Watch plasma nanofilm detection method and device based on image processing

Abstract

The present invention discloses a detection method and device for plasma nanofilm of watches based on image processing, wherein the detection method for plasma nanofilm of watches based on image processing collects dual polarization light intensity data of plasma nanofilm of watches through photobleaching effect excitation processing and phase modulation analysis, uses dynamic recovery characteristics and regional segmentation technology, selects the maximum point of photobleaching depth as a reference benchmark for differential processing, realizes the precise decoupling of geometric curvature effect and material effect, and thus accurately evaluates the intrinsic performance of nanofilm. The technical solution of the present invention can effectively solve the nonlinear interference problem caused by the curved surface of watches and provide reliable nanofilm quality detection results.

Description

Technical Field

[0001] The present invention relates to the technical field of watch surface treatment detection, and in particular to a method and device for detecting watch plasma nanofilm based on image processing. Background Art

[0002] Watch plasma nanofilm is a new type of functional nanostructured coating, which is usually formed by precious metal (such as gold, silver, etc.) nanoparticles deposited on the watch surface under specific process conditions. This nanofilm structure not only gives the watch surface unique optical properties, but also provides excellent waterproof and anti-scratch protection. At the nanoscale, the metal particles in the plasma nanofilm will produce a surface plasma resonance effect, which makes the nanofilm produce special selective absorption and reflection of incident light, thereby forming a unique optical response characteristic, which is also an important basis for the functionality and decorativeness of the watch plasma nanofilm.

[0003] In the watch production process, the quality of plasma nanofilm is directly related to the performance and appearance of the product, so it needs to be strictly inspected for quality. The current inspection equipment usually includes a light source system, an image acquisition device, a sample fixing platform and a data processing system. The inspection method is to illuminate the watch surface with a fixed light source, and then collect the reflection image of the nanofilm by the camera for analysis. However, this traditional inspection method has encountered serious technical bottlenecks in practical applications. Specifically, when the light source is irradiated on the watch surface, due to the complex three-dimensional curvature of the watch surface (including spherical, cylindrical or irregular surfaces), the incident light will produce different degrees of reflection and scattering in different areas. At the same time, the surface plasma resonance effect of the plasma nanofilm will produce nonlinear light interference phenomenon related to the observation angle in the edge area of ​​the curved surface. This interference phenomenon is completely different from the simple reflection of ordinary metal surfaces, and it contains complex optical response information. For example, when observing the nanofilm at the junction of the watch bezel and the watch mirror, even if the quality of the nanofilm is exactly the same, the difference in optical response caused by the change in curvature will cause significant deviations in the measurement results. Traditional image processing methods cannot accurately distinguish whether these differences are caused by geometric distortion due to surface curvature or by performance differences in the nanofilm itself, which seriously affects the accuracy and reliability of the detection results. Summary of the invention

[0004] The main purpose of the present invention is to solve the technical problem that the existing watch plasma nanofilm detection technology cannot accurately distinguish between the geometric distortion caused by the curved surface and the difference in the intrinsic performance of the nanofilm.

[0005] The first aspect of the present invention provides a method for detecting plasma nanofilm of a watch based on image processing, and the method for detecting plasma nanofilm of a watch based on image processing comprises:

[0006] Collecting dual-polarization light intensity distribution data of the plasma nanofilm of the watch before and after photobleaching within a preset incident angle range, wherein the dual-polarization light intensity distribution data includes s-polarization and p-polarization intensity data before photobleaching and s-polarization and p-polarization intensity data after photobleaching;

[0007] Phase modulating the dual-polarization light intensity distribution data, recording a phase-intensity response curve within a preset phase range, and obtaining amplitude-phase characteristic data characterizing local plasma resonance characteristics through Fourier transform processing;

[0008] Extracting the dynamic recovery characteristics of the photobleaching area based on the dual-polarization light intensity distribution data and the amplitude-phase characteristic data, and generating a regional segmentation feature map according to the spatial distribution law of the dynamic recovery characteristics;

[0009] According to the regional segmentation characteristic map, the maximum photobleaching depth point in each region is selected as a reference benchmark, and the optical response of other points in the region is differentially processed with the reference benchmark to obtain pure material effect distribution data that eliminates the influence of geometric curvature;

[0010] Performing dimensionality reduction processing on characteristic parameters in the pure material effect distribution data, extracting main characteristic components characterizing plasma resonance characteristics, and obtaining performance evaluation data;

[0011] The spatial distribution relationship of the regional performance parameters is established based on the performance evaluation data, and the quantitative detection results characterizing the intrinsic performance of the nanofilm are generated.

[0012] Optionally, the collecting of dual-polarization light intensity distribution data of the plasma nanofilm of the watch before and after photobleaching within a preset incident angle range includes:

[0013] Forming a photobleaching area array with a preset spacing on the sample surface, wherein the photobleaching area array includes a plurality of independent photobleaching spots;

[0014] Scanning the photobleaching area array at an angle of 0-75 degrees, collecting a set of polarization state light intensity data every 5 degrees, and obtaining original light intensity data before photobleaching;

[0015] Performing photobleaching excitation on the photobleaching area array, scanning in the same angle stepping manner as the 0-75 degree angle scanning, collecting polarization state light intensity data of the photobleaching area array, and obtaining excited state light intensity data after photobleaching;

[0016] Perform polarization state separation according to the original light intensity data and the excited state light intensity data to obtain s-polarization intensity data and p-polarization intensity data before photobleaching;

[0017] Perform polarization state separation according to the excited state light intensity data to obtain s-polarization intensity data and p-polarization intensity data after photobleaching;

[0018] The s-polarization and p-polarization intensity data before photobleaching and the s-polarization and p-polarization intensity data after photobleaching are combined into dual-polarization state light intensity distribution data.

[0019] Optionally, the dual polarization light intensity distribution data is phase modulated, a phase-intensity response curve is recorded within a preset phase range, and amplitude-phase characteristic data characterizing local plasma resonance characteristics is obtained through Fourier transform processing, including:

[0020] Performing phase scanning on the dual polarization light intensity distribution data in the range of 0-2π with a step length of 0.1π to obtain a phase scanning sequence before and after photobleaching;

[0021] Calculating local polarization state ellipse parameters according to the s-polarization and p-polarization intensity data of each phase point in the phase scanning sequence to obtain phase-intensity modulation data;

[0022] Performing discrete Fourier transform on the phase-intensity modulation data, extracting first-order and second-order Fourier components, and obtaining initial amplitude-phase characteristic quantities;

[0023] Calculate the local plasma resonance intensity coefficient and the phase delay coefficient according to the initial amplitude-phase characteristic quantity to obtain the resonance characteristic parameters;

[0024] Normalizing the resonance characteristic parameters to generate standardized amplitude-phase characteristic data;

[0025] The standardized amplitude-phase characteristic data is correlated and matched with the phase-intensity modulation data of the corresponding position to obtain the amplitude-phase characteristic data characterizing the local plasma resonance characteristics.

[0026] Optionally, extracting the dynamic recovery characteristics of the photobleaching area based on the dual-polarization light intensity distribution data and the amplitude-phase characteristic data, and generating a regional segmentation feature map according to the spatial distribution law of the dynamic recovery characteristics, includes:

[0027] Continuously sampling the dual-polarization light intensity distribution data at preset time intervals to obtain time-series intensity variation data of the photobleaching region;

[0028] Extract the recovery time constant and recovery amplitude parameter of each sampling point according to the time series intensity variation data to obtain initial dynamic characteristic data;

[0029] Performing cross-correlation analysis on the amplitude-phase characteristic data and the initial dynamic characteristic data to obtain comprehensive dynamic parameters characterizing the response characteristics of the material;

[0030] Performing spatial gradient analysis on the comprehensive dynamic parameters to obtain a dynamic characteristic difference distribution map of a local area;

[0031] According to the spatial continuity of the dynamic feature difference distribution map, adjacent regions are dynamically clustered to obtain a set of regional boundary feature points;

[0032] A closed curve is fitted to the region boundary feature point set to generate a region segmentation feature map.

