Artificial intelligence-based infrared spectroscopic image measurement alignment method and device

Abstract

The application discloses an infrared spectrum image measurement alignment method and device based on artificial intelligence, and relates to the technical field of image measurement.The technical solution points of the application comprise the following steps: obtaining infrared spectrum image data of a to-be-measured object, processing the infrared spectrum image data to obtain a to-be-aligned image area of the infrared spectrum image, and setting an image measurement reference point according to the to-be-aligned image area; processing the to-be-aligned image area according to the image measurement reference point to obtain a feature matching deviation coefficient, a gray distribution deviation coefficient, a feature abnormal reference point and a gray abnormal reference point; and forming a measurement abnormal reference point of the infrared spectrum image according to the feature abnormal reference point and the gray abnormal reference point.The effect is that the measurement error caused by the alignment deviation of the reference point is effectively reduced, and the measurement result based on the infrared spectrum image is more reliable.

Description

Technical Field

[0001] This invention relates to the field of image measurement technology, and more specifically, to an artificial intelligence-based infrared spectral image measurement alignment method and apparatus. Background Technology

[0002] Infrared spectral image measurement alignment is a core step in ensuring accurate and reliable measurement results. Traditional measurement alignment methods lack detailed and comprehensive analysis of this complex information, easily overlooking key deviations and leading to inaccurate reference point positioning. For example, focusing only on a single feature or grayscale dimension fails to comprehensively consider the overall deviation of the image, resulting in errors in subsequent measurement results. Furthermore, the correction process lacks priority determination, often handling abnormal reference points based on simple rules, and cannot dynamically adjust based on factors such as the degree of deviation and actual distance. This results in low correction efficiency and unreasonable resource allocation, thus affecting the alignment effect. Summary of the Invention

[0003] In view of the shortcomings of the existing technology, the purpose of this invention is to provide an infrared spectral image measurement alignment method and apparatus based on artificial intelligence.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] An artificial intelligence-based infrared spectral image measurement alignment method, comprising the following steps:

[0006] Acquire the infrared spectral image data of the object to be measured, process the infrared spectral image data to obtain the image region to be aligned in the infrared spectral image, and set the image measurement reference point based on the image region to be aligned.

[0007] Based on the image measurement reference points, the region of the image to be aligned is processed to obtain the feature matching deviation coefficient, gray-level distribution deviation coefficient, feature anomaly reference points, and gray-level anomaly reference points;

[0008] Measurement anomaly reference points for infrared spectral images are constructed based on feature anomaly reference points and grayscale anomaly reference points; the alignment deviation of the measurement anomaly reference points is obtained based on preset feature deviation weights and feature matching deviation coefficients, and preset grayscale deviation weights and grayscale distribution deviation coefficients.

[0009] A reference point correction intensity is generated based on the alignment deviation, and a reference point correction signal for the image measurement reference point is generated based on the reference point correction intensity. The signal feedback distance of the measurement anomaly reference point is obtained based on the measurement anomaly reference point and the preset alignment correction point.

[0010] The correction priority of the abnormal reference point is obtained based on the reference point correction intensity and the signal feedback distance; the alignment correction point is used to align and correct the abnormal reference point according to the correction priority and the reference point correction signal.

[0011] Preferably, the feature matching deviation coefficient, gray-level distribution deviation coefficient, feature anomaly reference point, and gray-level anomaly reference point are obtained by processing the image region to be aligned based on the image measurement reference point, specifically including the following steps:

[0012] Based on the image measurement reference point, image feature matching data of the image region to be aligned is obtained. The image feature matching data is analyzed to obtain the feature matching deviation coefficient of the image region to be aligned. Based on the feature matching deviation coefficient, the image measurement reference point corresponding to the image region to be aligned is recorded as the feature anomaly reference point.

[0013] Based on the image measurement reference point, obtain the image grayscale distribution data of the image area to be aligned, analyze the image grayscale distribution data to obtain the grayscale distribution deviation coefficient of the image area to be aligned, and record the image measurement reference point corresponding to the image area to be aligned as the grayscale anomaly reference point.

[0014] Preferably, the infrared spectral image data is processed to obtain the image region to be aligned in the infrared spectral image, specifically including the following steps:

[0015] The infrared spectral image data includes the image resolution and image size of the object to be measured;

[0016] Set a standard reference spacing, and divide the image size into regions based on the standard reference spacing to obtain the image region to be aligned in the infrared spectrum image.

[0017] Preferably, the feature matching deviation coefficient of the image region to be aligned is obtained by analyzing the image feature matching data, specifically including the following steps:

[0018] The image feature matching data includes contour feature data and texture feature data of the image region to be aligned;

[0019] The contour matching deviation coefficient of the image region to be aligned is obtained by analyzing the contour feature data.

[0020] The texture matching deviation coefficient of the image region to be aligned is obtained by analyzing the texture feature data.

[0021] Set contour deviation weights and texture deviation weights, and obtain the feature matching deviation coefficients of the image region to be aligned based on the contour deviation weights and contour matching deviation coefficients, and the texture deviation weights and texture matching deviation coefficients.

