An image processing and analysis method for industrial non-destructive testing

Abstract

The application discloses an image processing and analysis method for industrial nondestructive testing, and relates to the technical field of industrial testing and image processing, and comprises the following steps: collecting imaging pictures of a testing target and temperature change data of a corresponding testing environment, synchronously arranging the collected imaging pictures and temperature change data according to time sequence, and generating a time-continuous thermal refraction rhythm chart. Through time correlation of the thermal refraction rhythm chart and the refraction offset chart, time synchronization and space stability of the imaging pictures are realized, and pixel drift and virtual shift caused by thermal disturbance are avoided. Through a dynamic deviation correction inlet and an alternating sampling strategy, light refraction errors are compensated in real time, so that the image structure is accurate, the testing result is reliable, and the imaging accuracy and analysis stability of the industrial nondestructive testing are improved.

Description

Technical Field

[0001] This invention relates to the field of industrial inspection and image processing technology, specifically to an image processing and analysis method for industrial non-destructive testing. Background Technology

[0002] Image processing and analysis in industrial non-destructive testing (NDT) refers to a technical process that identifies, locates, and quantitatively assesses internal and surface defects in materials, components, or equipment without damaging the structure and performance of the object being inspected, utilizing image acquisition, digital processing, and intelligent analysis technologies. Its core idea is to acquire multi-source image data of the target object using imaging methods such as X-rays, ultrasound, infrared, lasers, and eddy currents, and then perform denoising, enhancement, segmentation, feature extraction, and pattern recognition processing on the images to reveal hidden microscopic defect information. Subsequently, artificial intelligence algorithms are used to classify and analyze the extracted features, enabling automatic identification and trend assessment of defects such as cracks, porosity, weld anomalies, and debonding, providing visualized and quantitative inspection data for industrial manufacturing, equipment operation, and material quality control.

[0003] The existing technology has the following shortcomings: In existing technologies, industrial non-destructive testing (NDT) largely relies on optical, infrared, or laser imaging for defect identification. When hot air flows within the inspection channel, the air temperature is unevenly distributed in space, and the refractive index changes instantaneously with the temperature gradient, causing the light propagation path to bend. Such refractive disturbances easily induce pixel shifts or local drift during image acquisition, causing the actual defect location within the inspected object to appear as a virtual shift or misalignment in the image. This phenomenon is particularly pronounced in dynamic inspection scenarios such as high-temperature metal castings and hot-rolled workpieces. The inspection system often misinterprets the virtual image caused by refraction as a new defect point, or misjudges the actual defect as a change in position, resulting in incorrect spatial positioning results. These problems are difficult to detect under real-time inspection conditions, often leading to decreased defect positioning accuracy or even distorted inspection results, becoming a hidden danger of imaging errors in high-temperature dynamic scenes that is difficult to avoid in existing technologies.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide an image processing and analysis method for industrial non-destructive testing to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an image processing and analysis method for industrial non-destructive testing, comprising the following steps: The system collects images of the target and temperature change data of the corresponding detection environment. The collected images and temperature change data are then synchronized and organized in chronological order to generate a time-continuous thermal refraction rhythm map, which is used to establish a time reference for subsequent image analysis and light refraction comparison. By using thermal refraction rhythm maps, brightness change analysis and edge offset detection are performed frame by frame on continuous imaging images. The abrupt change points of light refraction are extracted, and thermal disturbance trajectory sequences are generated according to the time series distribution to provide input data for subsequent refraction offset tracking. Based on the thermal perturbation trajectory sequence, spatially stable reference points are matched in continuous imaging images to calculate the light drift path caused by temperature fluctuations, generate a refraction offset map, and establish a rhythm with thermal refraction within the refraction offset map. Figure 1 A consistent time scale correlation is used for time-series analysis of refraction errors; By retrospectively analyzing the time segments in which the light shift is concentrated in the refraction shift image, high-risk areas of light drift are identified, the specific causes and temporal relationships of image shift are identified, and a dynamic correction input is output to guide the subsequent image correction process. Based on the dynamic correction input, the detection image is processed. During the thermal disturbance stage, reverse sampling and staggered exposure are alternately used according to the time sequence information of the refraction offset map to perform pixel-level position correction on the continuous imaging image and realize dynamic compensation for light offset. During the temperature cooling stage, the position of the imaging image is corrected and optical distortion correction is completed by pausing sampling and coordinating the sampling order.

[0007] Preferably, the steps for generating the thermal refraction rhythm diagram are as follows: Synchronous acquisition conditions are established for the imaging image of the detection target and the temperature change data of the detection environment. The time alignment between the detection image and the temperature sampling is determined by setting a unified reference for the acquisition timing. The acquired images and temperature change data are synchronized and organized in chronological order, and the image frames and corresponding temperature data are arranged according to timestamps to form a continuous time sequence. By analyzing the synchronized imaging images and temperature change data in a time series, a time-continuous thermal refraction rhythm map reflecting the trend of thermal disturbance changes is generated. By mapping the refraction change information at time nodes in the thermal refraction rhythm diagram to the image frame position, a time index system is formed that includes the correspondence between temperature, light shift, and image changes.

