Calibration method and calibration device for thermal reflectance coefficient for reflectance thermography

Abstract

This invention belongs to the field of reflectivity thermal imaging technology and provides a method and apparatus for calibrating thermal reflectivity coefficients. The method simultaneously acquires the light intensity sequence of the reflectivity thermal imaging region of the sample under test, the acoustic response of the phase transition reference material, and the magnetic relaxation response to obtain multimodal synchronous data; extracts reflectivity evolution features, acoustic phase transition features, and magnetic relaxation phase transition features to generate a phase transition identification feature set; determines the candidate time window for phase transitions based on the acoustic phase transition features and correlates it with the reflectivity evolution features and magnetic relaxation phase transition features to determine the phase transition reference time and establish an absolute temperature mapping relationship; and inverts and corrects the full-field thermal reflectivity coefficient to obtain the calibration result. This method improves the calibration accuracy and temperature measurement reliability in non-uniform material regions.

Description

Technical Field

[0001] This invention relates to the field of reflectivity thermal imaging technology, and in particular to a method and apparatus for calibrating the thermal reflectivity coefficient for reflectivity thermal imaging. Background Technology

[0002] Reflectivity thermal imaging is a non-contact temperature measurement technique that utilizes the change in surface reflectivity of a material with temperature. This technique typically uses an optical imaging system to acquire changes in reflected light intensity from the surface of the sample under test, and combines this with a pre-calibrated thermal reflectivity coefficient to convert the reflectivity change into a temperature change, thereby obtaining the temperature field distribution of the sample surface. Due to its advantages such as high spatial resolution, fast response speed, and the elimination of the need for contact temperature sensors in the test area, reflectivity thermal imaging is widely used in semiconductor chip thermal analysis, temperature field measurement of micro / nano devices, thermal response testing of thin film materials, material performance evaluation, thermal management design, and non-destructive testing.

[0003] In reflectivity thermal imaging, the accuracy of the thermal reflectivity coefficient directly affects the reliability of temperature inversion results. Existing thermal reflectivity coefficient calibration methods typically require the use of thermocouples, platinum resistance thermometers, hot-stage temperature sensors, or other external temperature measurement devices to establish a correspondence between external temperature readings and changes in optical reflectivity. However, in micro / nano devices, semiconductor chips, thin-film devices, and multi-material composite structures, there is often a spatial deviation between the external temperature measurement point and the actual reflectivity thermal imaging area. Furthermore, contact temperature measurement devices may alter the local thermal field distribution, making it difficult for external temperature readings to accurately represent the true temperature state of the measured area. Therefore, existing methods struggle to establish a stable and reliable absolute temperature reference without relying on external contact temperature measurements.

[0004] On the other hand, actual test samples typically possess multi-material, multi-layer film, multi-interface, and locally non-uniform structures, with differences in material composition, surface morphology, reflectivity, and thermal response processes across different regions. Traditional methods often use thermal reflectivity coefficients obtained from single-point or a few-point calibration as uniform coefficients for the entire imaging area, which fails to reflect the differences in thermal reflectivity response at different spatial locations. Furthermore, when relying solely on changes in optical reflection for calibration, the reflected light intensity is easily affected by ambient light disturbances, light source fluctuations, instrument noise, thermal radiation crosstalk, focal plane drift, sample surface roughness, scratches, oxide layers, and contaminant particles, making it difficult to distinguish the true thermal response from interference signals, further impacting the accuracy of the overall thermal reflectivity coefficient inversion results.

[0005] Therefore, existing methods for calibrating thermal reflectivity coefficients for reflectivity thermal imaging have the following shortcomings: First, they rely on external contact temperature measurement or single-point temperature readings, making it difficult to establish an accurate absolute temperature reference for micro-nano scale reflectivity thermal imaging; second, they are difficult to overcome the problems of material inhomogeneity and susceptibility to optical signal interference at the same time, resulting in insufficient accuracy and reliability of the full-field thermal reflectivity coefficient inversion results.

[0006] Therefore, it is necessary to provide a calibration method that can establish an absolute temperature mapping relationship under non-contact external temperature measurement conditions and adapt to the non-uniform thermal reflectivity response of the sample under test. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a method and apparatus for calibrating the thermal reflectivity coefficient for reflectivity thermal imaging. It establishes a non-contact absolute temperature reference using the known phase transition temperature of a phase transition reference material, and utilizes multi-modal collaborative verification of light intensity sequence, acoustic response, and magnetic relaxation response to achieve full-field thermal reflectivity coefficient inversion and correction. This improves the calibration accuracy, anti-interference capability, and temperature measurement reliability of reflectivity thermal imaging in non-uniform material regions of micro / nano devices.

[0008] This invention provides a method for calibrating the thermal reflectivity coefficient for reflectivity thermal imaging, the calibration method comprising the following steps: During the process of applying thermal excitation according to the preset thermal excitation program, the light intensity sequence of the reflectivity thermal imaging area of ​​the sample under test, the acoustic response of the phase transition reference material with a known phase transition temperature, and the magnetic relaxation response are collected in real time to obtain multimodal synchronous data. Based on the multimodal synchronization data, reflectivity evolution features, acoustic phase transition features, and magnetic relaxation phase transition features are extracted respectively to obtain a phase transition recognition feature set; Based on the acoustic phase transition features in the phase transition recognition feature set, a phase transition candidate time window is determined, and the phase transition candidate time window is time-synchronized with the reflectivity evolution feature and the magnetic relaxation phase transition feature to obtain the phase transition candidate window feature; Based on the phase transition candidate window features, the acoustic phase transition features and the magnetic relaxation phase transition features are subjected to phase transition consistency joint judgment to determine the phase transition reference time, and an absolute temperature mapping relationship is established based on the phase transition reference time and the known phase transition temperature. Based on the absolute temperature mapping relationship, the full-field thermal reflectivity coefficient is inverted on the reflectivity evolution characteristics, and the inversion result is corrected to obtain the thermal reflectivity coefficient calibration result.

[0009] Preferably, obtaining multimodal synchronization data includes: Based on the preset thermal excitation program and the known phase transition temperature of the phase transition reference material, the basic acquisition sequence is determined, and the acoustic response of the phase transition reference material is acquired under the basic acquisition sequence to obtain the acoustic acquisition indication quantity. Based on the acoustic leader acquisition indication, the acquisition frame rate of the light intensity sequence, the acquisition interval of the acoustic response, and the acquisition period of the magnetic relaxation response are adjusted in a coordinated manner to obtain the synchronous acquisition timing. According to the synchronous acquisition timing, the light intensity sequence of the reflectivity thermal imaging area of ​​the sample under test, the acoustic response of the phase change reference material, and the magnetic relaxation response are acquired in real time and synchronously. The acquisition results are then compensated for delay and time-aligned to obtain the multimodal synchronous data.

[0010] Preferably, the obtained phase transition recognition feature set includes: Based on the multimodal synchronization data and the preset thermal excitation program, the light intensity sequence, the acoustic response, and the magnetic relaxation response are converted from the acquisition time axis to the thermal excitation phase axis to obtain the phase synchronization sequence; Acoustic gating weights are constructed based on the acoustic response in the phase synchronization sequence, and phase transition sensitivity enhancement processing is performed on the light intensity sequence and the magnetic relaxation response based on the acoustic gating weights to obtain an enhanced multimodal sequence, wherein the acoustic gating weights are used to characterize the degree of phase transition sensitivity. Based on the enhanced multimodal sequence, reflectivity evolution features, acoustic phase transition features, and magnetic relaxation phase transition features are extracted respectively, and the phase transition recognition feature set is generated according to the correspondence of the thermal excitation phase axis.

[0011] Preferably, obtaining the phase transition candidate window features includes: Based on the acoustic phase transition characteristics, acoustic drift characteristics and periodic perturbation characteristics of the non-phase transition thermal excitation stage are extracted, and an acoustic perturbation template is established based on the acoustic drift characteristics and the periodic perturbation characteristics. Based on the acoustic disturbance template, pseudo-mutation removal is performed on the acoustic phase transition features to obtain the net acoustic features; Based on the net acoustic characteristics and the thermal phase constraint conditions corresponding to the preset thermal excitation program, determine the candidate time window for phase transition; Feature segments corresponding to the phase transition candidate time window are extracted from the reflectivity evolution features and the magnetic relaxation phase transition features, and the feature segments are time-aligned with the net acoustic features to generate the phase transition candidate window features.

[0012] Preferably, establishing the absolute temperature mapping relationship includes: Based on the phase transition candidate window features, the net acoustic features and the magnetic relaxation phase transition features are subjected to phase mutual verification processing to obtain phase mutual verification features. Based on the phase verification feature, the net acoustic feature is matched with the acoustic disturbance template, the residual acoustic disturbance component obtained by matching is removed, and the phase transition state of the removed net acoustic feature is verified by the magnetic relaxation phase transition feature to obtain the phase transition confirmation feature. Based on the phase transition confirmation characteristics and the thermal phase constraint conditions, the phase offset between the acoustic phase transition response and the magnetic relaxation phase transition response is determined, and the phase transition confirmation characteristics are anchored and corrected using the phase offset to obtain the phase transition reference time. Based on the phase transition reference time and the known phase transition temperature of the phase transition reference material, the absolute temperature mapping relationship is established.

[0013] Preferably, obtaining the thermal reflectivity coefficient calibration result includes: Based on the absolute temperature mapping relationship and the phase offset, the temperature axis of the reflectivity evolution characteristics is reconstructed to obtain the full-field temperature response characteristics; Based on the full-field temperature response characteristics, the thermal reflectivity coefficients corresponding to the spatial location of the thermal imaging region of the sample under test are inverted to obtain the initial coefficient field. Anchoring reliability weights are generated based on the phase transition confirmation features and the acoustic gating weights, and the initial coefficient field is corrected based on the anchoring reliability weights to obtain a corrected coefficient field. The thermal reflectivity coefficient calibration result is generated based on the correction coefficient field.

