The low-pass filter algorithm refers to a signal processing filter, which can pass signal components below a certain cutoff frequency while attenuating signal components above the cutoff frequency. The effective light digital signal change quantity refers to the quantized value obtained by calculating the absolute value of the difference between the effective light digital signals before and after displacement. It can be obtained by subtracting the second effective light digital signal from the first effective light digital signal and taking the absolute value of the result.

[0062] The above steps are the specific implementation details of analyzing the change trend of the effective light digital signal in step S4. By quantifying the change of the effective light digital signal when the target tongue surface area is displaced, the basis for subsequent adaptive rejection of tongue image background fluorescence digital signal values is provided. Specifically, when the target tongue surface area is displaced, the effective light digital signal changes from the first effective light digital signal to the second effective light digital signal. This change includes the change of the tongue image feature itself and the background fluorescence change caused by displacement. In order to distinguish between the two changes, the sliding average algorithm or low-pass filter algorithm is used to process these signals. These algorithms can smooth the signal and reduce transient noise interference, thereby capturing the overall change trend of the signal. By calculating the absolute value of the difference between the second effective light digital signal and the first effective light digital signal, a quantized effective light digital signal change quantity can be obtained. This change quantity directly reflects the signal amplitude change caused by the displacement of the tongue surface area, and excludes the influence of the signal direction. This quantization method enables the system to identify and distinguish the background fluorescence change caused by the displacement of the tongue surface, rather than judging it as a change of the tongue image feature itself.

[0063] In some preferred embodiments, the present application is implemented as follows: it is assumed that during the tongue image feature recognition process, the system continuously acquires effective light digital signals of the target tongue surface area. When the operator fine-tunes the probe position, causing displacement of the target tongue surface area, the system detects changes in the effective light digital signals. For example, before the displacement occurs, the system records a series of effective light digital signals, and defines one value or the average value over a period of time as the first effective light digital signal. After the displacement occurs, the system continues to acquire new effective light digital signals, and defines one value or the average value over a period of time as the second effective light digital signal. In order to quantify this change, the system can use a sliding average algorithm or a low-pass filtering algorithm for processing. For example, a sliding average window of length N can be used to smooth the second effective light digital signal sequence, or a second-order Butterworth low-pass filter can be used to filter the signal. The processed signal can reflect the trend of background fluorescence changes caused by displacement. Subsequently, the system calculates the absolute value of the difference between the second effective light digital signal and the first effective light digital signal. For example, if the first effective light digital signal is S1 and the second effective light digital signal is S2, the effective light digital signal change quantity can be calculated as |S2 - S1|. This calculation result, i.e. the effective light digital signal change quantity, can be used as an input parameter for the subsequent adaptive filtering algorithm, to adjust the rejection of tongue image background fluorescence digital signal values, so as to ensure that the final tongue image fluorescence digital signal can reflect the characteristics of the tongue image itself, even in the case of tongue surface displacement, the accuracy of recognition can be maintained.

[0064] As shown in Figure 3 , preferably, after obtaining the effective light digital signal change quantity, the tongue image background fluorescence digital signal values corresponding to the change trend are rejected to obtain the tongue image fluorescence digital signal, including the following steps: B1. Based on the correspondence between the target tongue surface area and the tongue image background fluorescence digital signal, the effective light digital signal change quantity is taken as the tongue image background fluorescence digital signal change quantity; B2. According to the first tongue image background fluorescence digital signal in the first effective light digital signal and the tongue image background fluorescence digital signal change quantity, the second tongue image background fluorescence digital signal is obtained as the tongue image background fluorescence digital signal value; B3. Calculate the difference between the second effective light digital signal and the tongue image background fluorescence digital signal value as the tongue image fluorescence digital signal.