[0008] Preferably, the steps for generating the thermal perturbation trajectory sequence are as follows: Brightness changes are analyzed frame by frame in continuous imaging, and regions of sudden brightness changes are extracted as light refraction change regions based on the time reference in the thermal refraction rhythm diagram. Based on the brightness change analysis results, edge offset detection is performed on the imaging area affected by thermal disturbance, and an edge displacement curve reflecting the degree of influence of light refraction is generated; By combining brightness changes with edge displacement curves, the locations where light refraction abruptly changes are identified, and these locations are recorded in chronological order to form a set of abrupt refraction change points. The refraction mutation points are organized according to their time series distribution to generate a continuous thermal disturbance trajectory sequence that reflects the changing trend of the light propagation path.

[0009] Preferably, the steps for generating the refraction offset map are as follows: By utilizing the temporal distribution characteristics of thermal perturbation trajectory sequences, spatially stable reference points are identified and matched in continuous imaging images, and a spatial coordinate system is established. The light drift path caused by temperature fluctuation is calculated and continuous light drift path data is obtained by using the light refraction information in the spatial reference point and the light disturbance trajectory sequence. The calculation results of the light drift path are visualized using spatial coordinates to form a refraction offset map; By correlating the refraction offset map with the thermal refraction rhythm map over time, a temporal matching system is formed that includes the correspondence between spatial change trajectory and temperature change rhythm.

[0010] Preferably, during the time-scale correlation establishment phase, the time-scale correlation is established by setting a time-scale correlation with the thermal refraction rhythm within the refraction offset diagram. Figure 1 The time scale is aligned with the time nodes of the light drift path and the time nodes of temperature fluctuation, and the spatial position change information of each time node is marked in the refraction offset diagram, thus forming a refraction offset time sequence structure containing time markers and spatial displacement relationships, which is used for continuous analysis and staged comparison of refraction errors.

[0011] The preferred dynamic correction input / output steps are as follows: Based on the time reference of the thermal refraction rhythm diagram, the time period of concentrated distribution of light offset in the refraction offset diagram is traced in reverse time to form a dynamic sequence of light offset changing with time. The dynamic sequence of light drift is used to identify spatial regions with frequent and unstable light drift and to determine the boundary range of high-risk areas of light drift. A time correlation analysis was performed on the formation process of light drift in high-risk areas to determine the specific causes of image motion and the temporal correspondence between temperature changes; By combining high-risk areas with virtual time relationships, a dynamic correction entry point containing time scale and spatial distribution information is formed and used to guide the image correction process.

[0012] Preferably, in the dynamic correction entry output stage, the dynamic correction entries are arranged in chronological order and marked with corresponding time nodes and spatial locations of light drift to form a correction entry mark sequence containing time scale and spatial range of action. The correction entry mark sequence is used to trigger image correction operation during the critical time period of light drift to reduce the imaging spatial positioning offset caused by thermal disturbance.

[0013] Preferably, the image processing and analysis steps performed on the detection image based on the dynamic correction entry are as follows: Based on the time node information in the dynamic correction inlet, the detection screen is time-synchronized and responded to, and the acquisition sequence and refraction change rhythm are kept consistent. Based on the time synchronization results, backsampling and staggered exposure operations are alternately performed during thermal disturbance to form a sequence of image frames containing different refraction states; Using the image frame sequence obtained by alternating sampling, the light drift path in the continuous imaging scene is corrected at the pixel level and the real spatial position is restored based on the temporal information in the refraction offset map; A short sampling pause is set during the temperature drop process, and the position correction and optical distortion compensation of the imaging image are completed by coordinating and adjusting the sampling order before and after the sampling.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention establishes a temporal correlation between a thermal refraction rhythm map and a refraction shift map, enabling the detection image to possess temporal synchronization characteristics under thermal disturbance conditions, thereby maintaining the continuity and stability of imaging data in dynamic high-temperature environments. By temporally matching the imaging image with temperature change data, the transient change law of air refractive index can be accurately reflected, making the light refraction path traceable on the time axis and avoiding the pixel drift accumulation effect caused by hot air flow. This method effectively improves the spatial positioning consistency of the imaging image, ensuring that the detection results remain clear, comparable, and repeatable in dynamic thermal fields, providing a stable and reliable image foundation for defect identification.

[0015] This invention introduces a dynamic correction inlet and an alternating sampling strategy to create an adaptive correction mechanism for image acquisition and light propagation, achieving pixel-level optical error compensation during the thermal disturbance stage. By combining backsampling with staggered exposure, the phase error in the light drift direction is balanced. Combined with sampling sequence adjustment during the cooling stage, image shift and spatial distortion caused by refraction disturbances are effectively suppressed. This method enables the imaging image to maintain geometric accuracy in real time, improving the spatial precision and reliability of the detection results and providing high-fidelity image analysis capabilities for dynamic scenarios in industrial non-destructive testing. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of an image processing and analysis method for industrial non-destructive testing according to the present invention. Detailed Implementation

