Method and system for reflection suppression in aircraft engine inspection based on industrial endoscope

Abstract

This application provides a method and system for suppressing reflections during aircraft engine inspection based on an industrial endoscope, relating to the field of endoscopic inspection technology. The method includes: acquiring an inspection image and environmental features of the interior of an aircraft engine using an industrial endoscope; determining brightness homogenization parameters for the inspection image based on the environmental features; performing brightness homogenization processing on the inspection image according to the brightness homogenization parameters to obtain a homogenized image; detecting whether reflections exist in the homogenized image; if reflections are present, adjusting the exposure parameters of the industrial endoscope's camera; during the adjustment of the exposure parameters, monitoring the overall brightness of the homogenized image in real time; when the overall brightness is lower than a preset brightness threshold, stopping the adjustment of the exposure parameters and outputting the inspection image after reflection suppression. The technical effect of this application is: improved inspection image quality.

Description

Technical Field

[0001] This application relates to the field of endoscopic inspection technology, specifically to a method, system, equipment, and medium for suppressing reflections during aircraft engine inspection based on an industrial endoscope. Background Technology

[0002] In the maintenance and inspection of aircraft engines, industrial endoscopes are widely used as key non-destructive testing equipment for visual inspection of the engine's internal structure to detect potential faults such as cracks, wear, and foreign objects. However, due to the large number of metal surfaces inside the engine, and the fact that these surfaces have strong reflective properties due to precision machining, strong reflections are easily generated when the endoscope's light source shines on these surfaces. This leads to overexposed areas and uneven brightness distribution in the inspection images. This reflection phenomenon not only seriously affects the visual quality of the images, but more importantly, it can obscure or distort critical defect features, making it difficult for inspectors to accurately identify and locate potential fault points, thus affecting the reliability and accuracy of the entire inspection process.

[0003] Currently, image post-processing algorithms are mainly used to address glare issues. Typical methods include histogram equalization, gamma correction, and brightness adjustment—traditional image enhancement techniques. While these methods improve the visual effects of glare by globally adjusting the brightness distribution of the image, thus alleviating overly bright areas and improving overall visual appeal to some extent, they lack effective quality control mechanisms during glare suppression. Often, in pursuit of glare suppression, they neglect the protection of overall image quality, easily leading to over-processing that results in excessively low overall image brightness and a decline in image quality. Summary of the Invention

[0004] This application provides a method, system, device, and medium for suppressing reflections during aircraft engine inspection based on an industrial endoscope, which improves the quality of inspection images.

[0005] In a first aspect, this application provides a method for suppressing reflections during aircraft engine inspection based on an industrial endoscope. The method includes: acquiring an inspection image of the interior of an aircraft engine through an industrial endoscope, and acquiring environmental features of the interior of the aircraft engine; determining a brightness uniformization parameter for the inspection image based on the environmental features; performing brightness uniformization processing on the inspection image according to the brightness uniformization parameter to obtain a uniform image; detecting whether there is a reflection in the uniform image; if there is a reflection in the uniform image, adjusting the exposure parameters of the camera of the industrial endoscope; during the adjustment of the exposure parameters, monitoring the overall brightness of the uniform image in real time; when the overall brightness is lower than a preset brightness threshold, stopping the adjustment of the exposure parameters and outputting the inspection image after reflection suppression.

[0006] By employing the aforementioned technical solution, and intelligently determining brightness uniformity parameters through acquiring and analyzing the internal environmental characteristics of the engine, targeted reflection suppression processing is achieved, effectively improving the accuracy and applicability of reflection suppression. This method uses a strategy of first performing brightness uniformity processing and then reflection detection, enabling more accurate identification of residual reflections while initially improving image quality, providing a reliable basis for subsequent exposure parameter adjustments. Further suppression of reflections is achieved by dynamically adjusting the exposure parameters of the industrial endoscope camera. This method combines image processing with optimized hardware parameter adjustments, achieving a hardware-software synergistic reflection suppression effect. More importantly, by real-time monitoring of the overall brightness of the uniformized image and setting a preset brightness threshold as the adjustment termination condition, this method establishes an effective quality protection mechanism. This ensures that while pursuing reflection suppression effects, the overall image brightness is not excessively reduced, avoiding the over-processing problem common in traditional methods. This guarantees that the output image after reflection suppression possesses both good reflection suppression effects and sufficient image quality, thus improving the overall quality of the detected image.