[0014] Preferably, the step of correcting the initial coefficient field based on the anchored reliability weights to obtain a corrected coefficient field includes: Based on the anchored reliability weights, the initial coefficient field is partitioned into reliability regions to obtain a partitioned coefficient field; The partitioned coefficient field is subjected to reliability segmentation and boundary detection, low reliability coefficient regions and boundary preservation regions are marked, and a coefficient correction mask is generated; Based on the coefficient correction mask, neighborhood consistency correction is performed on the low reliability coefficient region, and boundary preservation processing is performed on the boundary-preserving region to obtain the correction coefficient field.

[0015] The present invention also provides a thermal reflectivity coefficient calibration device for reflectivity thermal imaging, the calibration device comprising: The multimodal synchronous acquisition module is used to synchronously acquire, in real time, the light intensity sequence of the reflectivity thermal imaging area of ​​the sample under test, the acoustic response of the phase transition reference material with a known phase transition temperature, and the magnetic relaxation response of the phase transition reference material during the process of applying thermal excitation according to a preset thermal excitation program, so as to obtain multimodal synchronous data. The phase transition feature extraction module is used to extract reflectivity evolution features, acoustic phase transition features and magnetic relaxation phase transition features based on the multimodal synchronization data, respectively, to obtain a phase transition recognition feature set; The candidate window generation module is used to determine the phase transition candidate time window based on the acoustic phase transition features in the phase transition recognition feature set, and to time-synchronize the phase transition candidate time window with the reflectivity evolution feature and the magnetic relaxation phase transition feature to obtain the phase transition candidate window feature; The temperature mapping establishment module is used to perform phase transition consistency judgment on the acoustic phase transition feature and the magnetic relaxation phase transition feature according to the phase transition candidate window feature, determine the phase transition reference time, and establish an absolute temperature mapping relationship based on the phase transition reference time and the known phase transition temperature. The coefficient calibration output module is used to perform full-field thermal reflectivity coefficient inversion on the reflectivity evolution characteristics according to the absolute temperature mapping relationship, and to correct the inversion results to obtain thermal reflectivity coefficient calibration results.

[0016] Compared with related technologies, the thermal reflectivity coefficient calibration method and calibration device for reflectivity thermal imaging provided by the present invention have the following beneficial effects: This invention enables the coordinated processing of reflectivity changes, acoustic phase transition responses, and magnetic relaxation phase transition responses under a preset thermal excitation program by simultaneously acquiring light intensity sequences, acoustic responses, and magnetic relaxation phase transition responses. Simultaneously, it establishes a mapping relationship between the phase transition reference time and absolute temperature using a phase transition reference material with a known phase transition temperature, reducing reliance on external contact temperature probes such as thermocouples and platinum resistance thermometers, and lowering the spatial deviation between the temperature measurement point and the imaging area.

[0017] Furthermore, this invention enhances the phase transition sensitivity of optical, acoustic, and magnetic relaxation signals and performs mutual verification by employing acoustic leader acquisition, acoustic gating weights, thermal excitation phase axis transformation, acoustic disturbance templates, pseudo-mutation elimination, and phase cross-verification processing. This can suppress the influence of ambient light disturbances, mechanical vibrations, acoustic drift, sampling delays, and modal response differences on phase transition identification, thereby improving the reliability of determining the phase transition reference time.

[0018] Furthermore, this invention performs full-field thermal reflectivity coefficient inversion based on the absolute temperature mapping relationship to reflectivity evolution characteristics. Combined with anchored reliability weights, reliability partitioning, coefficient correction masks, neighborhood consistency correction, and boundary preservation processing, the initial coefficient field is corrected. This can correct the abnormal coefficients in low-reliability regions and retain the coefficient differences that actually exist at material boundaries, thereby improving the accuracy, stability, and adaptability of thermal reflectivity coefficient calibration results to non-uniform material regions. Attached Figure Description

[0019] Figure 1This is a flowchart of the thermal reflectivity coefficient calibration method of the present invention; Figure 2 This is a system block diagram of the thermal reflectivity coefficient calibration device of the present invention; Figure 3 This is a schematic diagram of the multimodal synchronous acquisition and linkage timing of the present invention; Figure 4 This is a schematic diagram of phase transition recognition feature extraction and candidate window generation according to the present invention; Figure 5 This is a schematic diagram of the inversion and correction of the full-field thermal reflectivity coefficient according to the present invention. Detailed Implementation

[0020] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the drawings, not all structures. Moreover, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0021] It should also be noted that, for ease of description, the accompanying drawings show only the parts relevant to the invention and not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but it may also have additional steps not included in the drawings. The process may correspond to a method, function, procedure, subroutine, subroutine, etc. Example 1

[0022] This invention provides a method for calibrating the thermal reflectivity coefficient for reflectivity thermal imaging. In this calibration method, the sample under test has a reflectivity thermal imaging region, which is used to acquire a light intensity sequence that changes with thermal excitation. A phase transition reference material is disposed near the sample under test, in a non-functional region, or in a thermally coupled region, and has a known phase transition temperature, capable of generating acoustic and magnetic relaxation response changes during the phase transition process. The light intensity sequence is used to characterize the reflectivity evolution process of the sample under test, and the acoustic and magnetic relaxation responses are used to collaboratively determine the phase transition reference time of the phase transition reference material, thereby establishing an absolute temperature mapping relationship. Based on the absolute temperature mapping relationship, the reflectivity evolution characteristics are inverted and corrected using a full-field thermal reflectivity coefficient, yielding the thermal reflectivity coefficient calibration result.

[0023] Meanwhile, the phase transition reference material can be a material with a stable phase transition point within a preset thermal excitation temperature range, and the phase transition reference material exhibits detectable acoustic state changes and magnetic relaxation state changes before and after the phase transition. Acoustic state changes can manifest as changes in resonant frequency, amplitude, phase, echo energy, or acoustic impedance; magnetic relaxation state changes can manifest as changes in relaxation time, relaxation curve morphology, or magnetic relaxation signal intensity. For phase transition materials with weak magnetic relaxation responses, a detection component capable of generating a magnetic relaxation response can be incorporated into the phase transition reference material, or a phase transition reference material that inherently possesses a detectable magnetic relaxation response can be selected.

[0024] refer to Figure 1 As shown, Figure 1 This is used to demonstrate the main flow of the thermal reflectivity coefficient calibration method of the present invention. Figure 1 As can be seen, this invention executes in the following order: "multimodal synchronous acquisition—phase change feature extraction—candidate window generation—temperature mapping establishment—coefficient inversion correction." Step S1 outputs multimodal synchronous data, which serves as the basis for step S2 to extract the phase change identification feature set; step S2 outputs the phase change identification feature set, which serves as the input for step S3 to determine the phase change candidate window features; step S3 outputs the phase change candidate window features, which are used in step S4 to perform phase change consistency collaborative judgment and establish an absolute temperature mapping relationship; step S4 outputs the absolute temperature mapping relationship, which is used in step S5 to perform full-field thermal reflectivity coefficient inversion and correction, ultimately obtaining the thermal reflectivity coefficient calibration result. This figure illustrates the continuous processing relationship where the output of the previous step serves as the input of the next step.

[0025] The calibration method includes the following steps: S1: During the process of applying thermal excitation according to the preset thermal excitation program, the light intensity sequence of the reflectivity thermal imaging area of ​​the sample under test, the acoustic response of the phase transition reference material with a known phase transition temperature, and the magnetic relaxation response are collected in real time to obtain multimodal synchronous data.

[0026] like Figure 3 As shown, Figure 3 This diagram illustrates the multimodal synchronous acquisition and linkage timing in step S1. Using a preset thermal excitation program as the timeline, the thermal excitation process is divided into an initial stage, a near-phase transition stage, a phase transition-sensitive stage, and a post-phase transition stabilization stage. In the normal stage, the light intensity sequence, acoustic response, and magnetic relaxation response are acquired at a normal pace. When the acoustic response triggers the phase transition-sensitive judgment, the system enters a high-density synchronous acquisition state, linkageically increasing the light intensity sequence acquisition frame rate, shortening the acoustic response acquisition interval, and shortening the magnetic relaxation response acquisition cycle. In the post-phase transition stabilization stage, the acquisition pace returns to normal. Subsequently, the system assigns timestamps under the same time base to the three types of acquisition results and performs delay compensation and time alignment to form multimodal synchronous data.

[0027] Specifically, step S1 includes the following steps: S11: Based on the preset thermal excitation program and the known phase transition temperature of the phase transition reference material, determine the basic acquisition timing sequence, and perform pilot acquisition on the acoustic response of the phase transition reference material under the basic acquisition timing sequence to obtain the acoustic pilot acquisition indication quantity.

[0028] In this embodiment, the preset thermal excitation program includes thermal excitation start conditions, target thermal excitation range, thermal excitation change mode, and thermal excitation duration. The thermal excitation change mode can be linear heating, segmented heating, step heating, or pulse heating.

[0029] Based on the known phase transition temperature of the phase transition reference material, a thermal excitation stage in which the phase transition reference material may undergo a phase transition is determined in the preset thermal excitation program, and a phase transition-sensitive acquisition interval is set before and after this thermal excitation stage. Outside the phase transition-sensitive acquisition interval, a conventional acquisition cycle with a lower acquisition density is used; within the phase transition-sensitive acquisition interval, a candidate acquisition cycle with a higher acquisition density is used, thereby forming the basic acquisition timing sequence. The basic acquisition timing sequence includes at least the initial acquisition frame rate of the light intensity sequence, the initial acquisition interval of the acoustic response, the initial acquisition period of the magnetic relaxation response, and the unified time reference corresponding to the three.