[0065] Among them, the correspondence between the target tongue surface area and the tongue image background fluorescence digital signal in step B1 refers to the mapping relationship reflecting the inherent background fluorescence signal characteristics of different positions or areas of the tongue surface, which can be specifically constructed by multi-point scanning and data acquisition on healthy people or standard tongue models to construct a tongue surface background fluorescence spectrum or database; The first tongue image background fluorescence digital signal in step B2 refers to the numerical value of the tongue surface background fluorescence signal separated or estimated from the effective light digital signal before the target tongue surface area is displaced or at an initial reference time. Specifically, the numerical value of the tongue surface background fluorescence signal can be measured at a fixed reference point on the tongue surface at the beginning of detection. The tongue image background fluorescence digital signal change refers to the value of the change in the tongue surface background fluorescence signal intensity or characteristics caused by the displacement of the target tongue surface area. Specifically, the overall change trend of the effective light digital signal before and after the displacement of the tongue surface can be analyzed, and a preset corresponding relationship model can be used to calculate the change. The second tongue image background fluorescence digital signal refers to the numerical value of the tongue surface background fluorescence signal calculated from the first tongue image background fluorescence digital signal and the tongue image background fluorescence digital signal change after the displacement of the target tongue surface area or at the current time. Specifically, the initial background fluorescence signal can be superimposed or corrected with the calculated background fluorescence change.

[0066] Steps B1 to B3 are the specific implementation details of obtaining the tongue image fluorescence digital signal in step S4. By establishing a corresponding relationship between the effective light digital signal change and the tongue image background fluorescence digital signal, and combining the initial background fluorescence signal to dynamically obtain the current background fluorescence value, the accurate estimation and elimination of the tongue image background fluorescence signal are achieved. This enables the acquisition of pure and accurate tongue image fluorescence digital signals even when the target tongue surface area is displaced, providing a reliable data basis for subsequent tongue image feature recognition and improving the accuracy of tongue image detection.

[0067] In some preferred embodiments, specifically, when the tongue image detection device is scanning the target tongue surface area, if the operator fine-tunes the probe position, causing the effective light digital signal collected by the sensor to change from a first effective light digital signal to a second effective light digital signal, and the effective light digital signal change amount is calculated by the sliding average algorithm. At this time, the system will first call the preset tongue surface background fluorescence spectrum, which records the background fluorescence intensity distribution of different areas of the tongue surface. Based on the change of the current probe position, the system maps this effective light digital signal change amount to the tongue image background fluorescence digital signal change amount. For example, if the probe moves from the tip of the tongue to the middle of the tongue, the system will determine according to the spectrum that this displacement usually causes the background fluorescence signal to increase or decrease by a certain amount. Then, assuming that the system has obtained the initial background fluorescence signal of the current detection area before the probe moves, i.e., the first tongue image background fluorescence digital signal. The system will superimpose this first tongue image background fluorescence digital signal with the tongue image background fluorescence digital signal change amount just determined, thereby dynamically calculating the tongue image background fluorescence digital signal value at the current position, i.e., the second tongue image background fluorescence digital signal. Finally, the system subtracts the second effective light digital signal collected at the current position from the dynamically calculated second tongue image background fluorescence digital signal value, thereby obtaining the tongue image fluorescence digital signal with the background fluorescence interference removed. For example, if the second effective light digital signal is 1000 units, and the dynamically estimated background fluorescence digital signal value is 300 units, then the final tongue image fluorescence digital signal obtained is 700 units. In this way, even if the tongue surface is displaced, the system can accurately separate the true tongue image fluorescence signal, avoiding the interference of background fluorescence on the detection result.

[0068] As Figure 4 shown, preferably, after fitting the original composite light frequency of a plurality of sampling points to obtain the original composite light frequency curve, the following steps are further included: C1. Record all fixed frequencies and corresponding periodic characteristics and waveform stability; C2. Calculate the similarity of each fixed frequency and the corresponding periodic characteristics and waveform stability with the preset biological criteria respectively to obtain the confusion risk index of each fixed frequency; C3. Prioritize each fixed frequency according to the confusion risk index from high to low; C4. Set the second digital notch filter based on the fixed frequency with the highest priority of the confusion risk index; C5. Input the original composite light digital signal into the second digital notch filter within the second preset period; C6. Output the effective light digital signal after filtering by the second digital notch filter.