[0022] Preferably, the contour matching deviation coefficient of the image region to be aligned is obtained by analyzing the contour feature data, specifically including the following steps:

[0023] The contour feature data packet is the contour coordinate value of the aligned image region;

[0024] Generate a contour matching curve corresponding to the image region to be aligned based on the contour coordinate values.

[0025] If the contour coordinate deviation value in the contour matching curve is greater than the preset contour coordinate threshold, the first contour deviation coefficient of the image region to be aligned is obtained based on the contour coordinate deviation value and the contour coordinate threshold.

[0026] Obtain the contour curve slope of the contour matching curve. If the contour curve slope is greater than the preset contour curve slope threshold, then obtain the second contour deviation coefficient of the image region to be aligned based on the contour curve slope and the contour curve slope threshold.

[0027] The contour matching deviation coefficient of the image region to be aligned is obtained based on the first contour deviation coefficient and the second contour deviation coefficient, and the image measurement reference point corresponding to the image region to be aligned is recorded as the feature anomaly reference point.

[0028] Preferably, the texture matching deviation coefficient of the image region to be aligned is obtained by analyzing the texture feature data, specifically including the following steps:

[0029] The texture feature data includes the texture density value of the image region to be aligned, and a texture matching curve corresponding to the image region to be aligned is generated based on the texture density value.

[0030] If the texture density deviation value in the texture matching curve is greater than the preset texture density threshold, the first texture deviation coefficient of the image region to be aligned is obtained based on the texture density deviation value and the texture density threshold.

[0031] Obtain the slope of the texture matching curve; if the slope of the texture curve is greater than a preset texture curve slope threshold, then obtain the second texture deviation coefficient of the image region to be aligned based on the texture curve slope and the texture curve slope threshold.

[0032] The texture matching deviation coefficient of the image region to be aligned is obtained based on the first texture deviation coefficient and the second texture deviation coefficient.

[0033] Preferably, the grayscale distribution data of the image is analyzed to obtain the grayscale distribution deviation coefficient of the image region to be aligned, specifically including the following steps:

[0034] The image grayscale distribution data includes the grayscale values ​​of the image region to be aligned;

[0035] Generate a grayscale distribution curve for the image region to be aligned based on the grayscale values, and obtain the slope of the grayscale curve.

[0036] If the slope of the grayscale curve is greater than the preset grayscale curve slope threshold, then the grayscale distribution deviation coefficient of the image region to be aligned is obtained based on the grayscale curve slope and the grayscale curve slope threshold.

[0037] Preferably, obtaining the signal feedback distance of the measurement anomaly reference point based on the measurement anomaly reference point and the preset alignment correction point specifically includes the following steps:

[0038] Acquire the first moment when the reference point transmits the reference point correction signal for the measurement anomaly reference point, and acquire the second moment when the alignment correction point receives the reference point correction signal;

[0039] The signal feedback duration is obtained based on the first and second time points;

[0040] The signal transmission medium for acquiring infrared spectral images, and the signal feedback speed for obtaining reference point correction signals based on the signal transmission medium;

[0041] The signal feedback distance of the measurement anomaly reference point is obtained based on the signal feedback duration and signal feedback speed.

[0042] Preferably, the correction priority of the measurement anomaly reference point is obtained based on the reference point correction intensity and signal feedback distance, specifically including the following steps:

[0043] Set the correction intensity weight and feedback distance weight;

[0044] The benchmark correction coefficient for measuring abnormal benchmark points is obtained based on the benchmark correction strength and the correction strength weight.

[0045] The reference point distance coefficient for measuring the anomaly reference point is obtained based on the signal feedback distance and the feedback distance weight.

[0046] The alignment correction coefficient of the measurement abnormal reference point is obtained based on the reference point correction coefficient and the reference point distance coefficient;

[0047] The correction priority of the measurement abnormal reference point is set according to the alignment correction coefficient.

[0048] An artificial intelligence-based infrared spectral image measurement alignment device includes:

[0049] Acquisition module: Acquires infrared spectral image data of the object to be measured, processes the infrared spectral image data to obtain the image region to be aligned in the infrared spectral image, and sets the image measurement reference point based on the image region to be aligned.

[0050] The first processing module: Based on the image measurement reference points, it processes the image region to be aligned to obtain the feature matching deviation coefficient, gray-level distribution deviation coefficient, feature anomaly reference points, and gray-level anomaly reference points;

[0051] The second processing module: constructs measurement anomaly reference points for the infrared spectral image based on feature anomaly reference points and grayscale anomaly reference points; and obtains the alignment deviation of the measurement anomaly reference points based on preset feature deviation weights and feature matching deviation coefficients, and preset grayscale deviation weights and grayscale distribution deviation coefficients.

[0052] The third processing module generates a reference point correction intensity for the measurement anomaly reference point based on the alignment deviation, generates a reference point correction signal for the image measurement reference point based on the reference point correction intensity, and obtains the signal feedback distance of the measurement anomaly reference point based on the measurement anomaly reference point and the preset alignment correction point.