[0030] Under the basic acquisition sequence, the acoustic response of the phase change reference material is acquired first. Specifically, acoustic excitation can be applied to the region where the phase change reference material is located through a piezoelectric transducer, an ultrasonic transducer, or an equivalent acoustic excitation unit, and the returned resonant frequency, amplitude, phase, echo energy, or acoustic impedance change information can be acquired through an acoustic receiving unit.

[0031] Baseline normalization and short-time trend analysis are performed on the continuously acquired acoustic responses. When the acoustic response shows a trend approaching phase transition characteristics relative to the baseline state in the non-phase transition stage, an acoustic leader acquisition indicator is generated. This acoustic leader acquisition indicator is used to characterize the degree to which the current thermal excitation stage approaches the phase transition process and serves as input for subsequent adjustments to the light intensity sequence, acoustic response, and magnetic relaxation response acquisition timing.

[0032] S12: Based on the acoustic leader acquisition indication, the acquisition frame rate of the light intensity sequence, the acquisition interval of the acoustic response, and the acquisition period of the magnetic relaxation response are adjusted in a coordinated manner to obtain a synchronous acquisition timing sequence.

[0033] In this embodiment, the acoustic leader acquisition indication obtained in step S11 is received, and the acquisition status of the current thermal excitation stage is determined based on the acoustic leader acquisition indication. When the acoustic leader acquisition indication indicates that the phase transition reference material is still in the non-phase transition stable stage, the basic acquisition timing is maintained, and the light intensity sequence, acoustic response, and magnetic relaxation response are acquired according to the normal rhythm. When the acoustic leader acquisition indication indicates that the phase transition reference material has entered the phase transition sensitive stage, the acquisition frame rate of the light intensity sequence is increased, the acquisition interval of the acoustic response is shortened, and the acquisition period of the magnetic relaxation response is shortened, so that the three types of responses have higher time resolution before and after the phase transition. When the acoustic leader acquisition indication indicates that the acoustic response has passed the phase transition sensitive stage and tends to stabilize, the normal acquisition rhythm is gradually restored to reduce the amount of invalid data.

[0034] The aforementioned coordinated adjustment does not involve adjusting the three acquisition channels independently, but rather uses the same time reference as a constraint to ensure that the light intensity sequence, acoustic response, and magnetic relaxation response maintain a corresponding sampling relationship within the same thermal excitation stage.

[0035] In practice, multiple acquisition levels can be pre-set. Each acquisition level corresponds to a set of light intensity sequence acquisition frame rates, acoustic response acquisition intervals, and magnetic relaxation response acquisition cycles. The acquisition is then switched or smoothly transitioned between different acquisition levels based on the acoustic leader acquisition indication, thus obtaining a synchronous acquisition sequence. This synchronous acquisition sequence includes at least the start time, sampling interval, sampling duration, sampling timestamp rules, and trigger correspondence between different acquisition channels for each acquisition channel, serving as the basis for executing step S13, real-time synchronous acquisition.

[0036] S13: According to the synchronous acquisition timing, the light intensity sequence of the reflectivity thermal imaging area of ​​the sample under test, the acoustic response of the phase change reference material, and the magnetic relaxation response are acquired in real time, and the acquisition results are compensated for delay and time aligned to obtain the multimodal synchronous data.

[0037] In this embodiment, the optical acquisition channel, acoustic acquisition channel, and magnetic relaxation acquisition channel are driven to operate according to the synchronous acquisition timing obtained in step S12. The optical acquisition channel faces the thermal imaging area of ​​the sample under test and acquires a light intensity sequence that changes with the thermal excitation process. The light intensity sequence consists of continuous image frames or corresponding pixel grayscale value sequences. The acoustic acquisition channel faces the phase change reference material and acquires the acoustic response. The acoustic response includes information reflecting the changes in the acoustic state of the phase change reference material, such as resonant frequency, amplitude, phase, echo energy, or acoustic impedance. The magnetic relaxation acquisition channel faces the phase change reference material and acquires the magnetic relaxation response. The magnetic relaxation response includes information reflecting the changes in the internal state of the phase change reference material, such as relaxation time, relaxation curve shape, or magnetic relaxation signal intensity.

[0038] During the acquisition process, all three types of responses are assigned an acquisition timestamp under the same time base. Since optical acquisition, acoustic acquisition, and magnetic relaxation acquisition may differ in triggering links, signal propagation paths, sampling integration times, and processing delays, after obtaining the acquisition results, delay compensation is performed on the light intensity sequence, acoustic response, and magnetic relaxation response based on the pre-calibrated fixed delay of each acquisition channel and the dynamic delay estimate during the acquisition process. Subsequently, using the time reference in the synchronous acquisition sequence, the compensated light intensity sequence, acoustic response, and magnetic relaxation response are time-aligned to form corresponding data sets of optical, acoustic, and magnetic relaxation changes under the same thermal excitation stage. After delay compensation and time alignment, the light intensity sequence, acoustic response, magnetic relaxation response, and corresponding timestamps are combined to obtain the multimodal synchronous data. This multimodal synchronous data serves as the foundation for subsequent extraction of reflectivity evolution features, acoustic phase transition features, and magnetic relaxation phase transition features.

[0039] S2: Based on the multimodal synchronization data, extract reflectivity evolution features, acoustic phase transition features and magnetic relaxation phase transition features respectively to obtain a phase transition recognition feature set.

[0040] Specifically, step S2 includes the following steps: S21: Based on the multimodal synchronization data and the preset thermal excitation program, the light intensity sequence, the acoustic response, and the magnetic relaxation response are converted from the acquisition time axis to the thermal excitation phase axis to obtain the phase synchronization sequence.

[0041] In this embodiment, the multimodal synchronization data obtained in step S1 is read. The multimodal synchronization data includes the light intensity sequence, acoustic response, magnetic relaxation response and their respective acquisition timestamps under the same time reference. At the same time, the thermal excitation start time, thermal excitation end time, thermal excitation change mode, thermal excitation stage division and target thermal state corresponding to each stage are read from the preset thermal excitation program.

[0042] First, a thermal excitation phase axis is established according to a preset thermal excitation program. The thermal excitation phase axis is used to characterize the progress position of the thermal excitation process from the initial stage, the heating stage, the near-phase change stage, the phase change sensitive stage to the stable stage after the phase change, rather than using ordinary acquisition time as the basis for data arrangement.

[0043] Subsequently, each frame of the light intensity sequence or each set of pixel gray values, each set of resonant frequency, amplitude, phase, echo energy or acoustic impedance change data in the acoustic response, and each set of relaxation time, relaxation curve shape or magnetic relaxation signal intensity data in the magnetic relaxation response are mapped to the corresponding thermal excitation phase position according to their acquisition timestamp.

[0044] For cases where the sampling intervals of different acquisition channels are inconsistent, the thermal excitation phase axis is used as a unified reference. Data at adjacent phase positions are interpolated, resampled, or matched with neighboring phases to ensure that the light intensity sequence, acoustic response, and magnetic relaxation response have corresponding data units at the same thermal excitation phase position.

[0045] After the above transformation, the three types of responses originally arranged according to the acquisition time axis are converted into a phase synchronization sequence arranged according to the thermal excitation process, so that subsequent feature extraction can reduce the impact of heating rate fluctuations, sampling period differences and channel response delays on phase transition identification.

[0046] S22: Construct acoustic gating weights based on the acoustic response in the phase synchronization sequence, and perform phase transition sensitivity enhancement processing on the light intensity sequence and the magnetic relaxation response based on the acoustic gating weights to obtain an enhanced multimodal sequence, wherein the acoustic gating weights are used to characterize the degree of phase transition sensitivity.

[0047] In this embodiment, the acoustic response is extracted from the phase synchronization sequence obtained in step S21, and the change of the acoustic response on the thermal excitation phase axis is used as a leading criterion for judging the phase transition sensitivity.

[0048] Specifically, the resonant frequency shift, amplitude change, phase change, echo energy change, or acoustic impedance change in the acoustic response can be smoothed and the baseline normalized to obtain an acoustic change trajectory that reflects the strength of the acoustic state change. Subsequently, the degree of deviation between the acoustic change trajectory and the acoustic baseline in the non-phase transition stage is compared, and the change trend between adjacent thermal excitation phase positions is combined to generate acoustic gating weights.

[0049] The acoustic gating weights are assigned a lower weight during the non-phase-change stable phase, gradually increase as the acoustic response begins to deviate from the baseline and approaches the phase-change sensitive phase, and are assigned a higher weight in the phase interval where the acoustic response changes most significantly.

[0050] Phase transition-sensitive enhancement processing is applied to the light intensity sequence and magnetic relaxation response in the phase synchronization sequence based on acoustic gating weights: For thermally excited phase intervals with high acoustic gating weights, the effective contribution of reflectivity variation trends in the corresponding light intensity sequence and relaxation state changes in the magnetic relaxation response is enhanced; for thermally excited phase intervals with low acoustic gating weights, the impact of ambient light fluctuations, random noise, magnetic relaxation sampling fluctuations, and non-phase transition thermal responses on subsequent feature extraction is reduced. Enhancement processing can be achieved through weighted smoothing, gated amplification, noise suppression, phase interval weighting, or downweighting of abnormal sampling points.