[0007] Secondly, this application provides a reflection suppression system for aircraft engine inspection based on an industrial endoscope, the system comprising: an acquisition module, a determination module, a processing module, a detection module, and an adjustment module; wherein, The acquisition module is used to acquire an inspection image of the interior of an aircraft engine through an industrial endoscope, and to acquire the environmental characteristics inside the aircraft engine; the determination module is used to determine the brightness uniformity parameters of the inspection image based on the environmental characteristics; the processing module is used to perform brightness uniformity processing on the inspection image according to the brightness uniformity parameters to obtain a uniform image; the detection module is used to detect whether there is a reflection in the uniform image, and if there is a reflection, adjust the exposure parameters of the camera of the industrial endoscope; the adjustment module is used to monitor the overall brightness of the uniform image in real time during the adjustment of the exposure parameters, and when the overall brightness is lower than a preset brightness threshold, stop adjusting the exposure parameters and output the inspection image after reflection suppression.

[0008] Thirdly, this application provides an electronic device that adopts the following technical solution: it includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to make the electronic device execute a computer program such as any of the above-mentioned methods for suppressing reflections in aircraft engines based on industrial endoscopes.

[0009] Fourthly, this application provides a computer-readable storage medium that employs the following technical solution: storing a computer program capable of being loaded by a processor and executing any of the above-mentioned methods for suppressing reflections in aircraft engines based on industrial endoscopes.

[0010] In summary, this application includes at least one of the following beneficial technical effects: By acquiring and analyzing the internal environmental characteristics of the engine to intelligently determine brightness homogenization parameters, targeted reflection suppression processing is achieved, effectively improving the accuracy and applicability of reflection suppression. This method employs a strategy of first performing brightness homogenization processing and then reflection detection, enabling more accurate identification of residual reflections while initially improving image quality, providing a reliable basis for subsequent exposure parameter adjustments. Further suppression of reflections is achieved by dynamically adjusting the exposure parameters of the industrial endoscope camera. This method combines image processing with optimized hardware parameter adjustments, achieving a hardware-software collaborative reflection suppression effect. More importantly, by real-time monitoring of the overall brightness of the homogenized image and setting a preset brightness threshold as the adjustment termination condition, an effective quality protection mechanism is established. This ensures that while pursuing reflection suppression effects, the overall image brightness is not excessively reduced, avoiding the over-processing problem common in traditional methods. This guarantees that the output image after reflection suppression possesses both good reflection suppression effects and sufficient image quality, thus improving the overall quality of the detected image. Attached Figure Description

[0011] Figure 1 This is a schematic flowchart of a method for suppressing reflection during aircraft engine inspection based on an industrial endoscope, provided in an embodiment of this application. Figure 2 This is a schematic diagram of an intelligent anti-reflection system architecture for engine internal detection provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an aircraft engine inspection reflection suppression system based on an industrial endoscope provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0012] Explanation of reference numerals in the attached figures: 1000, electronic device; 1001, processor; 1002, communication bus; 1003, user interface; 1004, network interface; 1005, memory. Detailed Implementation

[0013] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0014] In the description of the embodiments in this application, words such as "illustrative," "for example," or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "illustrative," "for example," or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Rather, the use of words such as "illustrative," "for example," or "for example" is intended to present the relevant concepts in a specific manner.

[0015] Figure 1 This is a schematic flowchart illustrating a method for suppressing reflections during aircraft engine inspection based on an industrial endoscope, as provided in an embodiment of this application. Figure 1 As shown, the method includes S101-S105: S101, acquire images of the interior of an aircraft engine using an industrial endoscope, and acquire environmental characteristics of the interior of the aircraft engine.