[0033] Optionally, according to the regional segmentation characteristic map, the maximum point of photobleaching depth in each region is selected as a reference benchmark, and the optical response of other points in the region is differentially processed with the reference benchmark to obtain pure material effect distribution data eliminating the influence of geometric curvature, including:

[0034] Performing light intensity gradient calculation on each region in the region segmentation feature map to obtain photobleaching depth distribution data within the region;

[0035] Perform extreme point detection on each region according to the photobleaching depth distribution data, and select a candidate point set with the maximum photobleaching depth;

[0036] Calculating the local curvature of each point in the candidate point set to obtain curvature feature data of the candidate point;

[0037] Optimizing the candidate point set according to the curvature feature data to determine the final reference reference point in each area;

[0038] Performing a ratio operation on the light intensity data of other points in the area and the light intensity data of the reference point to obtain normalized light intensity distribution data;

[0039] The normalized light intensity distribution data is subjected to curvature compensation processing to generate pure material effect distribution data that eliminates the influence of geometric curvature.

[0040] Optionally, the step of performing dimensionality reduction processing on characteristic parameters in the pure material effect distribution data, extracting main characteristic components characterizing plasma resonance characteristics, and obtaining performance evaluation data includes:

[0041] Performing nonlinear transformation on the resonance peak position, peak width and peak intensity in the pure material effect distribution data to obtain a feature point set in a multidimensional feature space;

[0042] Performing principal component-extreme value decomposition according to the feature point set, extracting the change trend in the main feature direction, and obtaining an orthogonal feature vector group;

[0043] Performing singular value decomposition on each component of the orthogonal eigenvector group to obtain feature weight distribution data;

[0044] Reordering the feature weight distribution data according to weights to obtain a sorted feature component sequence;

[0045] Reconstructing the feature components whose contribution rates in the feature component sequence exceed a preset threshold value to obtain feature data with reduced dimensions;

[0046] The feature data after dimension reduction is mapped and matched with the plasma resonance theory features to generate performance evaluation data.

[0047] Optionally, performing singular value decomposition on each component of the orthogonal eigenvector group to obtain feature weight distribution data includes:

[0048] The orthogonal eigenvector group is constructed into a characteristic matrix, and singular value decomposition is performed to obtain a left singular matrix, a singular value matrix and a right singular matrix;

[0049] Performing orthogonality analysis on the left singular matrix and the right singular matrix to obtain a basis vector group of the feature space;

[0050] Calculate the energy contribution coefficient of each characteristic component according to the size of the diagonal elements of the singular value matrix;

[0051] Performing hierarchical clustering analysis based on the energy contribution coefficient to obtain a feature hierarchical tree structure;

[0052] Pruning and optimizing the feature hierarchical tree structure to obtain a key feature node sequence;

[0053] The weight coefficients are recalculated according to the key feature node sequence to generate feature weight distribution data.

[0054] Optionally, establishing a spatial distribution relationship of regional performance parameters according to the performance evaluation data to generate a quantitative test result characterizing the intrinsic performance of the nanofilm includes:

[0055] Performing local response characteristic analysis on the performance evaluation data to obtain a plasma resonance intensity index, a kinetic stability index, and a uniformity index of each region;

[0056] The plasma resonance intensity index, the kinetic stability index and the uniformity index are combined according to a preset weighting coefficient to obtain a regional comprehensive performance index;

[0057] Performing Delaunay triangulation on the test area according to the comprehensive performance index of the area to obtain a spatial interpolation grid of performance parameters;

[0058] Performing thin plate spline interpolation on the node data of the spatial interpolation grid to obtain a continuous performance distribution function;

[0059] Calculating local gradients and curvatures based on the performance distribution function to generate a performance change characteristic graph;

[0060] The performance change characteristic diagram is fused and superimposed with the performance distribution function to obtain a quantitative detection result characterizing the intrinsic performance of the nanofilm.

[0061] A second aspect of the present invention provides a detection device for plasma nanofilm of a watch based on image processing, the detection device for plasma nanofilm of a watch based on image processing comprising:

[0062] A data acquisition module, used to collect dual-polarization light intensity distribution data of the plasma nanofilm of the watch before and after photobleaching within a preset incident angle range, wherein the dual-polarization light intensity distribution data includes s-polarization and p-polarization intensity data before photobleaching and s-polarization and p-polarization intensity data after photobleaching;

[0063] A phase modulation module is used to phase modulate the dual-polarization light intensity distribution data, record a phase-intensity response curve within a preset phase range, and obtain amplitude-phase characteristic data characterizing the local plasma resonance characteristics through Fourier transform processing;

[0064] A dynamic feature extraction module, used to extract the dynamic recovery feature of the photobleaching area based on the dual-polarization light intensity distribution data and the amplitude-phase feature data, and generate a regional segmentation feature map according to the spatial distribution law of the dynamic recovery feature;

[0065] A decoupling analysis module is used to select the maximum photobleaching depth point in each region as a reference benchmark according to the regional segmentation characteristic map, and perform differential processing on the optical response of other points in the region and the reference benchmark to obtain pure material effect distribution data that eliminates the influence of geometric curvature;

[0066] A dimensionality reduction processing module is used to perform dimensionality reduction processing on the characteristic parameters in the pure material effect distribution data, extract the main characteristic components that characterize the plasma resonance characteristics, and obtain performance evaluation data;

[0067] The performance evaluation module is used to establish the spatial distribution relationship of regional performance parameters according to the performance evaluation data, and generate quantitative detection results characterizing the intrinsic performance of the nanofilm.

[0068] The technical solution provided by the embodiment of the present application has at least the following advantages:

[0069] First, the photobleaching effect excitation treatment is combined with the dual-polarization light intensity distribution data collection method to make the material effect produce significant differences before and after photobleaching, while the geometric curvature effect remains unchanged, providing a physical basis for subsequent decoupling analysis. Secondly, the amplitude-phase characteristic data is obtained through phase modulation and Fourier transform processing, and the regional segmentation is performed in combination with the spatial distribution law of the dynamic recovery characteristics, effectively identifying regions with similar optical response characteristics. Finally, a local reference benchmark method based on the maximum point of photobleaching depth is introduced, and the influence of geometric curvature is eliminated through differential processing, realizing the accurate extraction of pure material effects. This multi-level decoupling analysis method not only solves the problem of difficulty in separating curvature distortion and material properties in traditional detection, but also provides a reliable technical means for the quantitative evaluation of nanofilm performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0071] Figure 1 A schematic diagram of an embodiment of a method for detecting plasma nanofilm of a watch based on image processing in an embodiment of the present invention;

[0072] Figure 2 Schematic diagram of an embodiment of a detection device for plasma nanofilm of a watch based on image processing in an embodiment of the present invention.

[0073] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0074] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0075] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back...), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0076] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, "and / or" in the full text includes three solutions. Taking A and / or B as an example, it includes technical solution A, technical solution B, and technical solution that satisfies both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, which must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0077] An embodiment of the present application provides a method for detecting plasma nanofilms in watches based on image processing. Figure 1 A flow chart of a method for detecting plasma nanofilm of a watch based on image processing provided in an embodiment of the present application. In this embodiment, the method includes:

[0078] See also Figure 1 , collecting dual-polarization light intensity distribution data of the plasma nanofilm of the watch before and after photobleaching within a preset incident angle range, wherein the dual-polarization light intensity distribution data includes s-polarization and p-polarization intensity data before photobleaching and s-polarization and p-polarization intensity data after photobleaching;

[0079] In one embodiment of the present invention, the dual polarization state light intensity distribution data of the plasma nanofilm of the watch before and after photobleaching is collected within a preset incident angle range, including: forming a photobleaching area array with a preset spacing on the sample surface, the photobleaching area array including a plurality of independent photobleaching spots; performing an angle scan of 0-75 degrees on the photobleaching area array, collecting a set of polarization state light intensity data every 5 degrees, and obtaining the original light intensity data before photobleaching; performing photobleaching excitation on the photobleaching area array, at the same angle as the 0-75 degree angle scan Scanning is performed in a stepping manner to collect polarization state light intensity data of the photobleaching area array to obtain excited state light intensity data after photobleaching; polarization state separation is performed according to the original light intensity data and the excited state light intensity data to obtain s polarization intensity data and p polarization intensity data before photobleaching; polarization state separation is performed according to the excited state light intensity data to obtain s polarization intensity data and p polarization intensity data after photobleaching; and the s polarization and p polarization intensity data before photobleaching and the s polarization and p polarization intensity data after photobleaching are combined into dual polarization state light intensity distribution data.