[0051] After processing, the enhanced light intensity sequence, acoustic response, and magnetic relaxation response are recombine according to the thermal excitation phase axis to obtain an enhanced multimodal sequence. This enhanced multimodal sequence not only preserves the phase correspondence between the three types of responses but also highlights the effective changes within the phase transition sensitive range, providing a more stable data foundation for subsequent extraction of reflectivity evolution features, acoustic phase transition features, and magnetic relaxation phase transition features.

[0052] S23: Based on the enhanced multimodal sequence, extract reflectivity evolution features, acoustic phase transition features, and magnetic relaxation phase transition features respectively, and generate the phase transition recognition feature set according to the correspondence of the thermal excitation phase axis.

[0053] In this embodiment, the enhanced multimodal sequence obtained in step S22 is used as input, and the features required for three types of phase transition identification are extracted under the same thermal excitation phase axis.

[0054] For the light intensity sequence, light intensity normalization, background subtraction, and adjacent phase change analysis are performed on each pixel or preset local area within the thermal imaging area of ​​the reflectance of the sample under test. Reflectance evolution features that can characterize the change of reflectance with the thermal excitation process are extracted. The reflectance evolution features may include information such as reflectance change trend, reflectance change rate, local plateau change, grayscale stability, and reflectance differences between pixels.

[0055] For acoustic response, the resonant frequency shift, amplitude attenuation, phase change, echo energy change or acoustic impedance change of the phase change reference material in the enhanced multimodal sequence are extracted and combined with the change position and intensity on the thermal excitation phase axis to form acoustic phase change characteristics, which are used to characterize the changes in the macroscopic acoustic state of the phase change reference material.

[0056] For magnetic relaxation response, the changes in relaxation time, slope of relaxation curve, intensity of relaxation signal, extreme values ​​of relaxation curve or stable range of relaxation state are extracted to form magnetic relaxation phase transition characteristics, which are used to characterize the changes in the internal state or microscopic motion state of phase transition reference material.

[0057] After extracting the three types of features, reflectivity evolution features, acoustic phase transition features, and magnetic relaxation phase transition features are bound together according to the correspondence of the thermal excitation phase axis, at the same thermal excitation phase position or within adjacent allowable phase ranges, forming feature groups containing phase position, optical features, acoustic features, and magnetic relaxation features. Multiple feature groups are arranged in the order of thermal excitation phase, thus generating a phase transition recognition feature set. This phase transition recognition feature set is used to subsequently determine the candidate time window for phase transition based on the acoustic phase transition features and to perform time synchronization association with the reflectivity evolution features and magnetic relaxation phase transition features.

[0058] like Figure 4As shown, a continuous feature processing link is formed between steps S2 and S3. First, the multimodal synchronization data is converted to the thermally excited phase axis to obtain a phase synchronization sequence. Then, acoustic gating weights are constructed based on the acoustic response to enhance the phase transition sensitivity of the light intensity sequence and magnetic relaxation response, resulting in an enhanced multimodal sequence. Next, reflectivity evolution features, acoustic phase transition features, and magnetic relaxation phase transition features are extracted and bound from the enhanced multimodal sequence to generate a phase transition recognition feature set. Based on this, an acoustic perturbation template is further established to remove pseudo-mutations from the acoustic phase transition features, resulting in net acoustic features. Then, phase transition candidate time windows are determined through thermal phase constraint screening, and feature segments within the time-aligned window are extracted to generate phase transition candidate window features. Therefore, subsequent phase transition consistency collaborative judgment no longer relies on a single acoustic mutation or a single optical change, but is based on the enhanced, de-perturbed, and windowed multimodal features.

[0059] S3: Based on the acoustic phase transition features in the phase transition identification feature set, determine the phase transition candidate time window, and time-synchronize the phase transition candidate time window with the reflectivity evolution feature and the magnetic relaxation phase transition feature to obtain the phase transition candidate window feature.

[0060] Specifically, step S3 includes the following steps: S31: Based on the acoustic phase transition characteristics, extract the acoustic drift characteristics and periodic disturbance characteristics of the non-phase transition thermal excitation stage, and establish an acoustic disturbance template based on the acoustic drift characteristics and the periodic disturbance characteristics.

[0061] In this embodiment, the phase transition identification feature set obtained in step S2 is used as input, and acoustic phase transition features arranged along the thermal excitation phase axis are read from the phase transition identification feature set. Since the phase transition reference material usually has a relatively stable non-phase transition thermal excitation stage before the actual phase transition and after the phase transition is completed, the acoustic feature segments far from the known phase transition temperature are selected as reference segments for the non-phase transition thermal excitation stage according to the thermal excitation stage division corresponding to the preset thermal excitation program. The reference segments may include the initial stage of thermal excitation, the stable heating stage below the phase transition temperature, and the stable stage after the phase transition is completed.

[0062] The acoustic phase transition characteristics are analyzed in the reference segment to detect the slow shift trend of the acoustic phase transition characteristics with the thermal excitation phase change, and the acoustic drift characteristics are extracted. In specific implementation, the acoustic drift characteristics can be obtained by smoothing, moving average, or low-frequency trend extraction of the acoustic phase transition characteristics in the non-phase transition thermal excitation stage. These characteristics are used to characterize the baseline shift of the acoustic response caused by slow temperature changes, slow changes in coupling state, or device drift when no phase transition occurs.

[0063] Simultaneously, the repetitive fluctuations in the reference segment caused by mechanical vibration, acoustic coupling fluctuations, transducer operating cycles, or external environmental disturbances are analyzed to extract periodic disturbance characteristics. In specific implementation, these periodic disturbance characteristics can be obtained by performing peak-valley period statistics, frequency component analysis, or periodic fluctuation amplitude statistics on the acoustic phase transition characteristics during the non-phase transition thermal excitation stage, and are used to characterize the acoustic disturbance patterns that may recur in the non-phase transition state.

[0064] The acoustic drift features and the periodic perturbation features are combined to form an acoustic perturbation template that can characterize the acoustic variation law without phase change. The acoustic perturbation template is used to describe the baseline drift and periodic perturbation that may occur in the acoustic response when no phase change occurs, and serves as the basis for subsequent pseudo-mutation elimination of acoustic phase change features.

[0065] S32: Based on the acoustic disturbance template, perform pseudo-mutation removal on the acoustic phase transition features to obtain net acoustic features.

[0066] In this embodiment, the acoustic disturbance template obtained in step S31 is matched with the acoustic phase transition features to identify pseudo-mutation components in the acoustic phase transition features that are not caused by the phase transition. Specifically, the acoustic phase transition features are analyzed segment by segment according to the thermal excitation phase axis. Within each thermal excitation phase interval, the changing trend, amplitude, and periodic fluctuation form of the current acoustic phase transition features are compared with the acoustic drift features and periodic disturbance features in the acoustic disturbance template.

[0067] When an acoustic change is consistent with the baseline drift trend in the acoustic disturbance template, or with the repetitive fluctuation pattern of a periodic disturbance feature, the acoustic change is marked as a non-phase-change disturbance. The non-phase-change disturbance may include acoustic fluctuations caused by slow acoustic baseline drift, periodic mechanical vibration, transducer coupling changes, or environmental disturbances.

[0068] When an acoustic change significantly deviates from the acoustic disturbance template, and its location is within the phase transition sensitive stage defined by the preset thermal excitation program, the acoustic change is retained as a valid acoustic change that may be related to the phase transition. The valid acoustic change can manifest as a sudden change in the acoustic response, a local rapid change, an increase in the rate of change, or a change process that tends to stabilize again after a sudden change.

[0069] Subsequently, the portions marked as non-phase transition perturbations are subjected to subtraction, weight reduction, smoothing replacement, or invalidation. The retained effective acoustic changes are continuously organized and their phase positions recorded to obtain the net acoustic features. These net acoustic features reflect the acoustic phase transition response after eliminating acoustic drift, periodic perturbations, and mechanical pseudo-mutations, and can serve as a direct basis for subsequently determining the candidate phase transition time window.

[0070] S33: Determine the candidate time window for phase transition based on the net acoustic characteristics and the thermal phase constraint conditions corresponding to the preset thermal excitation program.

[0071] In this embodiment, the net acoustic characteristics obtained in step S32 are used as input, and the thermal phase constraint conditions corresponding to the preset thermal excitation program are read. The thermal phase constraint conditions are used to limit the reasonable thermal excitation phase range in which the phase change reference material may undergo a phase change under the preset thermal excitation program.

[0072] The thermal phase constraint conditions can be determined based on the known phase transition temperature of the phase transition reference material, the thermal excitation initiation conditions, the thermal excitation change mode, the thermal excitation duration, and the thermal coupling relationship between the phase transition reference material and the heat source. For example, when the preset thermal excitation program is a heating process, the possible thermal phase range of the phase transition can be determined based on the theoretical reach range of the known phase transition temperature during the heating process; when the preset thermal excitation program is a segmented heating or step heating, the thermal excitation stage corresponding to the phase transition temperature can be set as the allowable candidate range.

[0073] Within the range defined by the thermal phase constraint, phase transition candidates are determined for the net acoustic features. Specifically, intervals in the net acoustic features that show increased amplitude of change, increased rate of change, acoustic state that changes from stable to abrupt, or abrupt change that tends to stabilize again are identified, and these intervals are cross-validated with the thermal phase constraint.

[0074] If a certain range of net acoustic feature changes exhibits both phase transition-related acoustic changes and falls within the limits allowed by thermal phase constraints, then this range is designated as a candidate time window for the phase transition. If multiple candidate ranges exist, they can be ranked according to the intensity, duration, continuity, and proximity to the thermal excitation stage corresponding to the known phase transition temperature. The range with the highest confidence level can then be selected as the candidate time window for the phase transition, or multiple candidate ranges can be retained for further confirmation by subsequent magnetic relaxation phase transition features.