[0053] The positioning module obtains the correction priority of the abnormal reference point based on the reference point correction intensity and signal feedback distance; the alignment and correction point performs alignment, correction and positioning of the abnormal reference point based on the correction priority and the reference point correction signal.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] This invention first divides the infrared spectral image data into alignment regions and establishes measurement reference points. It then conducts in-depth analysis from the dimensions of feature matching and grayscale distribution. By using feature matching deviation coefficients to identify feature anomaly reference points, and by using grayscale distribution deviation coefficients to find grayscale anomaly reference points, it comprehensively captures anomalies in both features and grayscale levels within the image, improving the comprehensiveness and accuracy of anomaly identification before alignment. The alignment deviation is calculated by combining preset features, grayscale deviation weights, and corresponding deviation coefficients. This quantitative method extends anomaly identification from qualitative to quantitative measurement, clearly distinguishing the magnitude of deviations and providing a precise basis for subsequent correction intensity settings and priority determination, making the alignment operation more closely reflect actual deviation conditions. This application effectively reduces measurement errors caused by reference point alignment deviations, making measurement results based on infrared spectral images more reliable. Attached Figure Description

[0056] Figure 1 This is a schematic diagram illustrating the steps of the infrared spectral image measurement alignment method based on artificial intelligence proposed in this invention;

[0057] Figure 2 This is a schematic diagram illustrating the steps of obtaining the feature matching deviation coefficient in the artificial intelligence-based infrared spectral image measurement alignment method proposed in this invention;

[0058] Figure 3 This is a schematic diagram of the module of the infrared spectral image measurement alignment device based on artificial intelligence proposed in this invention. Detailed Implementation

[0059] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0060] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0061] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0062] Reference Figures 1-3 As shown.

[0063] The embodiments further illustrate the infrared spectral image measurement alignment method and apparatus based on artificial intelligence proposed in this invention.

[0064] An artificial intelligence-based infrared spectral image measurement alignment method, comprising the following steps:

[0065] Acquire the infrared spectral image data of the object to be measured, process the infrared spectral image data to obtain the image region to be aligned in the infrared spectral image, and set the image measurement reference point based on the image region to be aligned.

[0066] Based on the image measurement reference points, the region of the image to be aligned is processed to obtain the feature matching deviation coefficient, gray-level distribution deviation coefficient, feature anomaly reference points, and gray-level anomaly reference points;

[0067] Measurement anomaly reference points for infrared spectral images are constructed based on feature anomaly reference points and grayscale anomaly reference points; the alignment deviation of the measurement anomaly reference points is obtained based on preset feature deviation weights and feature matching deviation coefficients, and preset grayscale deviation weights and grayscale distribution deviation coefficients.

[0068] The system generates a reference point correction intensity for the measurement anomaly reference point based on the alignment deviation, and generates a reference point correction signal for the image measurement reference point based on the reference point correction intensity. The system obtains the signal feedback distance of the measurement anomaly reference point based on the measurement anomaly reference point and the preset alignment correction point.

[0069] The correction priority of the abnormal reference point is obtained based on the reference point correction intensity and signal feedback distance; the alignment and correction point is aligned and corrected to locate the abnormal reference point based on the correction priority and the reference point correction signal.

[0070] This application first acquires infrared spectral image data of the object to be measured, including image resolution and size. The image size is then divided into regions according to a pre-set standard reference spacing to obtain the image regions to be aligned. Image measurement reference points are set within each image region to be aligned.

[0071] Each image region to be aligned is processed based on the image measurement reference point. Image feature matching data for that region is acquired, including contour feature data and texture feature data. For contour features, a corresponding contour matching curve is generated based on the contour coordinate values. By comparing the contour coordinate deviation value in the curve with a preset contour coordinate threshold, and the contour curve slope with a preset contour curve slope threshold, a first contour deviation coefficient and a second contour deviation coefficient are calculated respectively. These two coefficients are combined to obtain the contour matching deviation coefficient. If this coefficient exceeds a reasonable range, the corresponding image measurement reference point is marked as a feature anomaly reference point. A texture matching curve is generated based on the texture density value. The first texture deviation coefficient and the second texture deviation coefficient are calculated by comparing the texture density deviation value and the texture curve slope with their respective preset thresholds, thus obtaining the texture matching deviation coefficient. Finally, the feature matching deviation coefficient for the entire image region to be aligned is calculated by combining the contour deviation weight and the texture deviation weight. The gray values ​​of the image region to be aligned are obtained in the gray-scale distribution deviation analysis dimension, thereby generating a gray-scale distribution curve. By judging the relationship between the slope of the gray-scale curve and the preset gray-scale curve slope threshold, if the slope exceeds the threshold, the gray-scale distribution deviation coefficient is calculated based on the difference between the two, and the corresponding reference point is marked as a gray-scale abnormal reference point.

[0072] The feature anomaly reference points and grayscale anomaly reference points are integrated to form the measurement anomaly reference points of the infrared spectral image. In order to quantify the deviation degree of each measurement anomaly reference point, preset feature deviation weights and grayscale deviation weights are introduced. The feature matching deviation coefficient is combined with the feature deviation weights, and the grayscale distribution deviation coefficient is combined with the grayscale deviation weights. The alignment deviation degree of each measurement anomaly reference point is obtained by weighted calculation. The deviation degree value reflects the degree to which the reference point needs to be corrected.