[0080] Specifically, in the detection process of watch plasma nanofilm, the first step is to construct a photobleaching area array on the sample surface. The construction of the photobleaching area array adopts an array laser irradiation method. By controlling the irradiation time and energy density of the laser, a regularly distributed photobleaching area is formed on the watch surface. The so-called photobleaching area refers to the area where the plasma resonance effect in the nanomaterial is temporarily suppressed under strong light irradiation. The design of the size and spacing of each photobleaching area needs to consider two factors: one is to ensure that a single area can fully reflect the local material properties, and the other is to avoid interference between adjacent areas. The preset value of the spacing of the photobleaching area is mainly based on two key factors: the spatial attenuation characteristics of the plasma resonance effect and the change gradient of the watch surface curvature. Specifically, according to experiments, when the size of the photobleaching area is set at 50-100 microns and the spacing is set at 200 microns, it can not only ensure that the photobleaching effects of adjacent areas will not interfere with each other, but also ensure that the sampling density is sufficient to reflect the changing characteristics of the surface curvature. For example, when detecting the nanofilm on the bezel of a watch, if the photobleaching area is too large, it may span areas of different curvatures, causing distortion of the measurement results; if the spacing is too small, the photobleaching effects of adjacent areas may affect each other. This array design has obvious advantages: it not only ensures the spatial resolution of the detection, but also provides enough reference points for subsequent decoupling analysis.

[0081] After obtaining the photobleaching area array, angle scanning data acquisition is required. The angle scan starts from 0 degrees (perpendicular to the sample surface) and collects a set of data every 5 degrees until 75 degrees. At each angle position, the light intensity data of two polarization states, s polarization (vibration direction perpendicular to the incident plane) and p polarization (vibration direction parallel to the incident plane), are collected respectively. The s polarization intensity data and p polarization intensity data here respectively reflect the response characteristics of the material to light in different polarization states. For example, when detecting the nanofilm on the edge of the watch mirror, at an incident angle of 45 degrees, p polarization light will show stronger sensitivity to the plasma resonance effect, while s polarization light is mainly affected by geometric curvature. This full-angle, dual-polarization state data acquisition method can provide rich material response information and lay the foundation for subsequent characteristic decoupling.

[0082] Next, the array of marked photobleached areas is photobleached. Photobleaching excitation refers to the use of higher intensity laser irradiation to put the material into a photobleached state. In this state, the plasma resonance effect of the material is significantly suppressed, while the geometric features of the surface remain unchanged. Subsequently, the data must be re-collected in exactly the same angle scanning method as before. This requirement for angle step consistency stems from the need for decoupling analysis. Only at the same observation angle can the changes in material response be accurately compared to ensure that the data before and after photobleaching have a strict spatial correspondence. For example, for the curved area at the edge of a watch dial, the data before photobleaching contains information on both curvature scattering and material resonance, while the data after photobleaching mainly reflects the influence of geometric curvature. This before-and-after comparison acquisition method has an important advantage: it can selectively modulate the optical response of the material without changing the geometric morphology of the sample.

[0083] When performing polarization state separation processing on the collected data, a polarization optical system is used to decompose the mixed optical signal into two independent components, s-polarization and p-polarization. The physical basis of this separation process is that p-polarized light will experience significant reflectivity suppression under plasma resonance conditions, while s-polarized light is mainly affected by geometric scattering. For example, when detecting the concave and convex edges of the watch surface, by comparing the response differences between the two polarization states, it is possible to effectively distinguish which signal changes are caused by material properties and which are caused by surface morphology. The advantage of this polarization analysis method is that it provides two independent but complementary information channels, enhancing the reliability of feature extraction.

[0084] The last step is data combination, which requires constructing a data matrix according to the three dimensions of spatial position, polarization state, and photobleaching state, and combining the s-polarization and p-polarization intensity data before and after photobleaching into a complete dual-polarization light intensity distribution data set according to the time sequence and spatial position. For each detection point, a four-dimensional data set containing spatial coordinates (xy), polarization state (sp), and states before and after photobleaching is finally formed. This structured data organization method needs to ensure the precise correspondence between different data. For example, for each detection point on the watch surface, a complete data set containing two polarization states before and after photobleaching is formed. The advantage of this multidimensional data combination is that it provides sufficient information redundancy for subsequent decoupling analysis and improves the accuracy of feature extraction. Through this systematic data acquisition and combination method, it is ultimately possible to achieve an accurate evaluation of the performance of the plasma nanofilm of the watch.

[0085] Please continue reading Figure 1 , performing phase modulation on the dual-polarization light intensity distribution data, recording a phase-intensity response curve within a preset phase range, and obtaining amplitude-phase characteristic data characterizing local plasma resonance characteristics through Fourier transform processing;

[0086] In one embodiment of the present invention, the phase modulation of the dual polarization light intensity distribution data, recording the phase-intensity response curve within a preset phase range, and obtaining amplitude-phase characteristic data characterizing the local plasma resonance characteristics through Fourier transform processing includes: performing phase scanning on the dual polarization light intensity distribution data in the range of 0-2π according to a step size of 0.1π to obtain a phase scanning sequence before and after photobleaching; calculating the local polarization state ellipse parameters according to the s-polarization and p-polarization intensity data of each phase point in the phase scanning sequence to obtain phase-intensity modulation data; performing discrete Fourier transform on the phase-intensity modulation data to extract the first-order and second-order Fourier components to obtain initial amplitude-phase characteristic quantities; calculating the local plasma resonance intensity coefficient and the phase delay coefficient according to the initial amplitude-phase characteristic quantities to obtain resonance characteristic parameters; normalizing the resonance characteristic parameters to generate standardized amplitude-phase characteristic data; and correlating and matching the standardized amplitude-phase characteristic data with the phase-intensity modulation data at the corresponding position to obtain amplitude-phase characteristic data characterizing the local plasma resonance characteristics.

[0087] Specifically, the process of phase characteristic analysis of dual polarization light intensity distribution data first requires phase scanning. Phase scanning refers to modulating the phase of the incident light in the range of 0-2π with a step size of 0.1π. This fine step size setting is to accurately capture the response characteristics of the material to phase changes. At each phase point, it is necessary to record the light intensity response before and after photobleaching to form a complete phase scanning sequence. The physical meaning of phase scanning is that by changing the phase of the incident light, the optical response of the material under different phase conditions can be stimulated, and these responses contain the intrinsic characteristic information of the material. For example, when scanning a certain area on the surface of a watch, if there is a plasma resonance effect there, then there will be obvious light intensity modulation at a specific phase point. The advantage of this phase scanning method is that it can comprehensively obtain the phase response characteristics of the material.

[0088] According to the obtained phase scanning sequence, the local polarization state ellipse parameters need to be calculated. Polarization state ellipse parameters are important indicators that describe the change of polarization state of light waves during propagation, including parameters such as ellipticity and azimuth. In the specific calculation, the s-polarization and p-polarization intensity data of each phase point are used to construct the polarization state ellipse through the Stokes parameter method. These parameters are directly related to the optical anisotropy and plasma resonance characteristics of the material. For example, when detecting the uniformity of the nanofilm on the surface of a watch, the change in polarization state ellipse parameters can reflect the local differences in the material structure. The phase-intensity modulation data obtained by this method contains rich information on material properties.

[0089] Next, the phase-intensity modulation data is processed by discrete Fourier transform. In the Fourier transform process, the first-order and second-order Fourier components are extracted, because these two components correspond to the linear response and nonlinear response characteristics of the material, respectively. The first-order component mainly reflects the basic periodic changes, while the second-order component contains the nonlinear optical effect information of the material. For example, when analyzing the plasma resonance effect of the nanofilm on the surface of a watch, the intensity of the second-order Fourier component can directly reflect the resonance intensity. The advantage of this Fourier analysis method is that it can convert complex time domain signals into frequency domain features that are easy to analyze.

[0090] Based on the initial amplitude-phase characteristic quantities obtained, the local plasma resonance characteristics are calculated. This calculation process includes two key parameters: the resonance intensity coefficient and the phase delay coefficient. The resonance intensity coefficient reflects the strength of the plasma resonance effect of the material, while the phase delay coefficient characterizes the phase response characteristics during the resonance process. For example, when detecting defects in the nanofilm on the surface of a watch, abnormal changes in these coefficients can indicate the location and nature of the defects. The advantage of this parametric characterization is that it converts complex physical processes into quantifiable indicators.