[0075] The phase transition candidate time window is used to limit the intercept range of subsequent reflectivity evolution features and magnetic relaxation phase transition features, so that subsequent collaborative judgment is only performed within the effective range where phase transitions may occur.

[0076] S34: Extract feature segments corresponding to the phase transition candidate time window from the reflectivity evolution features and the magnetic relaxation phase transition features, and time-align the feature segments with the net acoustic features to generate the phase transition candidate window features.

[0077] In this embodiment, based on the phase transition candidate time window obtained in step S33, the thermal excitation phase range or acquisition time range corresponding to the phase transition candidate time window is searched in the phase transition identification feature set, and reflectivity evolution feature segments, magnetic relaxation phase transition feature segments and corresponding net acoustic feature segments are extracted within the range.

[0078] For reflectivity evolution characteristics, the reflectivity change trend, reflectivity change rate, local plateau changes, or pixel grayscale stability information of the thermal imaging region of the sample under test within the phase transition candidate time window are extracted. For magnetic relaxation phase transition characteristics, the relaxation time change, relaxation curve change, relaxation signal intensity change, or relaxation state stability interval information of the phase transition reference material within the phase transition candidate time window are extracted. For net acoustic characteristics, the acoustic change information after eliminating pseudo-mutations within the phase transition candidate time window is extracted.

[0079] Subsequently, using the thermal excitation phase position or timestamp of the net acoustic feature segment as a reference, the reflectivity evolution feature segment and the magnetic relaxation phase transition feature segment are time-aligned. When the sampling points of the three types of features are not completely corresponding, neighboring sampling point matching, phase interpolation, in-window resampling, or timestamp matching within the allowable error range can be used to establish a one-to-one correspondence or interval correspondence between the optical changes, net acoustic changes, and magnetic relaxation changes within the same phase transition candidate time window.

[0080] After time alignment is completed, the phase transition candidate time window, net acoustic feature segment, reflectivity evolution feature segment, magnetic relaxation phase transition feature segment and their corresponding time or thermal excitation phase information are combined to generate phase transition candidate window features.

[0081] The phase transition candidate window feature is used to subsequently perform a coordinated judgment on the consistency of acoustic phase transition features and magnetic relaxation phase transition features, so that the determination of the subsequent phase transition reference time no longer depends on a single acoustic abrupt change or a single optical change, but is confirmed by combining multimodal features within the candidate window.

[0082] S4: Based on the phase transition candidate window features, perform phase transition consistency judgment on the acoustic phase transition features and the magnetic relaxation phase transition features, determine the phase transition reference time, and establish an absolute temperature mapping relationship based on the phase transition reference time and the known phase transition temperature.

[0083] Specifically, step S4 includes the following steps: S41: Based on the phase transition candidate window features, perform phase verification processing on the net acoustic features and the magnetic relaxation phase transition features to obtain phase verification features.

[0084] In this embodiment, the phase transition candidate window features generated in step S3 are used as input to read the net acoustic features, magnetic relaxation phase transition features, and corresponding thermal excitation phase information of the same phase transition candidate time window. The net acoustic features characterize the acoustic phase transition response after eliminating acoustic drift and periodic perturbations, while the magnetic relaxation phase transition features characterize the magnetic relaxation response of the phase transition reference material as its internal state changes with the thermal excitation process.

[0085] Using the thermal excitation phase axis as a unified reference, the starting position of acoustic response change, the interval of change enhancement, and the peak or plateau interval in the net acoustic characteristics are analyzed in correspondence with the relaxation time change position, the interval of relaxation curve slope change, and the extreme or stable interval of relaxation signal in the magnetic relaxation phase transition characteristics.

[0086] When both the net acoustic characteristics and the magnetic relaxation phase transition characteristics show a trend consistent with the phase transition process within the same or adjacent allowable thermal excitation phase range, the phase range is marked as a mutually supportive phase transition response interval; when only one of them changes, or when the direction of change of both is inconsistent with the phase transition process, the validity marking of the phase range is reduced.

[0087] In practice, phase verification processing of net acoustic features and magnetic relaxation phase transition features can be performed by using phase position matching, peak interval matching, change trend matching, response intensity matching, or window correlation matching.

[0088] After completing the mutual verification process, the phase correspondence, response intensity correspondence, consistency of change trends, and effective phase transition response intervals are combined to obtain the phase mutual verification features. These phase mutual verification features characterize whether the net acoustic features and magnetic relaxation phase transition features both point to the same phase transition event within the same candidate phase transition time window, and serve as the basis for subsequent removal of residual acoustic disturbances and phase transition state verification.

[0089] S42: Based on the phase verification feature, the net acoustic feature is matched with the acoustic disturbance template, the residual acoustic disturbance component obtained by matching is removed, and the phase transition state of the removed net acoustic feature is verified by the magnetic relaxation phase transition feature to obtain the phase transition confirmation feature.

[0090] In this embodiment, the phase verification features obtained in step S41 are used as input, and the acoustic disturbance template established in step S31 and the net acoustic features obtained in step S32 are called. Since the net acoustic features have been eliminated by pseudo-mutation, local acoustic disturbances caused by transducer coupling changes, mechanical micro-vibrations, and thermal expansion may still remain within the phase transition candidate time window, the net acoustic features are further matched with the acoustic disturbance template.

[0091] Within the effective phase transition response range of the phase mutual verification feature marker, the local change morphology, duration of change, fluctuation period and change amplitude of the net acoustic feature are analyzed and compared with the acoustic drift feature and periodic disturbance feature in the acoustic disturbance template.

[0092] If an acoustic change is consistent with the change pattern of the acoustic disturbance template, then that part is marked as a residual acoustic disturbance component and is removed, downweighted, or smoothed out. If an acoustic change is inconsistent with the acoustic disturbance template, and its phase position is supported by the magnetic relaxation phase transition characteristics, then it is retained as an effective acoustic phase transition response.

[0093] Subsequently, the magnetic relaxation phase transition characteristics were used as the basis for verifying the phase transition state, and the net acoustic characteristics after removing residual acoustic disturbances were verified. When the magnetic relaxation phase transition characteristics exhibited abrupt changes in relaxation time, changes in the slope of the relaxation curve, changes in the intensity of the relaxation signal, extreme values ​​of the relaxation curve, or stable intervals after the phase transition within the corresponding thermal excitation phase range, the net acoustic change was confirmed to correspond to the true phase transition state; when the magnetic relaxation phase transition characteristics did not show corresponding changes, the corresponding net acoustic change was marked as a low-confidence change.

[0094] After completing the above matching, elimination, and verification, the effective acoustic phase transition response, corresponding magnetic relaxation state changes, phase correspondence, and phase transition state verification results are combined to obtain the phase transition confirmation features. These features characterize the true phase transition response after acoustic disturbance elimination and magnetic relaxation verification, providing a basis for subsequently determining the phase offset and phase transition reference time.

[0095] S43: Based on the phase transition confirmation characteristics and the thermal phase constraint conditions, determine the phase offset between the acoustic phase transition response and the magnetic relaxation phase transition response, and use the phase offset to anchor and correct the phase transition confirmation characteristics to obtain the phase transition reference time.

[0096] In this embodiment, the phase transition confirmation features obtained in step S42 are used as input, and combined with the thermal phase constraint conditions used in step S33, the phase deviation between the acoustic phase transition response and the magnetic relaxation phase transition response is analyzed. The thermal phase constraint conditions are used to limit the reasonable thermal excitation phase range of the phase transition reference material under a preset thermal excitation program, and to eliminate abnormal responses that exceed the laws of the thermal process.

[0097] The phase positions corresponding to the effective acoustic phase transition response and the magnetic relaxation phase transition response are extracted from the phase transition confirmation features. The phase position corresponding to the effective acoustic phase transition response can be determined by the starting position of the acoustic response change, the position of the maximum rate of change, the peak position of the response, or the stable initial position after the phase transition. The phase position corresponding to the magnetic relaxation phase transition response can be determined by the abrupt change position of the relaxation time, the extreme value position of the relaxation curve, the position of the change in the slope of the relaxation curve, or the stable position of the relaxation state.

[0098] Because acoustic and magnetic relaxation responses may differ in propagation path, acquisition period, signal processing delay, and the response speed of the material's internal state, their corresponding phase positions may not completely coincide. The phase offset between the acoustic phase transition response and the magnetic relaxation phase transition response is determined based on their phase sequence, phase interval, and whether they are within the allowable range of thermal phase constraints.

[0099] Subsequently, the phase transition confirmation features are anchored and corrected using the phase offset, that is, the phase positions in the net acoustic features and magnetic relaxation phase transition features that have differences in response sequence are corrected to the reference phase range corresponding to the same phase transition event.

[0100] Within the aforementioned reference phase range, a phase position that simultaneously satisfies the requirements of effective acoustic response, confirmed magnetic relaxation state, and thermal phase constraint is selected as the phase transition reference position. Finally, the phase transition reference position is converted into the corresponding acquisition time point or time position during the thermal excitation process to obtain the phase transition reference time.

[0101] The phase transition reference time refers to the moment when the phase transition reference material is confirmed to have undergone a phase transition during this thermal excitation process and can be used for absolute temperature anchoring.

[0102] S44: Based on the phase transition reference time and the known phase transition temperature of the phase transition reference material, establish the absolute temperature mapping relationship.

[0103] In this embodiment, the phase transition reference time obtained in step S43 is used as the temperature anchoring time, and the temperature corresponding to the phase transition reference time is set as the known phase transition temperature of the phase transition reference material. The known phase transition temperature can be the melting temperature, solidification temperature, crystallization transition temperature, glass transition temperature, or other phase transition temperatures that can be stably reproduced during thermal excitation of the phase transition reference material.