[0018] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0019] This invention provides, for example Figure 1 The image processing and analysis method for industrial non-destructive testing shown includes the following steps: The system collects images of the target and temperature change data of the corresponding detection environment. The collected images and temperature change data are then synchronized and organized in chronological order to generate a time-continuous thermal refraction rhythm map, which is used to establish a time reference for subsequent image analysis and light refraction comparison. The imaging image of the target and the corresponding temperature change data of the detection environment are simultaneously collected and processed to generate a time-continuous thermal refraction rhythm map, thereby establishing a time reference for subsequent image analysis and light refraction comparison. The specific implementation steps are as follows: During the data acquisition preparation phase, conditions for synchronous acquisition of the target image and the temperature change data of the detection environment are established. This phase determines the time alignment between the image and temperature sampling by setting a unified reference for the acquisition sequence. To ensure a strict temporal correspondence between the image and temperature change data, continuous sampling is used to acquire the target image during the detection process, while simultaneously recording the temperature changes of the detection environment within the same time period. At this point, each frame of the image forms a one-to-one temporal match with its corresponding temperature data, ensuring that changes in brightness and texture of the image and temperature fluctuations within the detection space remain consistent over time. This time-synchronized acquisition method yields a continuous image sequence containing temperature field change information, providing a reliable source of basic information for subsequent data processing.

[0020] During the data processing phase, the acquired imaging images and temperature change data are synchronized in chronological order. The purpose of this phase is to ensure a strict, continuous correspondence between data from different sources on the timeline, forming a complete time series structure. To this end, the imaging images are arranged according to their acquisition timestamps, and the temperature change data corresponding to each time point is sequentially inserted into the corresponding time node, forming a unified time series chain. During the processing, periods with abnormal sampling intervals are removed to ensure that each node in the time series has a complete matching relationship between image and temperature data. In this way, a set of time-continuous and complete image-temperature data pairs can be formed, making the temporal relationship between temperature fluctuations and image changes clearly visible. The processed data structure is continuous and traceable, fully reflecting the correspondence between temperature gradient changes caused by hot air flow and image brightness changes during the detection process.

[0021] In the rhythm map generation stage, the synchronized image data and temperature change data are analyzed according to a time series to form a continuous thermal refraction rhythm map. The key to this stage is integrating image brightness changes, edge shift changes, and temperature change trends through continuous temporal correlation, allowing the impact of thermal disturbances on the image to form an observable continuous rhythm on the time axis. Specifically, for each moment's corresponding image frame and temperature data, the corresponding information between the brightness change area caused by thermal refraction and the temperature change rate is extracted and combined in chronological order to form a continuous rhythm map with time as the horizontal axis and refraction influence characteristics as the vertical axis. This thermal refraction rhythm map can reflect the trend of light path changes caused by temperature fluctuations in a continuous time manner, thus providing a clear temporal reference for subsequent image analysis and light refraction comparison. The formation of the rhythm map not only records the periodic characteristics of thermal disturbances in the detection environment but also preserves the changing patterns of light propagation path shifts in different time periods, enabling subsequent analysis steps to quantitatively compare the refraction effects at different time periods based on this rhythm map.

[0022] In the time reference establishment phase, the generated thermal refraction rhythm map is used as a unified time reference for subsequent image analysis and light refraction comparison. In this phase, a mapping relationship is established between the refraction change information at each time node in the rhythm map and the corresponding image frame position, thus forming a time index system that includes the correspondence between temperature, light shift, and image change. Through this index system, the refraction state changes of the detected image can be tracked in different time periods based on the time reference of the rhythm map during subsequent image analysis, and light shift recognition and image correction analysis operations can be performed accordingly. The establishment of this time reference ensures that the entire detection process has a continuous time reference, guaranteeing that the optical state of each image frame can be mapped to a specific temperature change moment in subsequent analysis and tracking steps, thereby maintaining the temporal consistency of the detection data and the comparability of the analysis.

[0023] By using thermal refraction rhythm maps, brightness change analysis and edge offset detection are performed frame by frame on continuous imaging images. The abrupt change points of light refraction are extracted, and thermal disturbance trajectory sequences are generated according to the time series distribution to provide input data for subsequent refraction offset tracking. To address the temperature fluctuation information contained in the thermal refraction rhythm map, frame-by-frame analysis of continuous imaging images is performed to extract abrupt changes in light refraction caused by thermal disturbances and generate a continuous thermal disturbance trajectory sequence. This provides data input for subsequent refraction offset tracking. The specific implementation steps are as follows: In the brightness change extraction stage, frame-by-frame brightness change analysis is performed on continuous imaging images. This stage uses the time reference recorded in the thermal refraction rhythm map as a reference, comparing each frame with the previous frame to analyze the brightness distribution change trend within the same spatial area. Since hot air disturbances cause slight deflections in the light propagation path, creating unstable areas of brightness distribution in the imaging image, the continuity of brightness distribution is observed during frame-by-frame analysis, extracting areas of abrupt brightness changes as potential refraction change areas. By continuously tracking these areas, light disturbances caused by hot air flow can be distinguished from the overall brightness change, allowing subsequent edge detection to focus on the actual range of light offset. During this process, the temporal distribution of brightness changes remains consistent with the temperature fluctuation trend of the thermal refraction rhythm map, ensuring that the brightness change analysis results accurately reflect the rhythmic characteristics of thermal disturbances, providing fundamental data for subsequent edge offset detection.