[0069] The fixed frequency in step C1 refers to a frequency component with persistent existence characteristics identified by fast Fourier transform analysis in the original composite light frequency curve, which can represent environmental light interference or potential biologically relevant signals. The fixed frequency refers to a frequency component with persistent existence characteristics identified by fast Fourier transform analysis in the original composite light frequency curve, which can represent environmental light interference or potential biologically relevant signals. The waveform stability refers to the degree of change in the waveform shape and amplitude of the fixed frequency signal within a certain time window, which can be measured by calculating the root mean square error or standard deviation of the signal waveform to measure its stability. The preset biological criteria in step C2 refers to a set of predefined standards or models for judging whether the fixed frequency signal is related to biological characteristics, which can include frequency range, periodic pattern, waveform characteristics, and relevance to known biological processes of biological signals. Similarity calculation refers to the process of comparing and matching the recorded fixed frequency and corresponding periodic characteristics and waveform stability with the preset biological criteria, which can use correlation coefficient, Euclidean distance, or machine learning classification algorithm to quantify the similarity between the two. The confusion risk index refers to an index that quantifies the risk of each fixed frequency signal being misjudged as non-biological interference and filtered out, which can be a value between 0 and 1, with a high value indicating that the signal may carry biological information. The second digital notch filter in step C4 refers to the same kind of digital signal processing device as the first digital notch filter, which is used to filter out or attenuate signals in a specific frequency range, which can be achieved by designing a finite impulse response (FIR) or infinite impulse response (IIR) filter. The second preset period in step C5 refers to a pre-set time length, within which the original composite light digital signal is input to the second digital notch filter for processing. Steps C1 to C5 are specific implementation details after step A2, which record the periodic characteristics and waveform stability of the fixed frequency, and perform similarity calculation with the preset biological criteria, to obtain the confusion risk index of each fixed frequency, thereby identifying signals that may carry biological information. Based on the confusion risk index, priority is sorted, and the second digital notch filter is selectively set for filtering, which can avoid filtering out all fixed frequency signals with biological significance, prevent information loss, and distinguish and process different types of fixed frequency signals, so that the final effective light digital signal can reflect the tongue image characteristics, improving the accuracy of tongue image feature recognition.

[0070] In some preferred embodiments, after the system completes the fast Fourier transform analysis of the original composite light digital signal and curve fitting of the original composite light frequency curve, the system can automatically identify and record all fixed frequency components in the curve, for example, identify multiple fixed frequencies of 50 Hz, 60 Hz, 120 Hz, etc. through a peak detection algorithm. For each identified fixed frequency, the system can further analyze its periodicity characteristics, for example, determine the period by calculating the peak position of its autocorrelation function, and evaluate its waveform stability, for example, quantify the consistency of the waveform by calculating the root mean square error in the continuous time window. Subsequently, the system can calculate the similarity of the feature data of these fixed frequencies, such as the periodicity characteristic value and waveform stability value of the 50 Hz signal, with the preset biological criteria. The preset biological criteria can be a biological signal feature template library stored in a database, which contains information such as frequency, periodicity, waveform stability, etc. of the biological signal related to tongue image features. The similarity calculation can use the cosine similarity algorithm to calculate the similarity of each fixed frequency feature vector with each template vector in the biological criteria, thereby obtaining the confusion risk index of each fixed frequency, which can be a percentage value from 0 to 100, indicating the degree of similarity with the biological signal. Then, the system can prioritize all fixed frequencies from high to low according to these confusion risk indexes, for example, if the confusion risk index of the 50 Hz signal is 90 and that of the 60 Hz signal is 20, the priority of the 50 Hz signal is high. Finally, the system can set a second digital notch filter based on the fixed frequency with the highest confusion risk index, such as 50 Hz, which can be a digital IIR filter with a notch frequency set to 50 Hz. In the second preset period, for example, in the next 5 seconds, the original composite light digital signal is input into the second digital notch filter for filtering, and the output signal after the filter processing is the effective light digital signal, in which the fixed frequency components with high confusion risk have been selectively removed.

[0071] As shown in Figure 5 , preferably, before the ambient light digital signal in the original composite light digital signal is removed to obtain the effective light digital signal in the second preset time period, the following steps are included: D1. The preset first light source linearly scans the tongue tip to the tongue root at a preset fixed frequency to obtain a third tongue background fluorescence signal, and obtains a third tongue background fluorescence digital signal after preprocessing; D2. Fit the third tongue background fluorescence digital signal to generate a third tongue background fluorescence intensity curve as a tongue background fluorescence position reference map.