[0104] Based on the preset thermal excitation program, phase transition reference time, known phase transition temperature, and the time or thermal excitation phase sequence during the thermal excitation process, establish the correspondence between acquisition time, thermal excitation phase, and absolute temperature.

[0105] In practice, when the preset thermal excitation program is linear or approximately linear heating, the absolute temperature corresponding to each acquisition time point before and after the phase transition reference time can be determined by taking the phase transition reference time and the known phase transition temperature as anchor points, combined with the thermal excitation start conditions and the thermal excitation change rate.

[0106] When the preset thermal excitation program is segmented heating, stepped heating, or pulsed heating, the temperature correspondence between different stages can be established according to the set thermal state, stage duration, and stage of phase change reference time of each thermal excitation stage, and continuous processing or segmented connection processing can be performed at the stage boundary.

[0107] In cases where there is a thermal transfer delay or thermal coupling difference between the phase change reference material and the thermal imaging region of the sample under test, the acquisition time or the temperature corresponding to the thermal excitation phase can be corrected based on the spatial position, thermal coupling state, phase offset, or pre-calibrated thermal transfer delay between the phase change reference material and the sample under test.

[0108] After completing the above processing, the absolute temperature mapping relationship is obtained. This absolute temperature mapping relationship includes at least the correspondence between acquisition time and absolute temperature, or the correspondence between thermal excitation phase and absolute temperature, and may also include the correspondence between time, phase, and temperature simultaneously. This absolute temperature mapping relationship serves as the basis for subsequent temperature axis reconstruction of reflectivity evolution characteristics and inversion of the full-field thermal reflectivity coefficient.

[0109] S5: Based on the absolute temperature mapping relationship, perform full-field thermal reflectivity coefficient inversion on the reflectivity evolution characteristics, and correct the inversion results to obtain thermal reflectivity coefficient calibration results.

[0110] like Figure 5 As shown, step S5 is used to convert the absolute temperature mapping relationship into a thermal reflectivity coefficient calibration result that can be used for reflectivity thermal imaging temperature measurement. Specifically, firstly, the temperature axis is reconstructed based on the absolute temperature mapping relationship, phase offset, and reflectivity evolution characteristics to obtain the full-field temperature response characteristics; then, thermal reflectivity coefficients are inverted according to the spatial location of the thermal imaging area of ​​the sample under test to obtain the initial coefficient field; subsequently, anchoring reliability weights are generated by combining phase transition confirmation characteristics and acoustic gating weights, and the initial coefficient field is subjected to reliability partitioning, boundary detection, neighborhood consistency correction, and boundary preservation processing to obtain the corrected coefficient field; finally, thermal reflectivity coefficient calibration results are generated based on the corrected coefficient field. Thus, the calibration results can not only reflect the thermal reflectivity coefficients at each spatial location, but also reduce the influence of anomalous coefficients in low-reliability areas on the temperature measurement results, and retain the coefficient differences that truly exist at the boundaries of different materials or structures.

[0111] Specifically, step S5 includes the following steps: S51: Based on the absolute temperature mapping relationship and the phase offset, the temperature axis of the reflectivity evolution characteristics is reconstructed to obtain the full-field temperature response characteristics.

[0112] In this embodiment, the absolute temperature mapping relationship established in step S4 is used as the basis for temperature conversion, and the phase bias obtained in step S43 and the reflectivity evolution characteristics obtained in step S2 are read. The absolute temperature mapping relationship is used to characterize the correspondence between acquisition time, thermal excitation phase and absolute temperature, and the phase bias is used to compensate for the phase difference between acoustic phase transition response and magnetic relaxation phase transition response, as well as the phase shift caused by heat transfer delay, sampling period difference or signal processing delay.

[0113] First, based on the stated phase offset, phase correction is performed on each sampling point or thermal excitation phase position in the reflectivity evolution feature to ensure that the reflectivity evolution feature is within the same temperature reference system as the phase transition reference time confirmed by both acoustic and magnetic relaxation responses. For cases where the reflectivity evolution feature exhibits uneven sampling intervals, local frame gaps, or incomplete phase position correspondence, methods such as neighbor-to-neighbor phase matching, phase interpolation, in-window resampling, or timestamp compensation can be used to adjust the reflectivity evolution feature to a temperature axis position that matches the absolute temperature mapping relationship.

[0114] Subsequently, according to the absolute temperature mapping relationship, the reflectance evolution characteristics are converted from the original time arrangement or thermal excitation phase arrangement to an absolute temperature axis arrangement. Specifically, for each pixel position or preset local area in the thermal imaging region of the sample under test, the reflectance change trend, reflectance change rate, local plateau change or grayscale stability information at different sampling times are read, and the corresponding absolute temperature position is determined according to the absolute temperature mapping relationship, thereby forming the reflectance-temperature response relationship of that pixel position or local area.

[0115] After the above processing is completed, the reflectance-temperature response relationship of each pixel position or each preset local region is arranged according to its spatial position in the thermal imaging region of the reflectance of the sample under test, thus obtaining the full-field temperature response characteristics. The full-field temperature response characteristics are used to characterize the response process of reflectance at different spatial positions of the sample under test with changes in absolute temperature, and serve as the basis for subsequent inversion of the thermal reflectance coefficient corresponding to the spatial position.

[0116] S52: Based on the full-field temperature response characteristics, the thermal reflectivity coefficient corresponding to the spatial location of the thermal imaging region of the sample under test is inverted to obtain the initial coefficient field.

[0117] In this embodiment, the full-field temperature response feature obtained in step S51 is used as input, and a spatial index is established according to the spatial coordinates, pixel coordinates, or preset grid area of ​​the thermal imaging area of ​​the sample to be tested. The spatial index is used to bind each pixel position or each local area to its corresponding full-field temperature response feature, so that the thermal reflectivity coefficient inversion process can be performed separately for different spatial positions, rather than using a single fixed coefficient for the entire imaging area.

[0118] For each pixel location or preset local region, reflectance response data at different absolute temperatures is extracted from the global temperature response features, and local coefficient inversion is performed based on the reflectance-temperature response relationship. Specifically, linear fitting, piecewise fitting, robust regression, weighted least squares, or local window fitting can be used to determine the thermal reflectance coefficient corresponding to the spatial location (the thermal reflectance coefficient can be understood as the degree of normalized reflectance change caused by a unit temperature change, which can be determined by the local slope of the reflectance-temperature response relationship during inversion). For regions where the reflectance response changes approximately linearly within a specific temperature range, linear inversion can be performed within the corresponding temperature interval. For regions where the reflectance response exhibits piecewise changes or local nonlinear changes, piecewise inversion can be performed according to temperature intervals, and inversion results matching the actual thermal imaging operating temperature range can be selected.

[0119] During the inversion process, pixel grayscale stability, temperature sampling coverage, reflectance variation continuity, and neighborhood spatial consistency can be combined to deweight or label abnormal pixels, saturated pixels, low signal-to-noise ratio pixels, or locally contaminated areas. For areas with significant reflectance variations, such as material boundaries, thin film steps, and metal interconnect edges, their spatial correspondence is preserved to avoid misinterpreting real material differences as ordinary noise.

[0120] After inverting the thermal reflectivity coefficients at all spatial locations, the thermal reflectivity coefficients obtained at each pixel location or local region are rearranged according to their spatial location to form an initial coefficient field. This initial coefficient field characterizes the initial distribution of thermal reflectivity coefficients at different locations within the thermal imaging area of ​​the sample under test and serves as input for subsequent reliability correction.

[0121] S53: Generate anchoring reliability weights based on the phase transition confirmation features and the acoustic gating weights, and correct the initial coefficient field based on the anchoring reliability weights to obtain the corrected coefficient field.

[0122] In this embodiment, the phase transition confirmation feature obtained in step S42, the acoustic gating weight obtained in step S22, and the initial coefficient field obtained in step S52 are used as inputs. The phase transition confirmation feature is used to characterize the true phase transition response of the phase transition reference material after acoustic perturbation removal and magnetic relaxation verification. The acoustic gating weight is used to characterize the phase transition sensitivity at different thermal excitation phase positions. The initial coefficient field is used to characterize the initial distribution of thermal reflectivity coefficients in the thermal imaging region of the sample under test.

[0123] First, based on the phase transition state verification results in the phase transition confirmation features, the consistency between the net acoustic features and the magnetic relaxation phase transition features, the phase cross-verification results, and the stability of the phase transition reference time, the reliability of the phase transition anchoring process is determined. Simultaneously, combined with acoustic gating weights, the contribution of each thermally excited phase interval to phase transition anchoring and coefficient inversion is assessed. Phase intervals with high acoustic gating weights and stable phase transition confirmation features are assigned higher reliability; phase intervals with low acoustic gating weights, more residual acoustic disturbances, insufficient magnetic relaxation verification, or large phase offsets are assigned lower reliability. This generates the anchoring reliability weights.

[0124] Subsequently, the anchoring reliability weights are mapped to the spatial locations or local regions corresponding to the initial coefficient field. For spatial locations that mainly rely on high-reliability anchoring intervals during coefficient inversion, their initial coefficient values ​​are retained or only slightly smoothed. For spatial locations that are significantly affected by low-reliability anchoring intervals during coefficient inversion, neighborhood consistency correction, local revaluation, or weighted fusion are performed. Through this method, the initial coefficient field is corrected using the anchoring reliability weights to obtain the corrected coefficient field. The corrected coefficient field is used to characterize the thermal reflectivity coefficient distribution results after phase change anchoring reliability evaluation and spatial correction.