[0024] In the edge shift detection stage, based on the results of brightness change analysis, edge shift detection is performed on areas affected by thermal disturbances in continuous imaging images. The goal of this stage is to determine the displacement range caused by light refraction on the spatial structure of the image. Because light bends in its refraction direction with temperature gradients when propagating in an air medium with uneven temperature, the position of the edge of the same object in the image shifts in different frames. By detecting the changes in the edge contour position of the same area in consecutive frames, spatial drift information of light in the time series can be obtained. During the detection process, using the time nodes in the thermal refraction rhythm map as references, the changing trend of edge shift amplitude within each time period is organized to form an edge displacement curve reflecting the degree of influence of light refraction. In this way, optical drift caused by thermal disturbance can be separated from the natural brightness changes in the imaging image, thus providing a spatial positioning basis for subsequent abrupt change point extraction. In this stage, brightness change areas and edge shift areas are mapped to each other, keeping brightness abrupt changes and positional drift synchronously correlated in time, forming a complete light refraction influence feature.

[0025] In the refraction abrupt change point extraction stage, the location of the abrupt change in light refraction is identified by combining comprehensive information of brightness changes and edge displacements. The core of this stage lies in extracting the moment and spatial location of the abrupt change in the light path from the brightness change curves and edge displacement curves of consecutive frames. By comparing the temporal continuity of brightness changes with the spatial continuity of edge displacements, the transient occurrence point of refraction disturbance can be determined. The light refraction abrupt change point is usually manifested as the moment when a sudden increase or decrease in brightness gradient and a sudden change in edge position occur simultaneously. The spatial location corresponding to this moment is the location where the thermal disturbance has the strongest impact. The extracted refraction abrupt change points are recorded in chronological order and correspond one-to-one with the temperature change nodes in the thermal refraction rhythm diagram, thus forming a discrete set of nodes of light refraction changes on the time axis. Each refraction abrupt change point represents a refraction disturbance event of thermal disturbance in the imaging image. The temporal and spatial distribution of these discrete events constitutes the basic data for subsequent trajectory generation. Through this extraction stage, the originally scattered light refraction changes can be transformed into a traceable set of temporal features, making the disturbance characteristics of the light propagation path traceable in time.

[0026] In the thermal perturbation trajectory generation stage, the extracted light refraction abrupt change points are organized according to their time series distribution order to generate a continuous thermal perturbation trajectory sequence. The task of this stage is to connect the discrete abrupt change points obtained in the previous stage along the time dimension, so that the dynamic change process of light refraction is presented in the form of a continuous curve. To this end, according to the time reference order in the thermal refraction rhythm diagram, the time coordinates and spatial coordinates of each refraction abrupt change point are paired and arranged sequentially to form a time series of light refraction changes. Each thermal perturbation trajectory sequence reflects the trend of light propagation path changes under the influence of heated air disturbance. Through this sequence, the changes and shifts in the refraction direction of light in different time periods can be clearly observed, thus providing continuous and complete input data for subsequent refraction offset tracking. The thermal perturbation trajectory sequence not only contains the drift trajectory of light in space but also records the time response characteristics of light refraction with temperature changes, enabling subsequent steps to perform accurate tracking and time-matching analysis of light refraction offset based on this sequence.

[0027] Based on the thermal perturbation trajectory sequence, spatially stable reference points are matched in continuous imaging images to calculate the light drift path caused by temperature fluctuations, generate a refraction offset map, and establish a rhythm with thermal refraction within the refraction offset map. Figure 1 A consistent time scale correlation is used for time-series analysis of refraction errors; By matching spatially stable reference points in continuous imaging images using thermal perturbation trajectory sequences, the light drift path caused by temperature fluctuations is calculated, and a refraction offset map is generated. Simultaneously, a rhythmic relationship with thermal refraction is established within the refraction offset map. Figure 1 The established time scale correlation is used for subsequent time-series analysis of refraction errors. The specific implementation steps are as follows: In the spatial reference point matching stage, the temporal distribution characteristics of the thermal perturbation trajectory sequence are utilized to identify and match spatially stable reference points in continuous imaging. The purpose of this stage is to establish a spatial anchoring basis for ray drift analysis. Due to the continuous hot air flow and temperature fluctuations during the detection process, the texture and edges of some image areas may experience slight shifts due to refraction. Therefore, matching spatially stable reference points is crucial for the accurate derivation of subsequent ray drift paths. During the matching process, fixed structures, background details, or optical textures in continuous imaging are compared frame by frame to identify areas that maintain a constant position under the influence of thermal perturbation. These areas are then used as spatially stable reference points. These reference points maintain their spatial position on the time axis, providing a reliable reference benchmark for calculating the relative path of ray drift. By establishing a spatial coordinate system for these reference points, the relative displacement of ray drift can be transformed into absolute spatial changes in subsequent steps, forming a spatial reference framework for the ray drift path.