[0016] The system first acquires inspection images of the aircraft engine's interior using a high-resolution CCD camera on an industrial endoscope. This is because the internal structure of an aircraft engine is complex, containing numerous irregular blades, combustion chamber walls, turbine disks, and other metal components. These components are prone to varying degrees of reflection under endoscopic light. Traditional reflection suppression methods, lacking understanding of the specific inspection environment, often use fixed parameters, resulting in significant differences in effectiveness under different conditions. Therefore, this solution, while acquiring inspection images, also needs to obtain the environmental characteristics inside the aircraft engine to provide fundamental data support for subsequent adaptive reflection suppression.

[0017] In practical implementation, the industrial endoscope system is equipped with an environmental perception module. This module uses multiple sensors and algorithms to work together to acquire environmental characteristics. First, the system uses a spectral analyzer to identify the material of the components in the detection area and obtain component material information. Here, component material refers to the material type of different components inside the engine, such as titanium alloy blades, stainless steel combustion chambers, and nickel-based alloy turbine disks. Each material has a different reflectance coefficient. The reflectance coefficient is a quantitative indicator of the strength of a material surface's ability to reflect incident light. The higher the value, the stronger the reflectance and the easier it is to produce reflective phenomena.

[0018] Simultaneously, the system's built-in angle sensor detects the illumination angle of the endoscope's light source relative to the surface being inspected in real time. It then uses image geometric analysis algorithms to calculate the incident angle and potential reflection angle of the light on the component surface. This is because, according to optical principles, different incident angles result in varying intensities of specular reflection, which is the primary cause of strong glare. Furthermore, the system analyzes the cavity structure of the current inspection area using a 3D reconstruction algorithm. The cavity structure refers to the geometric features of the engine's internal space, including parameters such as wall curvature, opening size, and depth. These structural features affect the light propagation path and multiple reflections.

[0019] By acquiring the aforementioned environmental features, the system can establish a complete model of the current detection environment, providing accurate input data for determining subsequent brightness uniformity parameters. This environment-aware approach allows the reflection suppression algorithm to adaptively adjust according to the specific detection scenario, rather than simply applying fixed processing parameters. The acquired detection images are typically 24-bit RGB color images with a resolution of no less than 1920×1080 pixels to ensure clear display of minute defects and structural details inside the engine. The environmental feature data is stored in a structured data format, including material reflectance coefficient values, angle parameters, and geometric parameters. This data will be used in the next step to calculate targeted brightness uniformity parameters, thereby achieving a more accurate and effective reflection suppression effect, ultimately improving the accuracy and reliability of aircraft engine inspection.

[0020] S102, Determine the brightness uniformity parameters of the detected image based on environmental characteristics.

[0021] In practice, the system first extracts the reflectance coefficient of each component from environmental characteristic data. The reflectance coefficient is typically between 0 and 1, with the value closer to 1 indicating stronger reflectivity. For example, the reflectance coefficient of a polished titanium alloy blade may reach 0.8-0.9, while the reflectance coefficient of a surface-treated combustion chamber wall may only be 0.3-0.5. Next, the system uses the light source illumination angle information to determine the incident angle and reflection angle of light on the surface of each component through optical calculation formulas. The incident angle refers to the angle between the light ray and the surface normal, while the reflection angle is calculated according to the law of reflection. When the incident angle is close to 0 degrees, the strongest specular reflection effect is produced.

[0022] Subsequently, the system constructs a three-dimensional light propagation model based on the chamber structure information. This model considers the impact of the complex geometry inside the engine on light propagation, including direct propagation, wall reflection, and multiple scattering. Scattering characteristics refer to the physical phenomenon of light changing direction in a non-smooth surface or medium. In engine detection, scattered light can illuminate shadowed areas, but it can also create secondary reflections in some areas. The system simulates the propagation path of light within the chamber using a Monte Carlo ray tracing algorithm, calculating the intensity distribution of each ray reaching the image sensor after multiple reflections and scatterings.