[0091] In order to eliminate the measurement differences between different areas, the resonance characteristic parameters need to be normalized. The normalization process takes into account the reflection characteristics of the local substrate and the response function of the measurement system, making the data from different areas comparable. This step is crucial to ensure the reliability of the measurement results. For example, when analyzing the performance of nanofilms in different curvature areas on the watch surface, normalization can eliminate the impact of curvature differences. The advantage of this standardization process is that it provides a unified evaluation standard.

[0092] The last step is to perform correlation matching, which spatially corresponds and features the standardized amplitude-phase feature data with the original phase-intensity modulation data. This process requires the establishment of an accurate position mapping relationship to ensure that the feature data at each spatial position can accurately reflect the local plasma resonance characteristics. For example, when generating a performance distribution map of a nanofilm on the surface of a watch, this correlation matching ensures the accuracy of feature extraction. An important advantage of this step is that it integrates multi-dimensional measurement data into a complete feature description, providing a reliable data basis for subsequent performance evaluation. This systematic phase analysis method ultimately achieves accurate characterization of the nanofilm plasma resonance characteristics.

[0093] Please continue reading Figure 1 , extracting the dynamic recovery characteristics of the photobleaching area based on the dual-polarization light intensity distribution data and the amplitude-phase characteristic data, and generating a regional segmentation feature map according to the spatial distribution law of the dynamic recovery characteristics;

[0094] In one embodiment of the present invention, the method of extracting the dynamic recovery characteristics of the photobleaching area based on the dual-polarization light intensity distribution data and the amplitude-phase characteristic data, and generating a regional segmentation feature map according to the spatial distribution law of the dynamic recovery characteristics, includes: continuously sampling the dual-polarization light intensity distribution data at preset time intervals to obtain time-series intensity change data of the photobleaching area; extracting the recovery time constant and recovery amplitude parameter of each sampling point according to the time-series intensity change data to obtain initial dynamic feature data; performing cross-correlation analysis on the amplitude-phase characteristic data and the initial dynamic feature data to obtain comprehensive dynamic parameters characterizing the response characteristics of the material; performing spatial gradient analysis on the comprehensive dynamic parameters to obtain a dynamic feature difference distribution map of the local area; dynamically clustering adjacent areas according to the spatial continuity of the dynamic feature difference distribution map to obtain a regional boundary feature point set; performing closed curve fitting on the regional boundary feature point set to generate a regional segmentation feature map.

[0095] Specifically, it is first necessary to perform a time series analysis on the dual polarization state light intensity distribution data. The time series analysis uses a continuous sampling method to dynamically monitor the photobleached area at a preset time interval (typically 100 milliseconds). The purpose of this continuous sampling is to capture the complete process of the material gradually recovering from the photobleached state to the initial state. The time series intensity change data records the optical response intensity of the material at different time points, and these data contain information on the dynamic recovery characteristics of the material. For example, when monitoring a photobleached area of ​​the nanofilm on the surface of a watch, the gradual change of light intensity over time can be observed, which reflects the optical relaxation characteristics of the material. The advantage of this time series sampling method is that it can fully record the dynamic response process of the material.

[0096] When extracting features from the acquired time-series intensity variation data, we focus on two key parameters: the recovery time constant and the recovery amplitude parameter. The recovery time constant describes the characteristic time required for the material to recover from the photobleached state to the initial state, while the recovery amplitude parameter reflects the amplitude of the light intensity change during this process. These two parameters directly reflect the optical stability and response characteristics of the material. For example, when analyzing nanofilms in different areas of the watch surface, if the recovery time of a certain area is significantly longer or the recovery amplitude is abnormal, it means that there are differences in the material properties at that location. The initial dynamic characteristic data obtained through this parameterized description provides a basis for subsequent material property analysis.

[0097] Next, the dynamic features need to be cross-correlated with the amplitude-phase feature data obtained previously. This analysis process is actually studying the correlation between the transient response characteristics of the material and the steady-state optical properties. By calculating the correlation coefficients between different feature quantities, dynamic parameters that characterize the comprehensive response characteristics of the material can be obtained. For example, when analyzing complex curved areas on the surface of a watch, this cross-correlation analysis can help identify which optical response changes are due to the intrinsic properties of the material and which are due to geometric effects. This multi-dimensional feature correlation analysis method has an important advantage: it can verify the performance characteristics of the material from multiple angles.

[0098] When performing spatial gradient analysis on the obtained comprehensive dynamic parameters, the parameter change rate between adjacent regions is calculated to obtain the dynamic characteristic difference distribution of the local region. This analysis process is actually constructing a spatial distribution map of material properties. Gradient analysis can highlight areas where material properties change suddenly, which are often boundaries where material properties differ significantly. For example, when analyzing the uniformity of the nanofilm on the surface of a watch, the characteristic difference distribution map can clearly identify the boundaries of areas with uneven performance. The advantage of this gradient-based analysis method is that it can accurately locate the area where material properties change.

[0099] Based on the dynamic feature difference distribution map, it is necessary to perform dynamic clustering analysis on adjacent regions. The clustering process takes into account the spatial continuity constraint, that is, the regions formed by clustering are required to remain continuous in space. This process uses an adaptive threshold clustering algorithm to group adjacent regions together based on the similarity of feature parameters. For example, when analyzing the nanofilm in the transition area between the watch bezel and the watch mirror, this clustering method can accurately identify the boundary position where the performance characteristics change. The regional boundary feature point set obtained in this way accurately reflects the spatial distribution characteristics of material performance.

[0100] The last step is to fit a closed curve to the feature point set of the regional boundary to generate a complete regional segmentation feature map. The fitting process uses the spline interpolation method to ensure the smoothness and continuity of the boundary curve. This closed curve description can not only accurately express the spatial distribution of material properties, but also facilitate subsequent quantitative analysis. For example, when generating a performance partition map of the nanofilm on the surface of a watch, the closed curve clearly defines the range of different performance areas. This method of regional segmentation has important advantages: it not only ensures the accuracy of the segmentation results, but also provides a visual expression that is easy to understand and analyze. Through this systematic dynamic analysis method, the precise characterization of the spatial distribution of nanofilm performance is finally achieved.

[0101] The innovation of the entire analysis process is reflected in: combining time series dynamic analysis with spatial distribution analysis, and finally achieving accurate characterization of material properties through multi-dimensional feature extraction and correlation analysis. This method can not only identify subtle differences in material properties, but also accurately locate the spatial distribution of these differences, providing reliable technical support for the quality control of nanofilms.

[0102] Please continue reading Figure 1 , according to the regional segmentation characteristic map, the maximum photobleaching depth point in each region is selected as a reference benchmark, and the optical response of other points in the region is differentially processed with the reference benchmark to obtain pure material effect distribution data that eliminates the influence of geometric curvature;

[0103] In one embodiment of the present invention, the method of selecting the maximum photobleaching depth point in each region as a reference benchmark according to the region segmentation feature map, performing differential processing on the optical responses of other points in the region and the reference benchmark to obtain pure material effect distribution data that eliminates the influence of geometric curvature includes: performing light intensity gradient calculation on each region in the region segmentation feature map to obtain photobleaching depth distribution data in the region; performing extreme point detection on each region according to the photobleaching depth distribution data to screen a candidate point set with the maximum photobleaching depth; performing local curvature calculation on each point in the candidate point set to obtain curvature feature data of the candidate point; optimizing the candidate point set according to the curvature feature data to determine the final reference benchmark point in each region; performing ratio calculation on the light intensity data of other points in the region and the light intensity data of the reference benchmark point to obtain normalized light intensity distribution data; performing curvature compensation processing on the normalized light intensity distribution data to generate pure material effect distribution data that eliminates the influence of geometric curvature.

[0104] Specifically, it is first necessary to calculate the light intensity gradient for each region in the regional segmentation feature map. The light intensity gradient reflects the rate of change of light intensity in space. By calculating the intensity difference between adjacent measurement points, the spatial distribution characteristics of the photobleaching depth in the region can be obtained. When calculating the light intensity gradient, a two-dimensional gradient operator is used to calculate the rate of change of light intensity in the x and y directions, respectively, and finally obtain complete photobleaching depth distribution data. For example, when analyzing a certain curvature change area on the surface of a watch, the light intensity gradient data can reflect the spatial inhomogeneity of the photobleaching effect. This inhomogeneity includes both the intrinsic response differences of the material and the geometric effects caused by the curvature. The advantage of this gradient analysis method is that it can clearly display the spatial variation characteristics of the photobleaching effect.