[0028] In the ray drift calculation stage, the ray drift path caused by temperature fluctuations is calculated using the matched spatial reference points and ray refraction information from the thermal perturbation trajectory sequence. The task of this stage is to integrate the thermal perturbation features in the temporal dimension with the refraction changes in the spatial dimension, thereby obtaining the continuous change process of the ray propagation path. Specifically, by comparing the image brightness distribution and edge morphology changes at the same spatial location at different time points, the direction and distance of ray displacement in space can be determined. Since the ray deflection direction is closely related to the temperature gradient distribution, combined with the time series of refraction abrupt change points recorded in the thermal perturbation trajectory sequence, the continuous drift trend of ray with temperature fluctuations can be derived. To ensure the consistency of temporal and spatial data, the ray drift calculation uses the time nodes of the thermal refraction rhythm map as the main index, and the drift result at each moment can be mapped to the temperature change moment in the rhythm map. Through this joint analysis method of temporal synchronization and spatial matching, the propagation path of ray in the dynamic thermal field can be completely reconstructed, thereby obtaining continuous and stable ray drift path data, providing continuous input for the generation of the refraction offset map.

[0029] In the refraction offset map generation stage, the calculated light drift path results are visualized using spatial coordinates to form a refraction offset map. The core task of this stage is to record the spatiotemporal characteristics of the light drift path in a graphical manner, clearly presenting the trend and range of light drift in a two-dimensional plane. During the generation of the refraction offset map, using the spatial distribution of the detection image as a reference, the light drift direction and displacement amplitude at each time point are mapped to the corresponding position in the image according to spatial coordinates. The drift information at each time point is sequentially superimposed to form a continuous offset trajectory distribution map of the light during the detection period, thus constructing a visual result that reflects the law of refraction change over time. This refraction offset map not only includes information such as the spatial offset direction and offset distance of the light, but also reflects the differences in light path caused by temperature fluctuations in different regions, giving the impact of refraction on the imaging image spatial distribution characteristics. Through this generation process, the offset shape caused by thermal disturbances to optical imaging can be intuitively presented, providing an intuitive reference basis for the subsequent establishment of time scales and refraction error analysis.

[0030] In the time-scale correlation establishment phase, the refraction offset map and the thermal refraction rhythm map are correlated in time to form a complete time-series matching system. The task of this phase is to establish a one-to-one correspondence between the spatial change trajectory in the refraction offset map and the temperature change rhythm in the thermal refraction rhythm map, so as to analyze the refraction offset in the time dimension. Specifically, this is achieved by setting a correlation between the refraction offset map and the thermal refraction rhythm... Figure 1 A consistent time scale aligns the time nodes of the light drift path with the time nodes of temperature fluctuations, allowing each segment of the light drift path to be traced back to its corresponding temperature change stage. Thus, each point in the refraction drift map not only reflects spatial displacement but also carries temporal information, giving the refraction process a complete temporal marker. By establishing this time scale, a temporal comparison and staged analysis of the refraction drift can be performed, revealing the response patterns of light drift under different temperature fluctuation cycles. This time scale correlation not only provides a foundation for the temporal analysis of refraction errors but also provides a quantifiable temporal reference for subsequent refraction error correction and dynamic image correction, ensuring the entire light refraction process is continuous in time and comparable in space.

[0031] By retrospectively analyzing the time segments in which the light shift is concentrated in the refraction shift image, high-risk areas of light drift are identified, the specific causes and temporal relationships of image shift are identified, and a dynamic correction input is output to guide the subsequent image correction process. Retrospective analysis is performed on time segments where ray shift is concentrated in the refraction-shift image to identify high-risk areas of ray drift. Furthermore, the specific causes and temporal relationships of image artifacts are identified, ultimately outputting a dynamic correction input to guide subsequent image correction processes. The specific implementation steps are as follows: In the time segment backtracking phase, the target time periods are those with concentrated light shifts in the refraction offset map. These time segments are then traced back in reverse time based on the time reference of the thermal refraction rhythm map. The main task of this phase is to identify the time intervals with the most dramatic changes in light drift in the refraction offset map and compare these time periods with the temperature change nodes in the thermal perturbation rhythm map to determine the temporal correlation between thermal perturbation and changes in light refraction. In practice, areas with densely overlapping light shift trajectories in the refraction offset map are selected as the starting point for analysis. By tracing the formation process of these areas on the time axis, the start time, duration, and decay process of the light shift are identified. During the backtracking analysis, temporal continuity is the main focus. The direction and magnitude of light drift between different time nodes, as well as the changing trends between adjacent time frames, are compared to form a dynamic sequence of light shift changes over time. The output of this phase is a set of temporally continuous light shift records, used to describe the occurrence and change process of light drift, laying the temporal foundation for subsequent high-risk area identification.

[0032] In the high-risk area identification phase, the temporally continuous ray migration records obtained from backtracking analysis are used to identify spatial regions in the refraction migration map where ray drift is frequent and the direction is unstable. The task of this phase is to determine the areas where the ray propagation path experiences concentrated shifts under thermal disturbance, i.e., high-risk areas for ray drift. By observing the displacement trajectory of light in continuous time segments, it can be found that the ray trajectories in some areas exhibit concentrated intersections or bending patterns, indicating that these areas were strongly disturbed by hot air flow during the detection process. Analysis of the spatial distribution of these areas can determine the boundary range of the areas with the most concentrated ray drift. To ensure the consistency of the identification results, each high-risk area is mapped to its corresponding temperature node in the thermal refraction rhythm map, maintaining a one-to-one match between the spatial distribution of ray drift and the temporal process of temperature fluctuations. This allows high-risk areas to not only have spatial location significance but also background information on temperature changes, thereby revealing the heat source characteristics and periodicity of ray drift. In this phase, the joint analysis of temporal and spatial information provides reliable reference conditions for subsequent identification of the specific causes of image aliasing.