[0072] The first light source in step D1 refers to a specific light source used to excite the tongue surface background fluorescence, which can be implemented by using an LED light source, a laser diode, or a wide-spectrum lamp, etc. The preset fixed frequency refers to the working frequency or scanning rate of the light source that remains unchanged during the scanning process, which can be implemented by using a timer control, a pulse generator, or a stepper motor drive, etc. Linear scanning refers to the linear movement of the light source or sensor along the tongue surface from the tip to the root direction to cover the entire tongue surface area, which can be implemented by using a mechanical guide rail system, an electric sliding table, or an optical scanning galvanometer, etc. The third tongue image background fluorescence signal refers to the fluorescence response signal generated by the tongue surface tissue under the excitation of the first light source, which can be acquired by using a photodetector, a spectrometer, or a fluorescence sensor, etc. Preprocessing refers to the preliminary data processing of the original acquired fluorescence signal to eliminate noise, correct the baseline, or amplify the signal, which can be implemented by using filtering, gain adjustment, or analog-to-digital conversion, etc. The third tongue image background fluorescence digital signal refers to the tongue surface background fluorescence intensity data represented in digital form after preprocessing, which can be processed by using a digital signal processor, a microcontroller, or an application-specific integrated circuit, etc. The fitting in step D2 refers to finding a function or curve to approximate the discrete digital signal data points by mathematical methods to reveal the inherent trend or rule, which can be implemented by using polynomial fitting, spline interpolation, or least squares method, etc. The third tongue image background fluorescence intensity curve refers to the continuous curve obtained after fitting, reflecting the background fluorescence intensity distribution of different positions on the tongue surface, which can be stored and represented by using a data storage, a graphical display, or a database, etc. The tongue image background fluorescence position reference map refers to the third tongue image background fluorescence intensity curve as a kind of standard or benchmark for subsequent processing of tongue surface background fluorescence correction or compensation reference data set, which can be constructed in the form of a lookup table, a two-dimensional array, or an image file, etc. Steps D1 to D2 are specific implementation details before step S3, which is to generate a tongue image background fluorescence position reference map by pre-acquiring and modeling the background fluorescence intensity distribution of different positions on the tongue surface. This enables subsequent signal processing to be referenced and corrected according to the actual background fluorescence intensity of the tongue surface area, avoiding signal processing deviation caused by uneven tongue surface background fluorescence. Therefore, this scheme can eliminate the interference of background fluorescence on the target signal, improving the accuracy of tongue image feature recognition.

[0073] In some embodiments, the application is implemented as follows. In a second preset time period, before the ambient light digital signal in the original composite light digital signal is removed to obtain the effective light digital signal, a preset ultraviolet LED can be used as the first light source, and its emission wavelength can be set at about 365 nanometers to excite the background fluorescence of the tongue tissue. The ultraviolet LED can emit light in a pulsed manner at a preset fixed frequency of 100 times per second. Meanwhile, the first light source can be installed on a linear slide rail driven by a stepper motor, so that it can perform linear scanning from the tip of the tongue to the root of the tongue at a speed. During the scanning process, a photodiode array can receive the third tongue image background fluorescence signal emitted by the tongue surface in real time. These analog signals can be amplified by an amplifier and digitized by a 12-bit analog-to-digital converter, thereby obtaining the third tongue image background fluorescence digital signal. Subsequently, these digital signals can be input into a microcontroller, which can run a polynomial fitting algorithm based on the least squares method, such as a third-order polynomial fitting, to smooth the digital signal points and generate a third tongue image background fluorescence intensity curve. This curve can be stored in a lookup table as a tongue image background fluorescence position reference map for subsequent signal processing modules to query and use.

[0074] Preferably, after obtaining the tongue image background fluorescence position reference map, the following steps are further included: D3. The preset standard light source performs linear scanning from the tip of the tongue to the root of the tongue at a preset fixed frequency to obtain the effective light digital signal; D4. According to the current position during the linear scanning process, the tongue image background fluorescence position reference map is retrieved to obtain the tongue image background fluorescence reference value; D5. The difference between the effective light digital signal and the tongue image background fluorescence reference value is calculated as the compensated tongue image fluorescence digital signal.