[0105] Based on the obtained photobleaching depth distribution data, extreme point detection is required to screen the candidate point set. Extreme point detection uses a local maximum search algorithm to find the location points with the largest photobleaching depth in each region. These extreme points represent the locations where the photobleaching effect is most significant and often correspond to the areas where the material response is most sensitive. For example, when analyzing the nanofilm on the edge of a watch mirror, extreme point detection can be used to find several locations with the strongest response to the photobleaching effect. These candidate points will serve as the basis for the subsequent selection of reference reference points. The advantage of this screening method based on extreme point detection is that it can automatically identify the most representative measurement points.

[0106] For the selected candidate point set, local curvature calculation is required. The curvature calculation adopts the local quadratic surface fitting method. Specifically, with each candidate point as the center, the measurement points within the radius R (generally 2-3 mm) around it are selected to construct a local coordinate system. In this local coordinate system, the quadratic polynomial z=ax²+by²+cxy+dx+ey+f is used to fit the spatial distribution of these points. In this equation, z represents the height value of the measurement point, x and y are the plane coordinates of the measurement point in the local coordinate system, the coefficients a, b, and c represent the second-order curvature terms in the x² direction and y² direction and the cross term of xy, respectively, the coefficients d and e represent the first-order slope terms in the x and y directions, and f is a constant term representing the local translation. These coefficients are solved by the least squares method to minimize the sum of squared errors between the fitted surface and the actual measurement point. Based on the obtained surface equation, the principal curvature at the point can be calculated ( , ) and Gaussian curvature (K = × ) and other characteristic parameters, among which represents the maximum principal curvature, represents the minimum principal curvature, and the directions of the two are perpendicular to each other. These curvature feature data directly reflect the geometric characteristics of the sample surface. For example, when analyzing the candidate points in the transition area between the watch bezel and the dial, if the calculated principal curvature is =0.5 , =0.2 , then this indicates that this is an elliptical transition surface, and the curvature in one direction is 2.5 times that in the other direction. The advantage of this local curvature analysis is that it can quantitatively characterize the geometric characteristics of each candidate point, providing an accurate mathematical basis for the subsequent curvature effect compensation.

[0107] Based on the obtained curvature feature data, the candidate point set is optimized to determine the final reference datum point. The optimization process takes two aspects into consideration: first, the photobleaching depth of the point should be large enough to ensure a high signal-to-noise ratio; second, the curvature characteristics at the point should be typical and able to represent the geometric characteristics of the area. For example, when selecting a reference datum point for a partition on the surface of a watch, the location where the photobleaching effect is most obvious and the curvature characteristics are most representative in the area should be selected. The advantage of this multi-dimensional optimization strategy is that it can ensure that the selected reference datum points are well representative.

[0108] After selecting the reference point, it is necessary to perform a ratio operation on the light intensity data of other points in the area and the light intensity data of the reference point. This ratio operation is actually a normalization process, the purpose of which is to eliminate the influence of the response function of the measurement system and make the data at different locations comparable. For example, when analyzing the performance of nanofilms at different locations on the watch surface, this normalization process can directly compare the relative response intensity at different locations. The advantage of this reference point-based normalization method is that it provides a unified evaluation standard.

[0109] The normalized light intensity distribution data is subjected to curvature compensation. The compensation process is based on the local curvature characteristics obtained previously, and the compensation calculation is performed by establishing a relationship model between light intensity response and curvature. Specifically, for the light intensity response I at any observation point P, it can be expressed as I= ×f( , ),in represents the intrinsic light intensity response of the material, f( , ) represents the geometric modulation function caused by curvature, and are the maximum and minimum principal curvatures at this point respectively. The geometric modulation function f( , )=exp[-α( + )-β| - |], where α represents the influence coefficient of curvature size and β represents the influence coefficient of curvature anisotropy. These two coefficients can be determined by the calibration data of the reference area. By solving this equation, we can get =I / f( , ), thereby eliminating the effect of geometric curvature. For example, when analyzing nanofilms in complex curvature areas on a watch surface, if the principal curvature at a certain point =0.5 , =0.2 , measure the light intensity I=100 (normalized unit), and obtain f( , )=0.85, then the material intrinsic response at this point =117.6, which represents the pure material response after removing the geometric effect. The advantage of this decoupling analysis method is that it establishes a clear mathematical relationship and can quantitatively obtain pure material characteristic data that is not affected by geometric morphology.

[0110] Through this systematic reference point selection and decoupling analysis method, the precise extraction of the intrinsic properties of the material was finally achieved. The innovation of the entire analysis process is reflected in: the use of a multi-level screening strategy to ensure the representativeness of the reference points, the precise separation of geometric effects and material effects through rigorous mathematical processing, and the provision of a reliable technical basis for the quantitative evaluation of nanofilm performance.

[0111] Please continue reading Figure 1 , performing dimensionality reduction processing on the characteristic parameters in the pure material effect distribution data, extracting the main characteristic components characterizing the plasma resonance characteristics, and obtaining performance evaluation data;

[0112] In one embodiment of the present invention, the dimension reduction processing of the characteristic parameters in the pure material effect distribution data is performed to extract the main characteristic components characterizing the plasma resonance characteristics to obtain performance evaluation data, including: performing nonlinear transformation on the resonance peak position, peak width and peak intensity in the pure material effect distribution data to obtain a characteristic point set in a multidimensional characteristic space; performing principal component-extreme value decomposition according to the characteristic point set to extract the change trend in the main characteristic direction to obtain an orthogonal characteristic vector group; performing singular value decomposition on each component of the orthogonal characteristic vector group to obtain characteristic weight distribution data; reordering the characteristic weight distribution data according to the weight size to obtain a sorted characteristic component sequence; reconstructing the characteristic components whose contribution rate exceeds a preset threshold in the characteristic component sequence to obtain characteristic data after dimension reduction; mapping and matching the characteristic data after dimension reduction with the theoretical characteristics of plasma resonance to generate performance evaluation data.

[0113] Specifically, it is first necessary to perform a feature transformation on the pure material effect distribution data. This transformation process mainly focuses on three key parameters: resonance peak position, peak width and peak intensity. Among them, the resonance peak position reflects the resonance frequency characteristics of the material, the peak width reflects the energy loss characteristics of the material, and the peak intensity reflects the resonance intensity of the material. Through nonlinear transformation, these original parameters can be converted into data points in the feature space. For example, when analyzing the plasma resonance characteristics of the nanofilm on the surface of a watch, the resonance peak of a certain area may appear at a specific wavelength position with a certain peak width and intensity. These three parameters are transformed into a point in the feature space after nonlinear transformation. The advantage of this nonlinear transformation is that it can highlight the key characteristics of the material response and reduce the impact of noise.

[0114] The formed feature point set needs to be subjected to principal component-extreme value decomposition analysis. This decomposition method combines the advantages of principal component analysis and extreme value analysis, and can identify key change features while retaining the main information of the data. Specifically, first find the directions in which the data changes most significantly in the feature space, which constitute a set of orthogonal eigenvectors. For example, when analyzing the performance of nanofilms in different areas of the watch surface, certain characteristic directions correspond to the main change trends of material properties, such as systematic changes in resonance intensity. The advantage of this decomposition method is that it can extract the most representative characteristic directions.

[0115] Next, the obtained orthogonal eigenvector group is subjected to singular value decomposition. Singular value decomposition can decompose complex data structures into the product of three matrices, where the singular values ​​directly reflect the importance of each characteristic component. Through this decomposition, the weight distribution of each characteristic component can be obtained. For example, when analyzing the performance distribution of the nanofilm on the surface of a watch, some characteristic components may have large singular values, indicating that these features are more important for the characterization of material properties. The advantage of this weight analysis method based on singular values ​​is that it can quantitatively evaluate the importance of each feature.

[0116] According to the obtained feature weight distribution data, reordering is required. The reordering process rearranges the feature components according to the weight size to form an ordered sequence of feature components. The purpose of this process is to highlight the main features and provide a basis for subsequent dimensionality reduction. For example, when analyzing a large amount of measurement data on the watch surface, reordering can clearly show which feature components contribute the most to the characterization of material properties. The advantage of this weight-based sorting method is that it can effectively identify key features.