[0125] In step S53, "obtaining the correction coefficient field" further includes: First, the initial coefficient field is partitioned according to the anchored reliability weight to obtain the partitioned coefficient field.

[0126] In this embodiment, the reliability of each spatial location in the initial coefficient field is evaluated using the anchoring reliability weight and the initial coefficient field as input. Specifically, the initial coefficient field is divided into different reliability regions based on the anchoring reliability weight, coefficient inversion residual, reflectivity response continuity, neighborhood coefficient difference, and the degree of support from phase transition confirmation features corresponding to each spatial location. The reliability regions can include high reliability regions, medium reliability regions, and low reliability regions; where a high reliability region indicates that the coefficient inversion process in that region has relatively stable absolute temperature mapping support and good reflectivity response continuity, and a low reliability region indicates that the region may be affected by residual noise, abnormal reflection, local occlusion, heat transfer errors, or phase transition anchoring instability.

[0127] When performing reliability partitioning, threshold segmentation, region clustering, neighborhood consistency comparison, or reliability level mapping can be used. For spatially continuous regions with similar anchored reliability weights, they are divided into the same reliability region; for isolated abnormal pixels or coefficient points that are significantly inconsistent with their neighbors, they are marked as locations to be corrected. After partitioning, the coefficient values ​​in the initial coefficient field are bound to the corresponding reliability levels to obtain the partitioned coefficient field. This partitioned coefficient field is used to distinguish the correction methods for different spatial locations and serves as input for generating the coefficient correction mask.

[0128] Secondly, the partitioned coefficient field is subjected to reliability segmentation and boundary detection, low reliability coefficient regions and boundary preservation regions are marked, and a coefficient correction mask is generated.

[0129] In this embodiment, using a partitioned coefficient field as input, the partitioned coefficient field is first segmented according to its reliability level to identify low-reliability coefficient regions that require correction. These low-reliability coefficient regions may include regions where the anchoring reliability weight is lower than a preset requirement, regions where the coefficient value differs abnormally from the neighborhood, regions with discontinuous reflectivity response, regions with strong local noise, or regions with insufficient phase transition anchoring support. These regions are marked for subsequent neighborhood consistency correction.

[0130] Simultaneously, spatial boundary detection is performed on the partitioned coefficient field to identify material boundaries, structural boundaries, film steps, metal interconnect edges, or other real physical boundaries in the thermal imaging region of the sample's reflectance. In practice, the boundary preservation region can be determined by combining the spatial gradient of the initial coefficient field, image edge information in the light intensity sequence, spatial variations in reflectance evolution characteristics, and differences in neighboring coefficients. For boundary preservation regions, even if their coefficient values ​​differ significantly from adjacent regions, they are not directly considered anomalous noise but are marked as regions where boundary characteristics need to be preserved.

[0131] After marking the low-reliability coefficient regions and boundary-preserving regions, a coefficient correction mask is generated. This mask indicates which regions require neighborhood consistency correction, which require boundary preservation, and which regions can retain their original coefficient values. This mask serves as the control basis for subsequent differential correction processing.

[0132] Finally, based on the coefficient correction mask, neighborhood consistency correction is performed on the low reliability coefficient region, and boundary preservation processing is performed on the boundary-preserving region to obtain the correction coefficient field.

[0133] In this embodiment, a coefficient correction mask and a partitioned coefficient field are used as inputs, and different correction methods are applied to different marked regions. For low-reliability coefficient regions marked in the coefficient correction mask, high-reliability or medium-reliability coefficient points are selected within their neighborhood as references, and neighborhood consistency correction is performed on the low-reliability coefficient values ​​based on spatial distance, reliability level, similarity of reflectivity evolution, and material region consistency. In specific implementation, neighborhood weighted averaging, local fitting reestimation, interpolation of similar regions, or reliable neighborhood replacement methods can be used to ensure that the coefficient values ​​of low-reliability coefficient regions maintain reasonable continuity with the surrounding reliable regions, thereby reducing the impact of abnormal noise or anchoring errors on the coefficient field.

[0134] For the boundary preservation regions marked in the coefficient correction mask, a boundary preservation processing method is adopted. Specifically, when performing smoothing or neighborhood fusion, the propagation of coefficients across the real material boundary is restricted to avoid forcibly smoothing the thermal reflectivity coefficients of different material regions to similar values. For cases where there are stable coefficient distributions on both sides of the material boundary, local consistency processing is performed inside both sides of the boundary; for cases where there are transition pixels on the boundary line, local smoothing can be performed according to the boundary direction, or restricted fusion can be performed based on the coefficient values ​​on both sides of the boundary to maintain the differences of the real material.

[0135] After completing the neighborhood consistency correction for the low reliability coefficient region and the boundary preservation processing for the boundary-preserving region, the corrected coefficient values ​​are remapped to the spatial location of the thermal imaging region of the sample under test, forming a correction coefficient field. This correction coefficient field corrects the anomalous coefficients in the low reliability region while preserving the actual differences in thermal reflectivity coefficients at the boundaries of different materials or structures, and can serve as the core content of the final thermal reflectivity coefficient calibration result.

[0136] S54: Generate the thermal reflectivity coefficient calibration result based on the correction coefficient field.

[0137] In this embodiment, the calibration coefficient field obtained in step S53 is used as input to generate thermal reflectivity coefficient calibration results according to the usage requirements of reflectivity thermal imaging. The calibration coefficient field includes the thermal reflectivity coefficients corresponding to each spatial location in the thermal imaging region of the sample under test. After reliability correction and boundary preservation processing, it can be used to subsequently convert the real-time acquired reflectivity changes into temperature changes.

[0138] The calibration coefficient field can be converted into a two-dimensional coefficient matrix, a pixel-level lookup table, a partitioned coefficient table, a spatial coefficient cloud map, or a calibration file corresponding to the image coordinates of the sample under test. For pixel-level calibration applications, the thermal reflectivity coefficient corresponding to each pixel position is written into the two-dimensional coefficient matrix; for partitioned calibration applications, a partitioned coefficient table can be generated according to material regions, structural regions, or reliability partitions; for visualization and quality inspection applications, the calibration coefficient field can be converted into a thermal reflectivity coefficient cloud map, and low reliability coefficient regions, boundary preservation regions, and calibration mask information can be output simultaneously.

[0139] To facilitate subsequent use in reflectivity thermal imaging temperature measurement, the absolute temperature mapping relationship, phase transition reference time, known phase transition temperature of the phase transition reference material, anchoring reliability weight, correction coefficient field, and calibration time information can be encapsulated together to form a complete thermal reflectivity coefficient calibration result. This thermal reflectivity coefficient calibration result can be called by the reflectivity thermal imaging system to convert real-time reflectivity changes into corresponding temperature changes based on the correction coefficients at different spatial locations during actual temperature measurement, thereby obtaining a more accurate full-field temperature distribution. Example 2

[0140] The present invention also provides a thermal reflectivity coefficient calibration device for reflectivity thermal imaging. (Reference) Figure 2 As shown, the calibration device is suitable for full-field calibration of the thermal reflectivity coefficient of the thermal imaging area of ​​the sample under test. The calibration device is set up around the same thermal excitation process. The sample under test and the phase transition reference material with a known phase transition temperature are under the same preset thermal excitation program. The optical acquisition channel, acoustic acquisition channel, and magnetic relaxation acquisition channel are used to acquire the light intensity sequence, acoustic response, and magnetic relaxation response, respectively. The acquisition results are input into the subsequent processing module for synchronous processing, phase transition identification, temperature mapping, and coefficient calibration.

[0141] Specifically, the calibration device includes a thermal excitation execution unit, an optical acquisition channel, an acoustic acquisition channel, a magnetic relaxation acquisition channel, a multimodal synchronous acquisition module 100, a phase transition feature extraction module 200, a candidate window generation module 300, a temperature mapping establishment module 400, and a coefficient calibration output module 500. The thermal excitation execution unit applies thermal excitation to the sample under test and the phase transition reference material according to a preset thermal excitation program, causing the thermal imaging region of the sample under test and the phase transition reference material to form corresponding thermal excitation processes. The optical acquisition channel acquires the light intensity sequence of the thermal imaging region of the sample under test. The acoustic acquisition channel acquires the acoustic response of the phase transition reference material. The magnetic relaxation acquisition channel acquires the magnetic relaxation response of the phase transition reference material. All three acquisition channels are oriented towards the same thermal excitation process and are constrained by the same time reference to facilitate the subsequent formation of multimodal synchronous data.

[0142] The multimodal synchronous acquisition module 100 is connected to the optical acquisition channel, the acoustic acquisition channel, and the magnetic relaxation acquisition channel, respectively. It is used to synchronously acquire the light intensity sequence of the thermal imaging area of ​​the reflectivity of the sample under test, the acoustic response of the phase transition reference material with a known phase transition temperature, and the magnetic relaxation response of the phase transition reference material in real time during the process of applying thermal excitation according to the preset thermal excitation program. It also performs delay compensation and time alignment on the acquisition results to obtain multimodal synchronous data.

[0143] The phase transition feature extraction module 200 is connected to the multimodal synchronous acquisition module 100 and is used to extract reflectivity evolution features, acoustic phase transition features and magnetic relaxation phase transition features according to the multimodal synchronous data, and generate a phase transition recognition feature set according to the correspondence of the thermal excitation phase axis.

[0144] The candidate window generation module 300 is connected to the phase transition feature extraction module 200 and is used to determine the phase transition candidate time window based on the acoustic phase transition features in the phase transition recognition feature set, and to time-synchronize the phase transition candidate time window with the reflectivity evolution feature and the magnetic relaxation phase transition feature to obtain the phase transition candidate window feature.