[0075] The preset standard light source in step D3 refers to a light source with output characteristics and spectral distribution, which can be implemented by using an LED array, a laser diode, or a halogen lamp; the effective light digital signal refers to a digital signal that reflects the fluorescence information of the target tongue surface area after the ambient light is filtered out; The tongue image background fluorescence position reference map in step D4 refers to a previously established data set for characterizing the background fluorescence intensity distribution of the tongue surface position, which can be generated by scanning a large number of samples and performing data fitting; the tongue image background fluorescence reference value refers to the corresponding background fluorescence intensity value retrieved from the tongue image background fluorescence position reference map according to the current scanning position; The compensated tongue image fluorescence digital signal in step D5 refers to the tongue image fluorescence signal that has been removed from the influence of background fluorescence, which is obtained by calculating the difference between the real-time acquired effective light digital signal and the tongue image background fluorescence reference value; Steps D3 to D4 are specific implementation details after step D2, which realizes dynamic compensation of the tongue fluorescence signal by combining the real-time collected effective light digital signal with the pre-established tongue background fluorescence position reference map. The previous scheme provides a reference standard for the background fluorescence position, and on this basis, the present scheme further utilizes the reference standard to dynamically correct the real-time collected effective light digital signal, so that the final tongue fluorescence digital signal can accurately reflect the target fluorescence information, avoiding signal distortion caused by changes in background fluorescence, thereby improving the accuracy and reliability of tongue feature recognition.

[0076] In some embodiments, the present application is implemented as follows. When performing tongue feature recognition, first, a preset standard light source composed of multiple LEDs can be used, which linearly scans the subject's tongue tip to root area at a preset fixed frequency of, for example, 100 Hz, through a linear scanning mechanism driven by a stepper motor. During the scanning process, a photoelectric sensor array collects tongue surface reflection and fluorescence signals in real time, and after photoelectric conversion, analog-to-digital conversion and ambient light filtering, the effective light digital signal at the current position is obtained. At the same time, the system internally stores a pre-established tongue background fluorescence position reference map, which can be a two-dimensional lookup table, with the horizontal axis representing the tongue surface scanning position and the vertical axis representing the background fluorescence intensity reference value at that position. When the linear scanning mechanism moves to a certain position on the tongue surface, the system will quickly retrieve the tongue background fluorescence reference value corresponding to that position in the tongue background fluorescence position reference map according to the coordinate information of that position. Subsequently, a digital signal processor or microcontroller receives the real-time collected effective light digital signal and the tongue background fluorescence reference value retrieved from the map. The processor performs a subtraction operation, i.e. subtracts the tongue background fluorescence reference value corresponding to the current position from the effective light digital signal at that position, and takes the calculation result as the compensated tongue fluorescence digital signal. For example, if the effective light digital signal at a certain position is 500 units and the tongue background fluorescence reference value at that position is 150 units, the compensated tongue fluorescence digital signal is 350 units. This compensated signal can be further used for analysis and recognition of tongue features, such as evaluating oral health status by analyzing its intensity, distribution or spectral characteristics.

[0077] As shown in Figure 6 A tongue feature intelligent recognition system, comprising: A collection and preprocessing module for collecting the original composite light signal of the target tongue surface area by the preset standard light source within a first preset time period, and preprocessing the original composite light signal to obtain the original composite light digital signal; A signal processing module for performing fast Fourier transform analysis on the original composite light digital signal to obtain the ambient light digital signal; The signal denoising module is configured to remove the ambient light digital signal in the original composite light digital signal to obtain an effective light digital signal within a second preset time period. The signal screening module is configured to analyze a change trend of the effective light digital signal according to a preset adaptive filtering algorithm when the target tongue surface region is displaced to cause the effective light digital signal to change, remove the tongue image background fluorescence digital signal value corresponding to the change trend, and obtain a tongue image fluorescence digital signal for tongue image feature recognition.