[0117] For the sorted feature component sequence, the components whose contribution rate exceeds the preset threshold are selected for reconstruction. This preset threshold is usually set to a level that can retain 85%-95% of the original information. Through this dimensionality reduction, the complexity of the data can be significantly reduced while retaining the main information. For example, when processing large-scale measurement data of the watch surface, the original data may contain features of dozens of dimensions, but after dimensionality reduction, only 3-5 main dimensions may be needed to accurately characterize the material properties. The advantage of this dimensionality reduction method is that it can improve the efficiency of data processing while maintaining the accuracy of the characterization.

[0118] The final step is to perform feature mapping matching. This process compares and analyzes the reduced-dimensional feature data with the standard feature parameters of plasma resonance. The standard feature parameters include the resonance frequency , resonance bandwidth Δω and resonance intensity Q. Resonance frequency The resonance bandwidth Δω corresponds to the collective oscillation frequency of electrons in the material, and the resonance intensity Q represents the sharpness of the resonance. In the actual matching process, the measured data is first normalized to the standard parameter space, that is, the characteristic value of each measurement point is standardized: ω=ω / , Δω'=Δω / , Q'=Q / ,in and is the standard reference value. Then calculate the feature matching degree of each point M=exp[-((ω-1)² / +(Δω'-1)² / +(Q'-1)² / )],in , , is the tolerance parameter of the corresponding feature dimension. For example, when evaluating the performance of the nanofilm on the surface of a watch, if the normalized parameters of a certain area are ω=0.95, Δω'=1.1, Q'=0.9, the tolerance parameter is = = =0.1, then the feature matching degree M of the region can be calculated to be 0.86, indicating that the performance of the region has reached a good level. The advantage of this mapping matching method is that it provides an objective and quantitative performance evaluation standard, making the measurement results of different batches and different regions comparable.

[0119] The innovation of the entire feature analysis process is reflected in: adopting a multi-level feature extraction and dimensionality reduction strategy, achieving data dimensionality reduction and feature extraction through rigorous mathematical processing, and finally establishing a reliable performance evaluation system. This method can not only accurately reflect the performance characteristics of the material, but also effectively process large-scale measurement data, providing strong technical support for the quality control of nanofilm performance.

[0120] In one embodiment of the present invention, performing singular value decomposition on each component of the orthogonal eigenvector group to obtain feature weight distribution data includes: constructing the orthogonal eigenvector group into a feature matrix, performing singular value decomposition to obtain a left singular matrix, a singular value matrix and a right singular matrix; performing orthogonality analysis on the left singular matrix and the right singular matrix to obtain a basis vector group of the feature space; calculating the energy contribution coefficient of each feature component according to the size of the diagonal elements of the singular value matrix; performing hierarchical clustering analysis based on the energy contribution coefficient to obtain a feature hierarchical tree structure; performing pruning optimization on the feature hierarchical tree structure to obtain a key feature node sequence; recalculating the weight coefficient according to the key feature node sequence to generate feature weight distribution data.

[0121] Specifically, it is first necessary to construct the orthogonal eigenvector group into a feature matrix. This construction process actually arranges and combines the multiple independent eigenvectors obtained previously into a complete data matrix in columns. Performing singular value decomposition on this feature matrix means decomposing it into three independent matrices: the left singular matrix reflects the distribution characteristics of the features in the original data space, the singular value matrix contains the importance information of each feature component, and the right singular matrix reflects the relationship between the features. For example, when analyzing the nanofilm characteristics of a certain area on the surface of a watch, if the original data contains multiple parameters such as reflectivity, resonance frequency, peak width, etc., after these parameters are decomposed through the matrix, it can be clearly seen which parameter combinations are most important for the characterization of material properties. The advantage of this decomposition method is that it can fully reveal the structural characteristics implicit in the data.

[0122] The orthogonality analysis of the decomposed left and right singular matrices is to check whether the vectors in these matrices are perpendicular to each other and their lengths are normalized. This analysis ensures that the obtained feature space basis vector groups are independent of each other, and each basis vector reflects a unique dimension of the data. For example, when analyzing the multidimensional characteristics of a watch nanofilm, one basis vector may mainly reflect the resonance intensity characteristics of the material, while another basis vector may mainly reflect the change characteristics of the resonance frequency, and there is no information overlap between these basis vectors. The advantage of this orthogonality analysis is that it ensures the accuracy and independence of feature extraction.

[0123] Based on the size of the elements on the diagonal of the singular value matrix, the energy contribution coefficient of each characteristic component can be calculated. These coefficients actually reflect the proportion of each characteristic component in the overall characteristics. When calculating, the square of each singular value is divided by the sum of the squares of all singular values ​​to obtain the energy contribution ratio of the corresponding characteristic component. For example, when analyzing the characteristics of the nanofilm on the surface of a watch, it may be found that the first characteristic component contributes 70% of the information, the second contributes 20%, and the remaining characteristic components contribute 10% together. The advantage of this energy analysis method is that it can clearly identify the main features.

[0124] Next, a hierarchical cluster analysis is performed, which is a process of gradually classifying feature components with similar energy contributions. Specifically, each feature component is first regarded as an independent class, and then similar classes are gradually merged according to the similarity of their energy contribution coefficients, and finally a tree-like hierarchy is formed. For example, when analyzing the characteristics of watch nanofilms, three main levels are formed: the top layer contains those features that play a decisive role in material performance, the middle layer contains features with important influences, and the bottom layer contains features with relatively small influences. The advantage of this hierarchical clustering method is that it can clearly show the importance hierarchy between features.

[0125] Pruning the formed feature hierarchy tree means setting a threshold for energy contribution, retaining feature nodes whose cumulative energy contribution exceeds this threshold, and pruning other nodes. Usually this threshold is set at around 90%, which means that the retained feature nodes can express more than 90% of the information of the original data. For example, when analyzing the characteristics of watch nanofilms, if it is found that the cumulative contribution of the first three feature nodes has reached 92%, only these three key nodes can be retained, thereby greatly simplifying the data structure. The advantage of this pruning optimization is that it not only maintains the main information, but also reduces the complexity of data processing.

[0126] Finally, the weight coefficients are recalculated based on the sequence of retained key feature nodes. This calculation process divides the energy contribution of each retained node by the sum of the energy contributions of all retained nodes to obtain a new weight distribution. For example, if three feature nodes are retained, their weights in the final performance characterization may be 60%, 30%, and 10%, respectively. These weights directly reflect the importance of each feature in the material performance evaluation. The advantage of this weight distribution method is that it provides a simplified but accurate feature characterization system.

[0127] The innovation of the entire feature weight analysis process is reflected in: through rigorous mathematical analysis methods, feature screening and weight allocation are achieved, and a feature characterization system is established that maintains the essence of data while simplifying the calculation complexity. This method provides reliable technical support for the quantitative evaluation of nanomembrane performance.

[0128] Please continue reading Figure 1 , a spatial distribution relationship of regional performance parameters is established based on the performance evaluation data, and a quantitative detection result characterizing the intrinsic performance of the nanomembrane is generated.

[0129] In one embodiment of the present invention, the spatial distribution relationship of regional performance parameters is established according to the performance evaluation data to generate quantitative detection results characterizing the intrinsic performance of the nanomembrane, including: performing local response characteristic analysis on the performance evaluation data to obtain the plasma resonance intensity index, kinetic stability index and uniformity index of each region; combining the plasma resonance intensity index, kinetic stability index and uniformity index according to a preset weighting coefficient to obtain a regional comprehensive performance index; performing Delaunay triangulation on the test area according to the regional comprehensive performance index to obtain a spatial interpolation grid of performance parameters; performing thin plate spline interpolation on the node data of the spatial interpolation grid to obtain a continuous performance distribution function; calculating the local gradient and curvature based on the performance distribution function to generate a performance change characteristic diagram; fusing and superimposing the performance change characteristic diagram with the performance distribution function to obtain a quantitative detection result characterizing the intrinsic performance of the nanomembrane.

[0130] Specifically, the performance evaluation data first needs to be analyzed for local response characteristics. This analysis process focuses on three core indicators: the plasma resonance intensity index reflects the resonance ability of the material, that is, the response intensity of the material to the incident light; the kinetic stability index characterizes the stability of the material from the photobleaching state to the initial state; the uniformity index reflects the consistency level of performance distribution in the local area. For example, when analyzing the nanofilm in a certain area of ​​the watch surface, if the area shows a high resonance intensity (greater than 0.8, normalized value), good recovery stability (recovery time deviation is less than 5%) and excellent uniformity (local performance fluctuation is less than 3%), it means that the performance of the nanofilm in this area has reached a high level. The advantage of this multi-index analysis method is that it can fully reflect the performance characteristics of the material.