[0073] Subsequently, a correction signal is generated and the feedback distance is calculated. Based on the magnitude of the alignment deviation, a corresponding reference point correction intensity is generated for the measurement anomaly reference point; the greater the deviation, the higher the correction intensity. A specific reference point correction signal is generated based on the correction intensity. The time from transmission to reception of the reference point correction signal is recorded using the measurement anomaly reference point and a preset alignment correction point; this is the signal feedback time. The signal feedback distance to the measurement anomaly reference point is calculated by multiplying the time by the speed; this distance reflects the spatial span of the signal transmission.

[0074] Set the correction intensity weight and feedback distance weight. Multiply the correction intensity by the correction intensity weight to obtain the reference point correction coefficient for the abnormal reference point; multiply the feedback distance by the feedback distance weight to obtain the reference point distance coefficient. Add these two coefficients to obtain the alignment correction coefficient. Set the correction priority for the abnormal reference points based on the magnitude of the alignment correction coefficient; the higher the coefficient, the higher the priority. Alignment correction points are then used to align and correct the abnormal reference points one by one according to the correction priority and the reference point correction signal, ensuring that the infrared spectral image measurement reference points are in accurate positions.

[0075] Based on the image measurement reference points, the region of the image to be aligned is processed to obtain the feature matching deviation coefficient, gray-level distribution deviation coefficient, feature anomaly reference points, and gray-level anomaly reference points. Specifically, this includes the following steps:

[0076] Based on the image measurement reference point, image feature matching data of the image region to be aligned is obtained. The image feature matching data is analyzed to obtain the feature matching deviation coefficient of the image region to be aligned. Based on the feature matching deviation coefficient, the image measurement reference point corresponding to the image region to be aligned is recorded as the feature anomaly reference point.

[0077] Based on the image measurement reference point, obtain the image grayscale distribution data of the image area to be aligned, analyze the image grayscale distribution data to obtain the grayscale distribution deviation coefficient of the image area to be aligned, and record the image measurement reference point corresponding to the image area to be aligned as the grayscale anomaly reference point.

[0078] Image feature matching data includes the contour and texture information of the region. For example, by comparing the features of the region to be aligned with the ideal state using a standard feature model or a normal feature pattern learned by an artificial intelligence algorithm, the difference between the features of the region to be aligned and the ideal state is determined, and a feature matching deviation coefficient is obtained. If this coefficient exceeds a reasonable range, it indicates that the features corresponding to the image measurement reference point of that region are abnormal, and this reference point is marked as a feature aberration reference point.

[0079] Based on image measurement reference points, image grayscale distribution data of the image area to be aligned is obtained, i.e., the grayscale values ​​at different locations within that area. This is used to determine whether the grayscale is uniformly distributed or conforms to normal grayscale variation patterns, thereby calculating the grayscale distribution deviation coefficient. When an abnormal grayscale distribution deviation coefficient is determined, the corresponding image measurement reference point is marked as a grayscale anomaly reference point.

[0080] The infrared spectral image data is processed to obtain the image region to be aligned, specifically including the following steps:

[0081] Infrared spectral image data includes the image resolution and image size of the object to be measured;

[0082] Set a standard reference spacing, and divide the image size into regions based on the standard reference spacing to obtain the image region to be aligned in the infrared spectrum image.

[0083] In infrared spectral image data, image resolution determines the fineness of image detail, while image size relates to the overall spatial range. A standard reference spacing is set to systematically segment the image. Once the standard reference spacing is established, the image size is divided into regions according to this spacing to obtain the image area to be aligned in the infrared spectral image.

[0084] The feature matching deviation coefficient of the image region to be aligned is obtained by analyzing the image feature matching data, specifically including the following steps:

[0085] Image feature matching data includes contour feature data and texture feature data of the image region to be aligned;

[0086] The contour matching deviation coefficient of the image region to be aligned is obtained by analyzing the contour feature data.

[0087] The texture matching deviation coefficient of the image region to be aligned is obtained by analyzing the texture feature data.

[0088] Set contour deviation weights and texture deviation weights, and obtain the feature matching deviation coefficients of the image region to be aligned based on the contour deviation weights and contour matching deviation coefficients, and the texture deviation weights and texture matching deviation coefficients.

[0089] Image feature matching data includes contour feature data and texture feature data of the image region to be aligned. Contour feature data outlines the shape and contour of the region, such as edge direction and shape structure; texture feature data presents the texture pattern of the region's surface.

[0090] For contour feature data, the contour of the region to be aligned is compared with the standard contour. The differences in shape, position, and curvature between the two are calculated to obtain the contour matching deviation coefficient, which measures the degree of matching of contour features. For texture feature data, the differences between the texture of the region to be aligned and the standard texture are compared in terms of texture density, direction, and periodicity to calculate the texture matching deviation coefficient, which reflects the matching status of texture features.

[0091] To comprehensively consider the impact of contour and texture on overall feature matching, contour deviation weights and texture deviation weights are set. The contour deviation weights are multiplied by the contour matching deviation coefficient, and the texture deviation weights are multiplied by the texture matching deviation coefficient. The two results are then added together to obtain the feature matching deviation coefficient of the image region to be aligned.