[0078] The acquisition preprocessing module is a unit responsible for initial acquisition and digital conversion of the light signal, which can be implemented by using hardware components such as an integrated optical sensor, an analog-to-digital converter, and a data buffer. The signal processing module is a unit for frequency domain analysis of the digitized original composite light signal, which can be implemented by using a digital signal processor or a general-purpose processor to perform a fast Fourier transform algorithm. The signal denoising module is a unit for removing specific interference signals from the original composite light digital signal, which can be implemented by using a digital filter such as a notch filter or a band-stop filter. The signal screening module is a unit for adaptive correction of the signal when the signal is affected by target displacement, which can be implemented by using a microcontroller or an application-specific integrated circuit to perform an adaptive filtering algorithm.

[0079] The scheme of the present application embodies each logical step in the tongue image feature intelligent recognition method as an independent system module. Through modular design, the system allocates acquisition preprocessing, signal processing, signal denoising, and signal screening functions to different units, so that the tongue image feature intelligent recognition method can be effectively and reliably executed.

[0080] In the embodiments provided in the present application, it should be understood that the disclosed method and system can be implemented in other ways. The system embodiments described above are only illustrative. For example, the division of modules is only a logical function division, and actual implementation can have another division manner, such as: multiple modules or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed components can be indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.

[0081] In addition, each functional module in each embodiment of the present application can be integrated in one processor, or each module can be a separate device, or two or more modules can be integrated in one device. Each functional module in each embodiment of the present application can be implemented in the form of hardware or hardware plus software functional units.

[0082] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instructions and related hardware. The aforementioned program instructions can be stored in a computer readable storage medium. When the program instructions are executed, the steps of the above-mentioned method embodiments are executed. The aforementioned storage medium includes a mobile storage device, a read only memory (ROM), a magnetic disc or an optical disc, and various media that can store program codes.

[0083] Hereinafter, the terms "first" and "second" are only used for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first" and "second" can explicitly or implicitly include one or more of the features.

[0084] If a flowchart is used in the present application, the flowchart is used to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or subsequent operations are not necessarily executed in sequence. On the contrary, each step can be processed in reverse order or simultaneously. Meanwhile, other operations can be added to these processes, or one or more steps of operations can be removed from these processes.

[0085] The tongue feature intelligent recognition method and system provided by the present application are described in detail above. The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to the embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A tongue feature intelligent recognition method applied to an oral health detection scene, characterized in that, The method comprises the following steps: In a first preset time period, a preset standard light source is used to collect an original composite light signal of a target tongue surface area, the original composite light signal is preprocessed, and an original composite light digital signal is obtained; Fast Fourier transform analysis is performed on the original composite light digital signal, and an ambient light digital signal is obtained; In a second preset time period, the ambient light digital signal in the original composite light digital signal is removed, and an effective light digital signal is obtained; When the target tongue surface area is displaced, causing the effective light digital signal to change, the change trend of the effective light digital signal is analyzed according to a preset adaptive filtering algorithm, the tongue background fluorescence digital signal value corresponding to the change trend is removed, and a tongue fluorescence digital signal is obtained for realizing tongue feature recognition.

2. The tongue feature intelligent recognition method of claim 1, wherein, In the first preset time period, a preset standard light source is used to collect an original composite light signal of a target tongue surface area, the original composite light signal is preprocessed, and an original composite light digital signal is obtained, which comprises the following steps: In a first preset time period, an original composite light signal of a target tongue surface area is collected; Photoelectric conversion is performed on the original composite light signal, and an original composite electric signal is obtained; Analog-digital conversion is performed on the original composite electric signal, and the original composite light digital signal is obtained.

3. The tongue feature intelligent recognition method of claim 1, wherein, The fast Fourier transform analysis on the original composite light digital signal to obtain the ambient light digital signal comprises the following steps: In a preset data window, fast Fourier transform analysis is performed on the original composite light digital signal to obtain the original composite light frequency of a plurality of sampling points; The original composite light frequency of a plurality of sampling points is curve-fitted to obtain an original composite light frequency curve; It is judged whether a specific frequency in the original composite light frequency curve has a peak value; If so, the specific frequency peak value is defined as the ambient light digital signal.