[0131] When these three indexes are weighted and combined, the weighting coefficients are set according to the actual application requirements of the nanofilm. Generally, the plasma resonance intensity has the largest weight (about 0.5) because it directly reflects the core function of the material; the kinetic stability is second (about 0.3) because it is related to the service life of the material; the uniformity index has the smallest weight (about 0.2), but it is equally important for product quality control. Through this weighted combination, the regional comprehensive performance index obtained can fully reflect the comprehensive performance level of the material. For example, when evaluating nanofilms in different areas of the watch surface, if the comprehensive performance index of a certain area reaches above 0.9, it means that the overall performance of the area is very excellent. The advantage of this weighted evaluation method is that the evaluation criteria can be flexibly adjusted according to actual application needs.

[0132] Based on the regional comprehensive performance index, the Delaunay triangulation method is used to mesh the entire test area. The characteristic of Delaunay triangulation is that the generated triangles are as close to equilateral triangles as possible, which can avoid the appearance of triangles that are too narrow and long and improve the accuracy of interpolation. During the triangulation process, points with similar performance indices will form denser triangular meshes, while areas with large performance changes will form looser meshes. For example, when analyzing the nanofilm in the transition area between the watch bezel and the watch glass, the triangular mesh obtained by triangulation will become denser accordingly because there may be a gradient change in performance here. The advantage of this adaptive meshing method is that it can automatically adjust the analysis accuracy according to the performance distribution characteristics.

[0133] The spatial interpolation grid obtained by the division is interpolated by thin plate spline in order to obtain a continuous performance distribution function. The performance distribution function F(x, y) here represents the performance index value at any position (x, y) on the watch surface. It consists of the following parts: the first part is the basic term B(x, y), which reflects the overall trend of the performance; the second part is the correction term C(x, y), which is used to compensate for local performance fluctuations; the third part is the smoothing term S(x, y), which ensures the continuous change of performance in space. Therefore, F(x, y) = B(x, y) + C(x, y) + S(x, y). Thin plate spline interpolation simulates the deformation characteristics of thin elastic plates, which can minimize the bending energy of the surface while ensuring the smoothness and continuity of the surface. This interpolation method is particularly suitable for processing performance distribution data with slowly changing characteristics. For example, when analyzing the overall performance distribution of the watch surface, a smooth performance distribution surface can be obtained by thin plate spline interpolation, which clearly shows the gradual characteristics of the performance. The advantage of this interpolation method is that it can accurately reflect the continuous change characteristics of the performance.

[0134] Based on the obtained continuous performance distribution function, the local gradient and curvature characteristics are calculated. The gradient reflects the direction and rate of performance change, while the curvature reflects the severity of the performance change. These characteristic parameters are of great significance for identifying performance mutation areas and defect locations. For example, when testing the uniformity of the nanofilm on the surface of a watch, if a large gradient value or an abnormal curvature value appears somewhere, it may mean that there is a performance defect at that location. The advantage of this characteristic analysis method is that it can effectively identify performance abnormality areas.

[0135] The last step is to fuse and superimpose the performance change characteristic map with the performance distribution function. This process actually superimposes the local change characteristics on the overall distribution to form a comprehensive performance characterization result. This superposition adopts the feature enhancement method to modify the original performance distribution function F(x, y) through the modulation function T(x, y), that is, the final performance characterization result R(x, y) = F(x, y) × T(x, y), where the modulation function T(x, y) is dynamically adjusted according to the local gradient and curvature characteristics. For example, when generating the final inspection report of the nanofilm on the surface of a watch, the overall distribution of performance can be clearly displayed, while highlighting the areas with abnormal performance or significant gradient changes. The advantage of this feature fusion method is that it retains the overall performance distribution information and highlights the important local features.

[0136] The innovation of the entire performance evaluation process is reflected in: using a multi-dimensional index system to characterize performance, achieving continuous expression of performance through adaptive meshing and high-precision interpolation, and finally establishing a comprehensive performance evaluation system through feature fusion. This method can not only accurately evaluate material performance, but also effectively identify performance abnormality areas, providing reliable technical support for product quality control.

[0137] The above describes the detection method of the plasma nanofilm of a watch based on image processing in the embodiment of the present invention. The following describes the detection device of the plasma nanofilm of a watch based on image processing in the embodiment of the present invention. Figure 2 In one embodiment of the present invention, a detection device for a plasma nanofilm of a watch based on image processing comprises:

[0138] The data acquisition module 201 is used to collect dual-polarization light intensity distribution data of the plasma nanofilm of the watch before and after photobleaching within a preset incident angle range, wherein the dual-polarization light intensity distribution data includes s-polarization and p-polarization intensity data before photobleaching and s-polarization and p-polarization intensity data after photobleaching;

[0139] The phase modulation module 202 is used to perform phase modulation on the dual polarization light intensity distribution data, record a phase-intensity response curve within a preset phase range, and obtain amplitude-phase characteristic data characterizing the local plasma resonance characteristics through Fourier transform processing;

[0140] A dynamic feature extraction module 203 is used to extract the dynamic recovery feature of the photobleaching area based on the dual-polarization light intensity distribution data and the amplitude-phase feature data, and generate a regional segmentation feature spectrum according to the spatial distribution law of the dynamic recovery feature;

[0141] The decoupling analysis module 204 is used to select the maximum photobleaching depth point in each region as a reference benchmark according to the regional segmentation characteristic map, and perform differential processing on the optical response of other points in the region and the reference benchmark to obtain pure material effect distribution data that eliminates the influence of geometric curvature;

[0142] A dimensionality reduction processing module 205 is used to perform dimensionality reduction processing on the characteristic parameters in the pure material effect distribution data, extract the main characteristic components that characterize the plasma resonance characteristics, and obtain performance evaluation data;

[0143] The performance evaluation module 206 is used to establish the spatial distribution relationship of the regional performance parameters according to the performance evaluation data, and generate quantitative detection results that characterize the intrinsic performance of the nanofilm.

[0144] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the inventive concept of the present invention, or directly / indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A detection method for plasma nanofilm of watches based on image processing, characterized in that: include: Collecting dual-polarization light intensity distribution data of the plasma nanofilm of the watch before and after photobleaching within a preset incident angle range, wherein the dual-polarization light intensity distribution data includes s-polarization and p-polarization intensity data before photobleaching and s-polarization and p-polarization intensity data after photobleaching; The method collects dual polarization state light intensity distribution data of a watch plasma nanofilm before and after photobleaching within a preset incident angle range, comprising: forming a photobleaching area array with a preset spacing on a sample surface, wherein the photobleaching area array comprises a plurality of independent photobleaching white spots; performing an angle scan of 0-75 degrees on the photobleaching area array, collecting a set of polarization state light intensity data every 5 degrees, and obtaining original light intensity data before photobleaching; performing photobleaching excitation on the photobleaching area array, scanning in the same angle stepping manner as the 0-75 degree angle scan, collecting polarization state light intensity data of the photobleaching area array, and obtaining excited state light intensity after photobleaching. data; performing polarization state separation according to the original light intensity data and the excited state light intensity data to obtain s polarization intensity data and p polarization intensity data before photobleaching; performing polarization state separation according to the excited state light intensity data to obtain s polarization intensity data and p polarization intensity data after photobleaching; combining the s polarization and p polarization intensity data before photobleaching and the s polarization and p polarization intensity data after photobleaching into dual polarization state light intensity distribution data; the photobleaching excitation refers to irradiating the material with a higher intensity laser to put it into a photobleaching state, in which the plasma resonance effect of the material is significantly suppressed, while the geometric characteristics of the surface remain unchanged; Phase modulating the dual-polarization light intensity distribution data, recording a phase-intensity response curve within a preset phase range, and obtaining amplitude-phase characteristic data characterizing local plasma resonance characteristics through Fourier transform processing; Extracting the dynamic recovery characteristics of the photobleaching area based on the dual-polarization light intensity distribution data and the amplitude-phase characteristic data, and generating a regional segmentation feature map according to the spatial distribution law of the dynamic recovery characteristics; According to the regional segmentation characteristic map, the maximum point of the photobleaching depth in each region is selected as a reference benchmark, and the optical responses of other points in the region are differentially processed with the reference benchmark to obtain the pure material effect distribution data with the geometric curvature effect eliminated; According to the regional segmentation characteristic map, the maximum point of the photobleaching depth in each region is selected as a reference benchmark, and the optical responses of other points in the region are differentially processed with the reference benchmark to obtain the pure material effect distribution data with the geometric curvature effect eliminated, including: performing light intensity gradient calculation on each region in the regional segmentation characteristic map to obtain the photobleaching depth in the region. degree distribution data; performing extreme point detection on each region according to the photobleaching depth distribution data, and screening a candidate point set with the largest photobleaching depth; performing local curvature calculation on each point in the candidate point set to obtain curvature feature data of the candidate point; optimizing the candidate point set according to the curvature feature data, and determining the final reference reference point in each region; performing ratio calculation on the light intensity data of other points in the region and the light intensity data of the reference reference point to obtain normalized light intensity distribution data; performing curvature compensation processing on the normalized light intensity distribution data to generate pure material effect distribution data that eliminates the influence of geometric curvature; Performing dimensionality reduction processing on characteristic parameters in the pure material effect distribution data, extracting main characteristic components characterizing plasma resonance characteristics, and obtaining performance evaluation data; The spatial distribution relationship of the regional performance parameters is established based on the performance evaluation data, and the quantitative detection results characterizing the intrinsic performance of the nanofilm are generated.