[0023] Based on the light propagation path analysis, the system divides the detected image into several pixel regions, typically using 8×8 or 16×16 pixel square blocks as the basic processing unit. For each pixel region, the system finds the corresponding reflectance coefficient based on its location on the surface of the component, and then calculates the basic reflectance value of that region, which is the reflectance value that the region should have under standard illumination conditions. Next, the system calculates the reflection intensity coefficient by combining the incident angle and the reflection angle. This coefficient reflects the modulation effect of geometric optics on the reflection intensity. When the light is nearly perpendicular to the incident angle, the reflection intensity coefficient is close to 1; when the incident angle is large, the reflection intensity coefficient decreases significantly.

[0024] Simultaneously, the system calculates the intensity of scattered light received by each pixel region based on scattering characteristics. Scattered light intensity is typically much weaker than directly reflected light, but in some occluded areas, scattered light may be the primary light source. The system arithmetically multiplies the base reflected brightness value by the reflected intensity coefficient to generate a reflected modulation brightness value that considers geometric optical effects. Then, it arithmetically adds the reflected modulation brightness value to the scattered light intensity to obtain the expected brightness distribution value for each pixel region. This expected brightness distribution value represents the theoretically expected brightness level of the region under the current environmental characteristics.

[0025] The system combines the expected brightness distribution values ​​of all pixel regions to form a theoretical brightness distribution map of the entire detected image, and then compares it with a preset standard brightness value. The preset standard brightness value is a target brightness level determined based on the optimal visual effect of engine detection, typically set within a grayscale range of 128-160 to ensure that details are visible without excessive glare. The system calculates the difference between the expected brightness distribution value and the preset standard brightness value for each region, resulting in a brightness difference distribution map. Positive values ​​indicate that the region is too bright and needs to be reduced, while negative values ​​indicate that the region is too dark and needs to be increased.

[0026] Finally, based on the statistical characteristics of the brightness difference distribution, including parameters such as mean, variance, and peak value, the system calculates brightness homogenization parameters for the current detection environment. These parameters include a brightness homogenization coefficient to control the intensity of brightness adjustment, and regional processing weights to achieve differentiated processing for different areas. Through this environment-feature-based parameter determination method, the system can generate the most suitable brightness homogenization parameters for each detected image, thereby achieving more precise brightness adjustment in subsequent brightness homogenization processing. This effectively reduces inconsistencies in processing results caused by environmental differences, laying a solid foundation for high-quality reflection suppression.

[0027] Based on the above embodiments, as an optional implementation, in S102, the environmental characteristics include component material, light source illumination angle, and chamber structure. Determining the brightness uniformity parameters of the detected image based on these environmental characteristics specifically includes S21-S24: S21, obtain the reflection coefficient of the component material.

[0028] The system retrieves the reflectance coefficients of various component materials from a pre-built material database, which includes standard reflectance coefficient values ​​for common engine internal materials such as titanium alloys, stainless steel, and ceramic coatings. For components made of new materials or with special surface treatments, the system measures their reflectance coefficients in real time using a spectral analyzer. The measurement process involves illuminating the components with a standard light source from multiple angles, and the reflectance characteristics of the material are determined by comparing the ratio of incident light intensity to reflected light intensity, thereby establishing an accurate optical model foundation.

[0029] S22, calculate the incident angle and reflection angle of the light on the surface of the component material based on the angle of illumination from the light source.

[0030] The system calculates the incident angle of light reaching the surface of each component based on the specific location and emission direction of the industrial endoscope's light source, combined with the geometric coordinates of each component in three-dimensional space. The incident angle is the angle between the light direction and the surface normal vector, which the system quickly calculates through vector operations. Subsequently, based on the law of optical reflection—that the incident angle equals the reflection angle—the corresponding reflection angle is calculated. These angular parameters directly affect the intensity distribution of the reflected light, providing a geometrical optical basis for subsequent brightness prediction.

[0031] S23, Based on the chamber structure, determine the light propagation path and scattering characteristics.

[0032] The system analyzes the three-dimensional structural model of the engine chamber, identifying geometric elements that may affect light propagation, such as curved pipes, protruding structures, and inner wall roughness. The system employs a ray tracing algorithm to simulate the propagation path of light within the chamber, calculating the light intensity attenuation for different propagation paths, including direct illumination, single reflection, and multiple reflections. The determination of scattering characteristics considers the impact of surface roughness on light propagation. Rough surfaces produce diffuse reflection, causing light to scatter in multiple directions; the system uses a Lambertian scattering model to describe this scattering behavior.