4. The tongue feature intelligent recognition method of claim 3, wherein, In the second preset time period, the ambient light digital signal in the original composite light digital signal is removed to obtain the effective light digital signal, which comprises the following steps: A first digital notch filter is set based on a fixed frequency equal to the ambient light digital signal; In a second preset period, the original composite light digital signal is input into the first digital notch filter; The effective light digital signal is output after being filtered by the first digital notch filter.

5. The tongue feature intelligent recognition method of claim 1, wherein, When the target tongue surface area is displaced, causing the effective light digital signal to change, the change trend of the effective light digital signal is analyzed according to a preset adaptive filtering algorithm, which comprises the following steps: When the target tongue surface area is displaced, causing the first effective light digital signal to change into a second effective light digital signal; According to a sliding average algorithm or a low-pass filtering algorithm, the difference absolute value between the second effective light digital signal and the first effective light digital signal is calculated as an effective light digital signal change value.

6. The tongue feature intelligent recognition method of claim 5, wherein, After obtaining the effective light digital signal change value, the tongue background fluorescence digital signal value corresponding to the change trend is removed to obtain the tongue fluorescence digital signal, which comprises the following steps: According to the correspondence between the target tongue surface area and the tongue image background fluorescence digital signal, the effective light digital signal change amount is taken as a tongue image background fluorescence digital signal change amount; According to the first tongue image background fluorescence digital signal in the first effective light digital signal and the tongue image background fluorescence digital signal change amount, a second tongue image background fluorescence digital signal is obtained as the tongue image background fluorescence digital signal value; A difference between the second effective light digital signal and the tongue image background fluorescence digital signal value is calculated as the tongue fluorescence digital signal.

7. The tongue feature intelligent recognition method of claim 3, wherein, After the curve fitting of the original composite light frequency of the plurality of sampling points obtains an original composite light frequency curve, the following steps are further included: Record all fixed frequencies and corresponding periodic characteristics and waveform stability; Respectively, similarity calculation is performed on the fixed frequencies and corresponding periodic characteristics and waveform stability and preset biological criteria to obtain a confusion risk index of each fixed frequency; According to the confusion risk index, each fixed frequency is prioritized from high to low; Based on the fixed frequency with the highest priority of the confusion risk index, a second digital notch filter is set; In a second preset period, the original composite light digital signal is input into the second digital notch filter; After filtering by the second digital notch filter, the effective light digital signal is output.

8. The tongue feature intelligent recognition method of claim 1, wherein, Before the original composite light digital signal is obtained by eliminating the ambient light digital signal in the original composite light digital signal in the second preset time period, the following steps are further included: A preset first light source linearly scans the tongue tip to the tongue root at a preset fixed frequency to obtain a third tongue image background fluorescence signal, and a third tongue image background fluorescence digital signal is obtained after preprocessing; The third tongue image background fluorescence digital signal is fitted to generate a third tongue image background fluorescence intensity curve as a tongue image background fluorescence position reference map.

9. The tongueprint feature intelligent recognition method of claim 8, wherein, After obtaining the tongue image background fluorescence position reference map, the following steps are further included: The preset standard light source linearly scans the tongue tip to the tongue root at a preset fixed frequency to obtain the effective light digital signal; According to the current position in the linear scanning process, the tongue image background fluorescence position reference map is searched to obtain a tongue image background fluorescence reference value; The difference between the effective light digital signal and the tongue image background fluorescence reference value is calculated as the tongue fluorescence digital signal after compensation.

10. A tongue feature intelligent recognition system, characterized in that, It includes: The acquisition and preprocessing module is used for collecting the original composite light signal of the target tongue surface area by the preset standard light source in the first preset time period, and preprocessing the original composite light signal to obtain the original composite light digital signal; The signal processing module is used for performing fast Fourier transform analysis on the original composite light digital signal to obtain the ambient light digital signal; The signal noise reduction module is used for eliminating the ambient light digital signal in the original composite light digital signal in the second preset time period to obtain the effective light digital signal; The signal screening module is configured to, when the target tongue surface area is displaced and the effective light digital signal is changed, analyze a change trend of the effective light digital signal according to a preset adaptive filtering algorithm, eliminate a tongue image background fluorescence digital signal value corresponding to the change trend, and obtain a tongue image fluorescence digital signal for realizing tongue image feature recognition.

Images