[0033] In the image shift cause identification stage, a temporal correlation analysis is performed on the formation process of light drift in high-risk areas to determine the specific causes of image shift and their correspondence with temperature changes. The purpose of this stage is to reveal the formation mechanism of image shift caused by thermal disturbances from a temporal perspective. Based on the light drift sequence obtained in the retrospective analysis stage and the spatial distribution data of the high-risk areas identified in the stage, a corresponding analysis is performed on the start time, rate of change, and spatial direction of light drift. Through this temporal-spatial correlation analysis, it can be determined whether the image shift is caused by a sudden change in local temperature, uneven laminar flow of hot air, or the cumulative effect of refraction caused by continuous fluctuations. When the direction of light drift is consistent with the gradient trend of temperature change, it can be inferred that the bending of the light path is directly affected by the refraction of hot air; while when the light drift reverses direction or is interrupted in a short period of time, it indicates that there is unstable thermal disturbance in the area, leading to image shift in consecutive frames. Through this analysis process, the root cause of image shift and its temporal coupling relationship with temperature fluctuations can be clarified, providing a targeted basis for the output of the dynamic correction input. The output of this stage includes not only the formation mechanism of virtual shift, but also the time range and duration of the virtual shift, providing time coordinate input for the next stage.

[0034] In the dynamic correction input output stage, the identified high-risk areas are combined with the temporal relationship of the ray drift to form a dynamic correction input, which guides the subsequent image correction process. The main task of this stage is to transform the ray drift patterns and temporal characteristics obtained in the previous stage into correction reference inputs that can be used for image processing. To this end, the time nodes and spatial positions of ray drift are marked on the refraction offset map, forming a correction input mark sequence containing temporal scale and spatial distribution information. Each correction input corresponds to a specific thermal disturbance time segment and its spatial range of influence, used to guide subsequent image position correction and optical distortion compensation. During the output process, multiple correction inputs are arranged in chronological order, so that correction commands can be triggered step by step along the time axis in subsequent image processing stages, thereby performing correction operations in a timely manner during the critical time period when ray drift occurs. Through this dynamic output method, while ensuring image continuity, the impact of thermal disturbance on the spatial positioning of the image can be effectively reduced, giving the correction process both temporal matching characteristics and spatial specificity.

[0035] Based on the dynamic correction input, image processing and analysis operations are performed on the detection image. During the thermal disturbance stage, the reverse sampling and staggered exposure strategies are alternately used according to the time sequence information of the refraction offset map to perform pixel-level position correction on the continuously acquired imaging image, so that the offset caused by light refraction is dynamically compensated. During the temperature cooling stage, sampling is paused. By coordinating and adjusting the sampling order, the spatial position of the imaging image is kept consistent with the light propagation state, and the optical distortion caused by thermal disturbance is continuously corrected. Based on the dynamic correction input, image processing and analysis operations are performed on the detection image. During the thermal disturbance phase, a strategy of alternating backsampling and staggered exposure is employed, combined with temporal information from the refraction offset map, to perform pixel-level real-time correction on continuously acquired imaging frames. Simultaneously, a short sampling pause is set during the temperature cooling phase. By coordinating and adjusting the sampling order, the position of the imaging image is continuously corrected and optical distortion is compensated, thereby obtaining stable and accurate detection image analysis results. The specific implementation steps are as follows: During the thermal disturbance response phase, based on the time node information recorded in the dynamic correction inlet, the detection image undergoes time synchronization control and response processing during the thermal disturbance. The goal of this phase is to determine the start time, duration, and decay time of the light disturbance based on the time sequence markers in the refraction offset map, thereby maintaining consistency between the timing of image acquisition and the rhythm of refraction changes during imaging. During execution, continuous control of the detection image acquisition frequency ensures that the acquisition operation is synchronized with the refraction offset changes, guaranteeing that each frame of the image acquires accurate light projection information under the corresponding thermal disturbance state. To reduce the impact of pixel displacement caused by transient light refraction, during the intense thermal disturbance phase, the uniformity of the acquisition rhythm is controlled, ensuring a correspondence between the time interval between consecutive frames and the refraction offset period, thus fully recording the time span of light drift. Through this phase, the acquisition time of the detection image and the dynamic rhythm of the refraction disturbance are unified, providing a time synchronization basis for subsequent backsampling and staggered exposure, ensuring consistency between image acquisition and the light change process on the time axis.