2. The detection method of watch plasma nanofilm based on image processing according to claim 1, characterized in that: The phase modulation is performed on the dual polarization light intensity distribution data, a phase-intensity response curve is recorded within a preset phase range, and amplitude-phase characteristic data characterizing the local plasma resonance characteristics is obtained through Fourier transform processing, including: Performing phase scanning on the dual polarization light intensity distribution data in the range of 0-2π with a step length of 0.1π to obtain a phase scanning sequence before and after photobleaching; Calculating local polarization state ellipse parameters according to the s-polarization and p-polarization intensity data of each phase point in the phase scanning sequence to obtain phase-intensity modulation data; Performing discrete Fourier transform on the phase-intensity modulation data, extracting first-order and second-order Fourier components, and obtaining initial amplitude-phase characteristic quantities; Calculate the local plasma resonance intensity coefficient and the phase delay coefficient according to the initial amplitude-phase characteristic quantity to obtain the resonance characteristic parameters; Normalizing the resonance characteristic parameters to generate standardized amplitude-phase characteristic data; The standardized amplitude-phase characteristic data is correlated and matched with the phase-intensity modulation data of the corresponding position to obtain the amplitude-phase characteristic data characterizing the local plasma resonance characteristics.

3. The detection method of watch plasma nanofilm based on image processing according to claim 1, characterized in that: The method of extracting the dynamic recovery characteristics of the photobleaching region based on the dual-polarization light intensity distribution data and the amplitude-phase characteristic data, and generating a regional segmentation feature map according to the spatial distribution law of the dynamic recovery characteristics, comprises: Continuously sampling the dual-polarization light intensity distribution data at preset time intervals to obtain time-series intensity variation data of the photobleaching region; Extract the recovery time constant and recovery amplitude parameter of each sampling point according to the time series intensity variation data to obtain initial dynamic characteristic data; Performing cross-correlation analysis on the amplitude-phase characteristic data and the initial dynamic characteristic data to obtain comprehensive dynamic parameters characterizing the response characteristics of the material; Performing spatial gradient analysis on the comprehensive dynamic parameters to obtain a dynamic characteristic difference distribution map of a local area; According to the spatial continuity of the dynamic feature difference distribution map, adjacent regions are dynamically clustered to obtain a set of regional boundary feature points; A closed curve is fitted to the region boundary feature point set to generate a region segmentation feature map.

4. The detection method of watch plasma nanofilm based on image processing according to claim 1, characterized in that: The dimensionality reduction processing is performed on the characteristic parameters in the pure material effect distribution data to extract the main characteristic components characterizing the plasma resonance characteristics to obtain the performance evaluation data, including: Performing nonlinear transformation on the resonance peak position, peak width and peak intensity in the pure material effect distribution data to obtain a feature point set in a multidimensional feature space; Performing principal component-extreme value decomposition according to the feature point set, extracting the change trend in the main feature direction, and obtaining an orthogonal feature vector group; Performing singular value decomposition on each component of the orthogonal eigenvector group to obtain feature weight distribution data; Reordering the feature weight distribution data according to weights to obtain a sorted feature component sequence; Reconstructing the feature components whose contribution rates in the feature component sequence exceed a preset threshold value to obtain feature data with reduced dimensions; The feature data after dimension reduction is mapped and matched with the plasma resonance theory features to generate performance evaluation data.

5. The detection method of watch plasma nanofilm based on image processing according to claim 4 is characterized in that: The step of performing singular value decomposition on each component of the orthogonal eigenvector group to obtain feature weight distribution data includes: The orthogonal eigenvector group is constructed into a characteristic matrix, and singular value decomposition is performed to obtain a left singular matrix, a singular value matrix and a right singular matrix; Performing orthogonality analysis on the left singular matrix and the right singular matrix to obtain a basis vector group of the feature space; Calculate the energy contribution coefficient of each characteristic component according to the size of the diagonal elements of the singular value matrix; Performing hierarchical clustering analysis based on the energy contribution coefficient to obtain a feature hierarchical tree structure; Pruning and optimizing the feature hierarchical tree structure to obtain a key feature node sequence; The weight coefficients are recalculated according to the key feature node sequence to generate feature weight distribution data.

6. The detection method of watch plasma nanofilm based on image processing according to claim 1, characterized in that: The method of establishing the spatial distribution relationship of the regional performance parameters according to the performance evaluation data and generating the quantitative detection results characterizing the intrinsic performance of the nanofilm includes: Performing local response characteristic analysis on the performance evaluation data to obtain a plasma resonance intensity index, a kinetic stability index, and a uniformity index of each region; The plasma resonance intensity index, the kinetic stability index and the uniformity index are combined according to a preset weighting coefficient to obtain a regional comprehensive performance index; Performing Delaunay triangulation on the test area according to the comprehensive performance index of the area to obtain a spatial interpolation grid of performance parameters; Performing thin plate spline interpolation on the node data of the spatial interpolation grid to obtain a continuous performance distribution function; Calculating local gradients and curvatures based on the performance distribution function to generate a performance change characteristic graph; The performance change characteristic diagram is fused and superimposed with the performance distribution function to obtain a quantitative detection result characterizing the intrinsic performance of the nanofilm.

7. A detection device for plasma nanofilm of watches based on image processing, characterized in that: The detection device for plasma nanofilm of a watch based on image processing is applied to the detection method for plasma nanofilm of a watch based on image processing as claimed in any one of claims 1 to 6, and the detection device for plasma nanofilm of a watch based on image processing comprises: A data acquisition module, used to collect dual-polarization light intensity distribution data of the plasma nanofilm of the watch before and after photobleaching within a preset incident angle range, wherein the dual-polarization light intensity distribution data includes s-polarization and p-polarization intensity data before photobleaching and s-polarization and p-polarization intensity data after photobleaching; A phase modulation module is used to phase modulate the dual-polarization light intensity distribution data, record a phase-intensity response curve within a preset phase range, and obtain amplitude-phase characteristic data characterizing the local plasma resonance characteristics through Fourier transform processing; A dynamic feature extraction module, used to extract the dynamic recovery feature of the photobleaching area based on the dual-polarization light intensity distribution data and the amplitude-phase feature data, and generate a regional segmentation feature map according to the spatial distribution law of the dynamic recovery feature; A decoupling analysis module is used to select the maximum photobleaching depth point in each region as a reference benchmark according to the regional segmentation characteristic map, and perform differential processing on the optical response of other points in the region and the reference benchmark to obtain pure material effect distribution data that eliminates the influence of geometric curvature; A dimensionality reduction processing module is used to perform dimensionality reduction processing on the characteristic parameters in the pure material effect distribution data, extract the main characteristic components that characterize the plasma resonance characteristics, and obtain performance evaluation data; The performance evaluation module is used to establish the spatial distribution relationship of regional performance parameters according to the performance evaluation data, and generate quantitative detection results characterizing the intrinsic performance of the nanofilm.

Images