[0092] The contour matching deviation coefficient of the image region to be aligned is obtained by analyzing the contour feature data, specifically including the following steps:

[0093] The contour feature data package contains the contour coordinates of the aligned image region;

[0094] Generate a contour matching curve corresponding to the image region to be aligned based on the contour coordinate values.

[0095] If the contour coordinate deviation value in the contour matching curve is greater than the preset contour coordinate threshold, the first contour deviation coefficient of the image region to be aligned is obtained based on the contour coordinate deviation value and the contour coordinate threshold.

[0096] Obtain the contour curve slope of the contour matching curve. If the contour curve slope is greater than the preset contour curve slope threshold, then obtain the second contour deviation coefficient of the image region to be aligned based on the contour curve slope and the contour curve slope threshold.

[0097] The contour matching deviation coefficient of the image region to be aligned is obtained based on the first contour deviation coefficient and the second contour deviation coefficient, and the image measurement reference point corresponding to the image region to be aligned is recorded as the feature anomaly reference point.

[0098] The contour feature data package consists of contour coordinate values ​​of the image region to be aligned. These contour coordinate values ​​record the positional information of each point on the contour. Based on these coordinate values, a contour matching curve corresponding to the image region to be aligned is generated through mathematical fitting, transforming discrete coordinate points into a continuous curve form.

[0099] The contour coordinate deviation value in the contour matching curve is calculated, which is the difference between the actual contour coordinates and the ideal or standard contour coordinates. When this deviation value is greater than a preset contour coordinate threshold, it indicates that the contour has a significant deviation in position. At this time, the first contour deviation coefficient of the image area to be aligned is calculated based on the difference between the contour coordinate deviation value and the contour coordinate threshold. The first contour deviation coefficient quantifies the degree of contour position deviation. The slope of the contour curve of the contour matching curve is obtained, and the slope reflects the changing trend of the contour curve. If the slope of the contour curve is greater than a preset contour curve slope threshold, it means that the shape of the contour exceeds the normal range. Similarly, the second contour deviation coefficient of the image area to be aligned is calculated based on the comparison between the contour curve slope and the contour curve slope threshold. The second contour deviation coefficient measures the degree of contour shape deviation.

[0100] The contour matching deviation coefficient of the image region to be aligned is obtained by simply adding the first contour deviation coefficient and the second contour deviation coefficient. If the contour matching deviation coefficient exceeds a reasonable range, it indicates that there is a feature anomaly at the image measurement reference point corresponding to the image region to be aligned, and this reference point is recorded as the feature anomaly reference point.

[0101] The texture matching deviation coefficient of the image region to be aligned is obtained by analyzing the texture feature data, specifically including the following steps:

[0102] Texture feature data includes the texture density value of the image region to be aligned, and a texture matching curve corresponding to the image region to be aligned is generated based on the texture density value.

[0103] If the texture density deviation value in the texture matching curve is greater than the preset texture density threshold, the first texture deviation coefficient of the image region to be aligned is obtained based on the texture density deviation value and the texture density threshold.

[0104] Obtain the slope of the texture matching curve; if the slope of the texture curve is greater than the preset slope threshold, then obtain the second texture deviation coefficient of the image region to be aligned based on the slope of the texture curve and the slope threshold.

[0105] The texture matching deviation coefficient of the image region to be aligned is obtained based on the first texture deviation coefficient and the second texture deviation coefficient.

[0106] First, it's important to clarify that the core of texture feature data is the texture density value of the image region to be aligned, reflecting the density of the texture in that region, such as whether the texture is sparse or densely distributed. Based on the texture density value, a texture matching curve corresponding to the image region to be aligned is generated through data fitting.

[0107] The texture density deviation value in the texture matching curve is calculated, which is the difference between the actual texture density and the ideal or standard texture density. When this deviation value is greater than a preset texture density threshold, it indicates a significant abnormality in texture density. In this case, the first texture deviation coefficient for the image region to be aligned is calculated based on the difference between the texture density deviation value and the texture density threshold. This first texture deviation coefficient quantifies the degree of texture density deviation. The slope of the texture matching curve is obtained. The slope reflects the rate of change of texture density, such as whether the texture density changes rapidly or slowly. If the slope of the texture curve is greater than a preset texture curve slope threshold, it means that the texture distribution trend exceeds the normal range. Similarly, the second texture deviation coefficient for the image region to be aligned is calculated based on the comparison between the texture curve slope and the texture curve slope threshold. This coefficient is used to measure the abnormality of the texture distribution trend.

[0108] The texture matching deviation coefficient is obtained by adding the first texture deviation coefficient and the second texture deviation coefficient to obtain the texture matching deviation coefficient of the image region to be aligned. The texture matching deviation coefficient takes into account the deviation in both texture density and variation trend, and comprehensively reflects the degree of difference between the texture features of the region and the standard.

[0109] The grayscale distribution data of the image is analyzed to obtain the grayscale distribution deviation coefficient of the image region to be aligned. This includes the following steps:

[0110] Image grayscale distribution data includes the grayscale values ​​of the image region to be aligned;

[0111] Generate a grayscale distribution curve for the image region to be aligned based on the grayscale values, and obtain the slope of the grayscale curve.