[0036] During the alternating sampling phase, based on the time synchronization results of the previous phase, backsampling and staggered exposure operations are alternately performed during thermal disturbances to reduce the impact of hot air fluctuations on image stability. Backsampling acquires image frames at opposite times when the direction of light refraction changes, allowing the directional errors of light propagation deflection to cancel each other out in the temporal superposition. Staggered exposure adjusts the relative phase of the exposure start time and the rhythm of the thermal disturbance, ensuring that the acquired frames capture different stages of the thermal flow disturbance at different time points. This alternating approach allows for a complete mapping of the impact of thermal disturbance on light over time, forming a sequence of image frames containing different refraction states. To ensure coordination between backsampling and staggered exposure, the switching rhythm of the two sampling methods is controlled by the time scale of the refraction offset map in each set of alternating sampling, maintaining a complementary relationship between the change in the direction of light deflection and the exposure time. This alternating sampling strategy effectively separates the light bending information generated by thermal disturbances from the true structural features of the detected object, enabling subsequent image correction to distinguish the positional relationship between the virtual displacement caused by refraction offset and the actual physical form.

[0037] In the pixel-level correction stage, the image frame sequence obtained from the alternating sampling in the previous stage is used to correct the light drift path in the continuous imaging image at the pixel level. The core task of this stage is to restore the true propagation trajectory of light under thermal perturbation through temporal correlation and spatial registration, so that the pixel positions between consecutive frames are re-corresponding to the true spatial coordinates of the detected object. During the operation, the temporal information of the light drift path in the refraction offset map is used as a reference, and the pixel positions in each time frame are compared and adjusted one by one with their original spatial positions before thermal perturbation, thereby correcting the pixel offset error caused by light refraction. The pixel correspondence between consecutive frames is organized according to the temporal order of the thermal perturbation trajectory, so that the same spatial point is restored to the same imaging position in different time frames. This pixel-level correction process can effectively reduce the image shift phenomenon caused by light deflection and restore the true image of the surface and internal structure of the detected object. Through this two-dimensional pixel correction method of time and space, the consistency and hierarchical continuity of the image structure can be maintained in a dynamic thermal environment, so that subsequent optical distortion compensation operations have a precise pixel basis.

[0038] During the sequential adjustment and compensation phase of the cooling stage, a short sampling pause is set for the gradual temperature decrease. Through coordinated adjustment of the sampling sequence, the position of the image is continuously corrected and optical distortion is compensated. The task of this stage is to utilize the stable imaging conditions during the thermal disturbance dissipation period to repair and balance the residual effects of light refraction caused by the previous thermal disturbance. The short sampling pause allows for a temporary halt in acquisition when the refractive shift changes tend to stabilize, restoring the light to a uniform state along the propagation path. Subsequently, when sampling resumes, the image frames before and after the pause are rearranged chronologically to maintain continuity of the image on the time axis. Based on this, the spatial offset between the preceding and following sampling frames is coordinated to ensure that the trend of image position change matches the fading process of light drift, thus completing optical distortion compensation. Through this stage, the pixel positions of the image remain stable after the thermal disturbance dissipates, and the true shape of the detected object is restored in the image. Ultimately, through continuous processing throughout the entire process, the detected image achieved stable imaging results and accurate spatial positioning under thermal disturbance conditions, enabling the output detected image to have high temporal consistency and spatial accuracy, providing a reliable foundation for subsequent image analysis and defect identification.

[0039] This invention establishes a temporal correlation between a thermal refraction rhythm map and a refraction shift map, enabling the detection image to possess temporal synchronization characteristics under thermal disturbance conditions, thereby maintaining the continuity and stability of imaging data in dynamic high-temperature environments. By temporally matching the imaging image with temperature change data, the transient change law of air refractive index can be accurately reflected, making the light refraction path traceable on the time axis and avoiding the pixel drift accumulation effect caused by hot air flow. This method effectively improves the spatial positioning consistency of the imaging image, ensuring that the detection results remain clear, comparable, and repeatable in dynamic thermal fields, providing a stable and reliable image foundation for defect identification.

[0040] This invention introduces a dynamic correction inlet and an alternating sampling strategy to create an adaptive correction mechanism for image acquisition and light propagation, achieving pixel-level optical error compensation during the thermal disturbance stage. By combining backsampling with staggered exposure, the phase error in the light drift direction is balanced. Combined with sampling sequence adjustment during the cooling stage, image shift and spatial distortion caused by refraction disturbances are effectively suppressed. This method enables the imaging image to maintain geometric accuracy in real time, improving the spatial precision and reliability of the detection results and providing high-fidelity image analysis capabilities for dynamic scenarios in industrial non-destructive testing.

[0041] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. An image processing and analysis method for industrial non-destructive testing, characterized in that, Includes the following steps: The system collects images of the target and temperature change data of the corresponding detection environment, and then synchronizes and organizes the collected images and temperature change data in chronological order to generate a time-continuous thermal refraction rhythm map. The brightness change and edge shift of continuous imaging images are analyzed frame by frame using thermal refraction rhythm diagrams. The abrupt change points of light refraction are extracted and thermal disturbance trajectory sequences are generated according to the time series distribution. Based on the thermal disturbance trajectory sequence, a spatially stable reference point is matched in the continuous imaging image. The light drift path caused by temperature fluctuation is calculated, a refraction offset map is generated, and a time scale correlation consistent with the thermal refraction rhythm map is established in the refraction offset map. By retrospectively analyzing the time segments in which the light shift is concentrated in the refraction shift image, high-risk areas of light drift are identified, the specific causes and temporal relationships of image motion are identified, and a dynamic correction input is output. Based on the dynamic correction input, the detection image is processed. During the thermal disturbance stage, reverse sampling and staggered exposure are alternately used according to the time sequence information of the refraction offset map to perform pixel-level position correction on the continuous imaging image and realize dynamic compensation for light offset. During the temperature cooling stage, the position of the imaging image is corrected and optical distortion correction is completed by pausing sampling and coordinating the sampling order.