[0145] The temperature mapping establishment module 400 is connected to the candidate window generation module 300 and is used to perform phase transition consistency judgment on the acoustic phase transition features and the magnetic relaxation phase transition features according to the phase transition candidate window features, determine the phase transition reference time, and establish an absolute temperature mapping relationship based on the phase transition reference time and the known phase transition temperature of the phase transition reference material.

[0146] The coefficient calibration output module 500 is connected to the temperature mapping establishment module 400. It is used to perform full-field thermal reflectivity coefficient inversion on the reflectivity evolution characteristics based on the absolute temperature mapping relationship, and to correct the inversion result to obtain a thermal reflectivity coefficient calibration result. The thermal reflectivity coefficient calibration result may include a correction coefficient field, an absolute temperature mapping relationship, a calibration file or lookup table, and can be called by the reflectivity thermal imaging temperature measurement process to obtain the full-field temperature distribution of the sample under test based on the correction coefficients at different spatial locations.

[0147] In this embodiment, the multimodal synchronous acquisition module 100, phase transition feature extraction module 200, candidate window generation module 300, temperature mapping establishment module 400, and coefficient calibration output module 500 are connected sequentially, so that the data output by the previous module serves as the processing input for the next module. Thus, the calibration device can realize a continuous processing flow from multimodal synchronous acquisition, phase transition feature extraction, candidate window generation, absolute temperature mapping establishment to the output of the full-field thermal reflectivity coefficient calibration result. The specific processing procedures performed by each module of the calibration device can be referred to the description of the corresponding method steps in Embodiment 1, and will not be repeated here.

[0148] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0149] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0150] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. A method for calibrating the thermal reflectivity coefficient for reflectivity thermal imaging, characterized in that, The calibration method includes the following steps: During the process of applying thermal excitation according to the preset thermal excitation program, the light intensity sequence of the reflectivity thermal imaging area of ​​the sample under test, the acoustic response of the phase transition reference material with a known phase transition temperature, and the magnetic relaxation response are collected in real time to obtain multimodal synchronous data. Based on the multimodal synchronization data, reflectivity evolution features, acoustic phase transition features, and magnetic relaxation phase transition features are extracted respectively to obtain a phase transition recognition feature set; Based on the acoustic phase transition features in the phase transition recognition feature set, a phase transition candidate time window is determined, and the phase transition candidate time window is time-synchronized with the reflectivity evolution feature and the magnetic relaxation phase transition feature to obtain the phase transition candidate window feature; Based on the phase transition candidate window features, the acoustic phase transition features and the magnetic relaxation phase transition features are subjected to phase transition consistency joint judgment to determine the phase transition reference time, and an absolute temperature mapping relationship is established based on the phase transition reference time and the known phase transition temperature. Based on the absolute temperature mapping relationship, the full-field thermal reflectivity coefficient is inverted on the reflectivity evolution characteristics, and the inversion result is corrected to obtain the thermal reflectivity coefficient calibration result.

2. The method for calibrating the thermal reflectivity coefficient for reflectivity thermal imaging according to claim 1, characterized in that, The obtained multimodal synchronization data includes: Based on the preset thermal excitation program and the known phase transition temperature of the phase transition reference material, the basic acquisition sequence is determined, and the acoustic response of the phase transition reference material is acquired under the basic acquisition sequence to obtain the acoustic acquisition indication quantity. Based on the acoustic leader acquisition indication, the acquisition frame rate of the light intensity sequence, the acquisition interval of the acoustic response, and the acquisition period of the magnetic relaxation response are adjusted in a coordinated manner to obtain the synchronous acquisition timing. According to the synchronous acquisition timing, the light intensity sequence of the reflectivity thermal imaging area of ​​the sample under test, the acoustic response of the phase change reference material, and the magnetic relaxation response are acquired in real time and synchronously. The acquisition results are then compensated for delay and time-aligned to obtain the multimodal synchronous data.

3. The method for calibrating the thermal reflectivity coefficient for reflectivity thermal imaging according to claim 2, characterized in that, The obtained phase transition recognition feature set includes: Based on the multimodal synchronization data and the preset thermal excitation program, the light intensity sequence, the acoustic response, and the magnetic relaxation response are converted from the acquisition time axis to the thermal excitation phase axis to obtain the phase synchronization sequence; Acoustic gating weights are constructed based on the acoustic response in the phase synchronization sequence, and phase transition sensitivity enhancement processing is performed on the light intensity sequence and the magnetic relaxation response based on the acoustic gating weights to obtain an enhanced multimodal sequence, wherein the acoustic gating weights are used to characterize the degree of phase transition sensitivity. Based on the enhanced multimodal sequence, reflectivity evolution features, acoustic phase transition features, and magnetic relaxation phase transition features are extracted respectively, and the phase transition recognition feature set is generated according to the correspondence of the thermal excitation phase axis.

4. The method for calibrating the thermal reflectivity coefficient for reflectivity thermal imaging according to claim 3, characterized in that, The obtained phase transition candidate window features include: Based on the acoustic phase transition characteristics, acoustic drift characteristics and periodic perturbation characteristics of the non-phase transition thermal excitation stage are extracted, and an acoustic perturbation template is established based on the acoustic drift characteristics and the periodic perturbation characteristics. Based on the acoustic disturbance template, pseudo-mutation removal is performed on the acoustic phase transition features to obtain the net acoustic features; Based on the net acoustic characteristics and the thermal phase constraint conditions corresponding to the preset thermal excitation program, determine the candidate time window for phase transition; Feature segments corresponding to the phase transition candidate time window are extracted from the reflectivity evolution features and the magnetic relaxation phase transition features, and the feature segments are time-aligned with the net acoustic features to generate the phase transition candidate window features.

5. A method for calibrating the thermal reflectivity coefficient for reflectivity thermal imaging according to claim 4, characterized in that, The establishment of the absolute temperature mapping relationship includes: Based on the phase transition candidate window features, the net acoustic features and the magnetic relaxation phase transition features are subjected to phase mutual verification processing to obtain phase mutual verification features. Based on the phase verification feature, the net acoustic feature is matched with the acoustic disturbance template, the residual acoustic disturbance component obtained by matching is removed, and the phase transition state of the removed net acoustic feature is verified by the magnetic relaxation phase transition feature to obtain the phase transition confirmation feature. Based on the phase transition confirmation characteristics and the thermal phase constraint conditions, the phase offset between the acoustic phase transition response and the magnetic relaxation phase transition response is determined, and the phase transition confirmation characteristics are anchored and corrected using the phase offset to obtain the phase transition reference time. Based on the phase transition reference time and the known phase transition temperature of the phase transition reference material, the absolute temperature mapping relationship is established.

6. A method for calibrating the thermal reflectivity coefficient for reflectivity thermal imaging according to claim 5, characterized in that, The obtained thermal reflectivity coefficient calibration results include: Based on the absolute temperature mapping relationship and the phase offset, the temperature axis of the reflectivity evolution characteristics is reconstructed to obtain the full-field temperature response characteristics; Based on the full-field temperature response characteristics, the thermal reflectivity coefficients corresponding to the spatial location of the thermal imaging region of the sample under test are inverted to obtain the initial coefficient field. Anchoring reliability weights are generated based on the phase transition confirmation features and the acoustic gating weights, and the initial coefficient field is corrected based on the anchoring reliability weights to obtain a corrected coefficient field. The thermal reflectivity coefficient calibration result is generated based on the correction coefficient field.

7. A method for calibrating the thermal reflectivity coefficient for reflectivity thermal imaging according to claim 6, characterized in that, The step of correcting the initial coefficient field based on the anchored reliability weight to obtain the corrected coefficient field includes: Based on the anchored reliability weights, the initial coefficient field is partitioned into reliability regions to obtain a partitioned coefficient field; The partitioned coefficient field is subjected to reliability segmentation and boundary detection, low reliability coefficient regions and boundary preservation regions are marked, and a coefficient correction mask is generated; Based on the coefficient correction mask, neighborhood consistency correction is performed on the low reliability coefficient region, and boundary preservation processing is performed on the boundary-preserving region to obtain the correction coefficient field.

8. A thermal reflectivity coefficient calibration apparatus for reflectivity thermal imaging, used to perform the calibration method as described in any one of claims 1 to 7, characterized in that, The calibration device includes: The multimodal synchronous acquisition module is used to synchronously acquire, in real time, the light intensity sequence of the reflectivity thermal imaging area of ​​the sample under test, the acoustic response of the phase transition reference material with a known phase transition temperature, and the magnetic relaxation response of the phase transition reference material during the process of applying thermal excitation according to a preset thermal excitation program, so as to obtain multimodal synchronous data. The phase transition feature extraction module is used to extract reflectivity evolution features, acoustic phase transition features and magnetic relaxation phase transition features based on the multimodal synchronization data, respectively, to obtain a phase transition recognition feature set; The candidate window generation module is used to determine the phase transition candidate time window based on the acoustic phase transition features in the phase transition recognition feature set, and to time-synchronize the phase transition candidate time window with the reflectivity evolution feature and the magnetic relaxation phase transition feature to obtain the phase transition candidate window feature; The temperature mapping establishment module is used to perform phase transition consistency judgment on the acoustic phase transition feature and the magnetic relaxation phase transition feature according to the phase transition candidate window feature, determine the phase transition reference time, and establish an absolute temperature mapping relationship based on the phase transition reference time and the known phase transition temperature. The coefficient calibration output module is used to perform full-field thermal reflectivity coefficient inversion on the reflectivity evolution characteristics according to the absolute temperature mapping relationship, and to correct the inversion results to obtain thermal reflectivity coefficient calibration results.

Images