[0112] If the slope of the grayscale curve is greater than the preset grayscale curve slope threshold, the grayscale distribution deviation coefficient of the image region to be aligned is obtained based on the grayscale curve slope and the grayscale curve slope threshold.

[0113] This application first clarifies that the basis of image grayscale distribution data is the grayscale values ​​of the image region to be aligned. The grayscale values ​​reflect the brightness of the image in that region, and the grayscale values ​​at different locations together constitute the grayscale distribution of the region. Based on these grayscale values, a grayscale distribution curve of the image region to be aligned is generated through mathematical fitting.

[0114] The extracted grayscale curve slope is compared with a preset grayscale curve slope threshold. If the grayscale curve slope is greater than this threshold, it indicates that the rate of grayscale change in that area exceeds the normal range, meaning there is an abnormal grayscale distribution. In this case, the grayscale distribution deviation coefficient of the image area to be aligned is obtained based on the difference between the grayscale curve slope and the grayscale curve slope threshold. The grayscale distribution deviation coefficient quantifies the degree of grayscale distribution abnormality. If the deviation coefficient is too large, the corresponding image measurement reference point of the image area to be aligned is marked as a grayscale abnormality reference point.

[0115] The signal feedback distance of the measurement anomaly reference point is obtained based on the measurement anomaly reference point and the preset alignment correction point, specifically including the following steps:

[0116] Acquire the first moment when the reference point transmits the reference point correction signal for the measurement anomaly reference point, and acquire the second moment when the alignment correction point receives the reference point correction signal;

[0117] Based on the signal feedback duration obtained at the first and second moments;

[0118] The signal transmission medium for acquiring infrared spectral images, and the signal feedback speed for obtaining reference point correction signals based on the signal transmission medium;

[0119] The signal feedback distance of the measurement anomaly reference point is obtained based on the signal feedback duration and signal feedback speed.

[0120] The system acquires the first moment when the reference point transmitting the reference point correction signal from the measurement anomaly reference point; this is the starting time of signal transmission. The system also acquires the second moment when the alignment correction point receives the reference point correction signal; this represents the arrival time of the signal. The signal feedback duration, i.e., the time taken for the signal to travel from transmission to reception, is calculated based on the difference between the first and second moments.

[0121] The signal transmission medium for acquiring infrared spectral images is crucial, as different transmission media have varying effects on signal transmission speed. These media include optical fibers and wireless channels. The signal feedback speed of the reference point correction signal within the transmission medium is determined based on its characteristics.

[0122] The signal feedback distance of the measurement anomaly reference point is obtained by calculating the signal feedback duration and speed, and then multiplying the two values ​​to get the final signal feedback distance of the measurement anomaly reference point. The signal feedback distance reflects the spatial transmission distance between the measurement anomaly reference point and the alignment correction point.

[0123] The correction priority of abnormal reference points is obtained based on the reference point correction intensity and signal feedback distance, specifically including the following steps:

[0124] Set the correction intensity weight and feedback distance weight;

[0125] The benchmark correction coefficient for measuring abnormal benchmark points is obtained based on the benchmark correction strength and the correction strength weight.

[0126] The reference point distance coefficient for measuring the anomaly reference point is obtained based on the signal feedback distance and the feedback distance weight.

[0127] The alignment correction coefficient of the measurement abnormal reference point is obtained based on the reference point correction coefficient and the reference point distance coefficient;

[0128] The correction priority of the measurement abnormal reference points is set according to the alignment correction coefficient.

[0129] This application first sets the correction intensity weight and the feedback distance weight. The purpose of the weight is to reflect the importance of the correction intensity and the feedback distance in influencing the correction priority, and it can be adjusted according to the actual measurement alignment requirements.

[0130] The benchmark correction strength and correction strength weight of the measured abnormal benchmark point are multiplied to obtain the benchmark correction coefficient of the measured abnormal benchmark point. This coefficient reflects the quantitative impact of correction strength in priority determination. The signal feedback distance of the measured abnormal benchmark point is multiplied with the feedback distance weight to obtain the benchmark distance coefficient of the measured abnormal benchmark point, which quantifies the impact of feedback distance on priority.

[0131] The alignment correction coefficient of the measurement abnormal reference point is obtained by adding the reference point correction coefficient and the reference point distance coefficient, which comprehensively reflects the overall priority of the reference point that needs to be corrected.

[0132] The abnormal reference points are sorted according to the magnitude of the alignment correction coefficient. Reference points with larger coefficients have higher correction priority and are corrected first; those with smaller coefficients are corrected later. This makes the correction of abnormal reference points during infrared spectral image measurement alignment more systematic, prioritizing those reference points that urgently need correction, improving the efficiency and scientific nature of alignment correction, and ensuring the accuracy and orderliness of the entire infrared spectral image measurement alignment process.

[0133] An artificial intelligence-based infrared spectral image measurement alignment device includes:

[0134] Acquisition module: Acquires infrared spectral image data of the object to be measured, processes the infrared spectral image data to obtain the image region to be aligned in the infrared spectral image, and sets the image measurement reference point based on the image region to be aligned.