2. The image processing and analysis method for industrial nondestructive testing according to claim 1, characterized in that, The steps for generating a thermal refraction rhythm diagram are as follows: Synchronous acquisition conditions are established for the imaging image of the detection target and the temperature change data of the detection environment. The time alignment between the detection image and the temperature sampling is determined by setting a unified reference for the acquisition timing. The acquired images and temperature change data are synchronized and organized in chronological order, and the image frames and corresponding temperature data are arranged according to timestamps to form a continuous time sequence. By analyzing the synchronized imaging images and temperature change data in a time series, a time-continuous thermal refraction rhythm map reflecting the trend of thermal disturbance changes is generated. By mapping the refraction change information at time nodes in the thermal refraction rhythm diagram to the image frame position, a time index system is formed that includes the correspondence between temperature, light shift, and image changes.

3. The image processing and analysis method for industrial nondestructive testing according to claim 2, characterized in that, The steps for generating the thermal perturbation trajectory sequence are as follows: Brightness changes are analyzed frame by frame in continuous imaging, and regions of sudden brightness changes are extracted as light refraction change regions based on the time reference in the thermal refraction rhythm diagram. Based on the brightness change analysis results, edge offset detection is performed on the imaging area affected by thermal disturbance, and an edge displacement curve reflecting the degree of influence of light refraction is generated; By combining brightness changes with edge displacement curves, the locations where light refraction abruptly changes are identified, and these locations are recorded in chronological order to form a set of abrupt refraction change points. The refraction mutation points are organized according to their time series distribution to generate a continuous thermal disturbance trajectory sequence that reflects the changing trend of the light propagation path.

4. The image processing and analysis method for industrial nondestructive testing according to claim 3, characterized in that, The steps for generating the refraction offset map are as follows: By utilizing the temporal distribution characteristics of thermal perturbation trajectory sequences, spatially stable reference points are identified and matched in continuous imaging images, and a spatial coordinate system is established. The light drift path caused by temperature fluctuation is calculated and continuous light drift path data is obtained by using the light refraction information in the spatial reference point and the light disturbance trajectory sequence. The calculation results of the light drift path are visualized using spatial coordinates to form a refraction offset map; By correlating the refraction offset map with the thermal refraction rhythm map over time, a temporal matching system is formed that includes the correspondence between spatial change trajectory and temperature change rhythm.

5. The image processing and analysis method for industrial nondestructive testing according to claim 4, characterized in that, In the time scale association establishment stage, by setting a time scale consistent with the thermal refraction rhythm diagram in the refraction offset diagram, the time nodes of the light drift path are aligned with the time nodes of temperature fluctuation, and the spatial position change information of each time node is marked in the refraction offset diagram, thereby forming a refraction offset time sequence structure containing time markers and spatial displacement relationships.

6. The image processing and analysis method for industrial nondestructive testing according to claim 4, characterized in that, The dynamic correction input and output steps are as follows: Based on the time reference of the thermal refraction rhythm diagram, the time period of concentrated distribution of light offset in the refraction offset diagram is traced in reverse time to form a dynamic sequence of light offset changing with time. The dynamic sequence of light drift is used to identify spatial regions with frequent and unstable light drift and to determine the boundary range of high-risk areas of light drift. A time correlation analysis was performed on the formation process of light drift in high-risk areas to determine the specific causes of image motion and the temporal correspondence between temperature changes; By combining high-risk areas with virtual time relationships, a dynamic correction entry point containing time scale and spatial distribution information is formed and used to guide the image correction process.

7. The image processing and analysis method for industrial nondestructive testing according to claim 6, characterized in that, In the dynamic correction entry output stage, the dynamic correction entries are arranged in chronological order and marked with corresponding time nodes and spatial locations of light drift, forming a correction entry mark sequence that includes time scale and spatial range of action. The correction entry mark sequence is used to trigger image correction operation during the critical time period of light drift to reduce the imaging spatial positioning offset caused by thermal disturbance.

8. The image processing and analysis method for industrial nondestructive testing according to claim 6, characterized in that, The image processing and analysis steps performed on the detection image based on the dynamic correction input are as follows: Based on the time node information in the dynamic correction inlet, the detection screen is time-synchronized and responded to, and the acquisition sequence and refraction change rhythm are kept consistent. Based on the time synchronization results, backsampling and staggered exposure operations are alternately performed during thermal disturbance to form a sequence of image frames containing different refraction states; Using the image frame sequence obtained by alternating sampling, the light drift path in the continuous imaging scene is corrected at the pixel level and the real spatial position is restored based on the temporal information in the refraction offset map; A short sampling pause is set during the temperature drop process, and the position correction and optical distortion compensation of the imaging image are completed by coordinating and adjusting the sampling order before and after the sampling.

Images