[0135] The first processing module: Based on the image measurement reference points, it processes the image region to be aligned to obtain the feature matching deviation coefficient, gray-level distribution deviation coefficient, feature anomaly reference points, and gray-level anomaly reference points;

[0136] The second processing module: constructs measurement anomaly reference points for the infrared spectral image based on feature anomaly reference points and grayscale anomaly reference points; and obtains the alignment deviation of the measurement anomaly reference points based on preset feature deviation weights and feature matching deviation coefficients, and preset grayscale deviation weights and grayscale distribution deviation coefficients.

[0137] The third processing module generates a reference point correction intensity for the measurement abnormal reference point based on the alignment deviation, generates a reference point correction signal for the image measurement reference point based on the reference point correction intensity, and obtains the signal feedback distance of the measurement abnormal reference point based on the measurement abnormal reference point and the preset alignment correction point.

[0138] Positioning module: Based on the benchmark correction intensity and signal feedback distance, the correction priority of the abnormal benchmark point is obtained; Alignment and correction point: Based on the correction priority and the benchmark correction signal, the abnormal benchmark point is aligned, corrected and positioned.

[0139] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0140] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An infrared spectral image measurement alignment method based on artificial intelligence, characterized in that, The method includes the following steps: Acquire the infrared spectral image data of the object to be measured, process the infrared spectral image data to obtain the image region to be aligned in the infrared spectral image, and set the image measurement reference point based on the image region to be aligned. Based on the image measurement reference points, the region of the image to be aligned is processed to obtain the feature matching deviation coefficient, gray-level distribution deviation coefficient, feature anomaly reference points, and gray-level anomaly reference points. Specifically, this includes the following steps: Image feature matching data of the image region to be aligned is obtained based on the image measurement reference point. The image feature matching data includes the contour feature data and texture feature data of the image region to be aligned. The image feature matching data is analyzed to obtain the feature matching deviation coefficient of the image region to be aligned. Based on the feature matching deviation coefficient, the image measurement reference point corresponding to the image region to be aligned is recorded as the feature anomaly reference point. Based on the image measurement reference point, obtain the image grayscale distribution data of the image area to be aligned, analyze the image grayscale distribution data to obtain the grayscale distribution deviation coefficient of the image area to be aligned, and record the image measurement reference point corresponding to the image area to be aligned as the grayscale anomaly reference point; Measurement anomaly reference points for infrared spectral images are constructed based on feature anomaly reference points and grayscale anomaly reference points; the alignment deviation of the measurement anomaly reference points is obtained based on preset feature deviation weights and feature matching deviation coefficients, and preset grayscale deviation weights and grayscale distribution deviation coefficients. A reference point correction intensity is generated based on the alignment deviation, and a reference point correction signal for the image measurement reference point is generated based on the reference point correction intensity. The signal feedback distance of the measurement anomaly reference point is obtained based on the measurement anomaly reference point and the preset alignment correction point. The correction priority of the abnormal reference point is obtained based on the reference point correction intensity and the signal feedback distance; the alignment correction point is used to align and correct the abnormal reference point according to the correction priority and the reference point correction signal.

2. The infrared spectral image measurement alignment method based on artificial intelligence according to claim 1, characterized in that, The infrared spectral image data is processed to obtain the image region to be aligned, specifically including the following steps: The infrared spectral image data includes the image resolution and image size of the object to be measured; Set a standard reference spacing, and divide the image size into regions based on the standard reference spacing to obtain the image region to be aligned in the infrared spectrum image.

3. The infrared spectral image measurement alignment method based on artificial intelligence according to claim 2, characterized in that, The feature matching deviation coefficient of the image region to be aligned is obtained by analyzing the image feature matching data, specifically including the following steps: The contour matching deviation coefficient of the image region to be aligned is obtained by analyzing the contour feature data. The texture matching deviation coefficient of the image region to be aligned is obtained by analyzing the texture feature data. Set contour deviation weights and texture deviation weights, and obtain the feature matching deviation coefficients of the image region to be aligned based on the contour deviation weights and contour matching deviation coefficients, and the texture deviation weights and texture matching deviation coefficients.

4. The infrared spectral image measurement alignment method based on artificial intelligence according to claim 3, characterized in that, The contour matching deviation coefficient of the image region to be aligned is obtained by analyzing the contour feature data, specifically including the following steps: The contour feature data packet is the contour coordinate value of the aligned image region; Generate a contour matching curve corresponding to the image region to be aligned based on the contour coordinate values. If the contour coordinate deviation value in the contour matching curve is greater than the preset contour coordinate threshold, the first contour deviation coefficient of the image region to be aligned is obtained based on the contour coordinate deviation value and the contour coordinate threshold. Obtain the contour curve slope of the contour matching curve. If the contour curve slope is greater than the preset contour curve slope threshold, then obtain the second contour deviation coefficient of the image region to be aligned based on the contour curve slope and the contour curve slope threshold. The contour matching deviation coefficient of the image region to be aligned is obtained based on the first contour deviation coefficient and the second contour deviation coefficient, and the image measurement reference point corresponding to the image region to be aligned is recorded as the feature anomaly reference point.

5. The infrared spectral image measurement alignment method based on artificial intelligence according to claim 4, characterized in that, The texture matching deviation coefficient of the image region to be aligned is obtained by analyzing the texture feature data, specifically including the following steps: The texture feature data includes the texture density value of the image region to be aligned, and a texture matching curve corresponding to the image region to be aligned is generated based on the texture density value. If the texture density deviation value in the texture matching curve is greater than the preset texture density threshold, the first texture deviation coefficient of the image region to be aligned is obtained based on the texture density deviation value and the texture density threshold. Obtain the slope of the texture matching curve; if the slope of the texture curve is greater than a preset texture curve slope threshold, then obtain the second texture deviation coefficient of the image region to be aligned based on the texture curve slope and the texture curve slope threshold. The texture matching deviation coefficient of the image region to be aligned is obtained based on the first texture deviation coefficient and the second texture deviation coefficient.

6. The infrared spectral image measurement alignment method based on artificial intelligence according to claim 5, characterized in that, The grayscale distribution data of the image is analyzed to obtain the grayscale distribution deviation coefficient of the image region to be aligned. This includes the following steps: The image grayscale distribution data includes the grayscale values ​​of the image region to be aligned; Generate a grayscale distribution curve for the image region to be aligned based on the grayscale values, and obtain the slope of the grayscale curve. If the slope of the grayscale curve is greater than the preset grayscale curve slope threshold, then the grayscale distribution deviation coefficient of the image region to be aligned is obtained based on the grayscale curve slope and the grayscale curve slope threshold.

7. The infrared spectral image measurement alignment method based on artificial intelligence according to claim 6, characterized in that, The signal feedback distance of the measurement anomaly reference point is obtained based on the measurement anomaly reference point and the preset alignment correction point, specifically including the following steps: Acquire the first moment when the reference point transmits the reference point correction signal for the measurement anomaly reference point, and acquire the second moment when the alignment correction point receives the reference point correction signal; Based on the signal feedback duration obtained at the first and second moments; The signal transmission medium for acquiring infrared spectral images, and the signal feedback speed for obtaining reference point correction signals based on the signal transmission medium; The signal feedback distance of the measurement anomaly reference point is obtained based on the signal feedback duration and signal feedback speed.

8. The infrared spectral image measurement alignment method based on artificial intelligence according to claim 7, characterized in that, The correction priority of abnormal reference points is obtained based on the reference point correction intensity and signal feedback distance, specifically including the following steps: Set the correction intensity weight and feedback distance weight; The benchmark correction coefficient for measuring abnormal benchmark points is obtained based on the benchmark correction strength and the correction strength weight. The reference point distance coefficient for measuring the anomaly reference point is obtained based on the signal feedback distance and the feedback distance weight. The alignment correction coefficient of the measurement abnormal reference point is obtained based on the reference point correction coefficient and the reference point distance coefficient; The correction priority of the measurement abnormal reference point is set according to the alignment correction coefficient.

9. An artificial intelligence-based infrared spectral image measurement and alignment device, applied to the artificial intelligence-based infrared spectral image measurement and alignment method according to any one of claims 1 to 8, characterized in that, include: Acquisition module: Acquires infrared spectral image data of the object to be measured, processes the infrared spectral image data to obtain the image region to be aligned in the infrared spectral image, and sets the image measurement reference point based on the image region to be aligned. The first processing module: Based on the image measurement reference points, it processes the image region to be aligned to obtain the feature matching deviation coefficient, gray-level distribution deviation coefficient, feature anomaly reference points, and gray-level anomaly reference points. Specifically, it includes the following steps: Image feature matching data of the image region to be aligned is obtained based on the image measurement reference point. The image feature matching data includes the contour feature data and texture feature data of the image region to be aligned. The image feature matching data is analyzed to obtain the feature matching deviation coefficient of the image region to be aligned. Based on the feature matching deviation coefficient, the image measurement reference point corresponding to the image region to be aligned is recorded as the feature anomaly reference point. Based on the image measurement reference point, obtain the image grayscale distribution data of the image area to be aligned, analyze the image grayscale distribution data to obtain the grayscale distribution deviation coefficient of the image area to be aligned, and record the image measurement reference point corresponding to the image area to be aligned as the grayscale anomaly reference point; The second processing module: constructs measurement anomaly reference points for the infrared spectral image based on feature anomaly reference points and grayscale anomaly reference points; and obtains the alignment deviation of the measurement anomaly reference points based on preset feature deviation weights and feature matching deviation coefficients, and preset grayscale deviation weights and grayscale distribution deviation coefficients. The third processing module generates a reference point correction intensity for the measurement anomaly reference point based on the alignment deviation, generates a reference point correction signal for the image measurement reference point based on the reference point correction intensity, and obtains the signal feedback distance of the measurement anomaly reference point based on the measurement anomaly reference point and the preset alignment correction point. The positioning module obtains the correction priority of the abnormal reference point based on the reference point correction intensity and signal feedback distance; the alignment and correction point performs alignment, correction and positioning of the abnormal reference point based on the correction priority and the reference point correction signal.

Images