[0033] S24, based on the light propagation path, combined with the reflection coefficient, incident angle, reflection angle and scattering characteristics, calculates the expected value of the brightness distribution of each region in the detected image.

[0034] The system comprehensively calculates the aforementioned parameters. For each pixel region in the detected image, the system first determines the corresponding reflectance coefficient based on its physical surface location. Then, it calculates the geometrical optical contribution by combining the incident angle and reflection angle at that location, and finally superimposes the contribution of scattered light to obtain the theoretical brightness value of that region. This calculation process uses photometric formulas and comprehensively considers multiple influencing factors such as light source intensity, propagation distance, reflection efficiency, and scattering intensity to generate a complete brightness distribution prediction map.

[0035] Based on the above embodiments, as an optional implementation, in S24, the calculation of the expected brightness distribution value of each region in the detected image based on the light propagation path, combined with the reflection coefficient, incident angle, reflection angle, and scattering characteristics, specifically includes S241-S246: S241, the detection image is divided into multiple pixel regions, and the position of the component surface corresponding to each pixel region is determined according to the light propagation path.

[0036] The system divides the entire detection image into regular pixel regions according to a preset grid density, typically using 16×16 or 32×32 pixel regions. This division ensures computational accuracy without excessively increasing the computational burden. Utilizing the established 3D geometric model and camera calibration parameters, the system transforms the 2D image coordinates of each pixel region into corresponding 3D physical space coordinates through perspective projection transformation, thereby determining the specific component surface position corresponding to that region. This coordinate mapping relationship is the foundation of optical computation, ensuring an accurate correspondence between theoretical calculations and actual imaging.

[0037] S242, calculate the basic reflectance value of each pixel area based on the reflectance coefficient of the component surface position.

[0038] The system extracts the corresponding reflection coefficient values ​​from the material database based on the component surface position corresponding to each pixel region, and considers the influence of surface treatment state on the reflection coefficient correction. The system uses the Lambertian reflection model to calculate the basic reflection brightness value. The calculation formula is: Basic reflection brightness value = Incident light intensity × Reflection coefficient × Cosine factor, where the cosine factor is determined by the angle between the surface normal vector and the light source direction. This calculation process establishes the basic optical relationship from light source characteristics to surface reflection response.

[0039] S243 calculates the reflection intensity coefficient of each pixel region by combining the incident angle and the reflection angle.

[0040] The system comprehensively considers the modulation effects of the incident angle and reflection angle on the intensity of reflected light to calculate the reflection intensity coefficient. The reflection efficiency is highest when the incident angle is close to perpendicular; as the incident angle increases, the reflection efficiency gradually decreases. The system uses a simplified form of the Fresnel reflection formula to describe this angle dependence. The calculation of the reflection intensity coefficient also considers the influence of the viewing angle, i.e., the geometric factors affecting whether the reflected light can effectively reach the camera sensor, and determines the proportion of effectively reflected light through solid angle calculation.

[0041] S244, calculate the intensity of scattered light received by each pixel region based on the scattering characteristics.

[0042] The system calculates the contribution of indirect scattered light to each pixel region based on the scattering characteristics determined by surface roughness and cavity geometry. Scattered light intensity includes diffuse reflection from the surrounding surface and scattered light after multiple reflections. The system uses Monte Carlo ray tracing to simulate the multiple scattering process of light in complex geometric environments, counts the number of scattered photons reaching each pixel region, and converts them into corresponding light intensity values. The calculation of scattered light is the most complex part of the entire optical model, as it determines the brightness levels of shadowed and indirectly illuminated areas in the image.

[0043] S245, the basic reflective brightness value is arithmetically multiplied by the reflective intensity coefficient to generate the reflective modulation brightness value, and the reflective modulation brightness value is arithmetically added to the scattered light intensity to generate the expected brightness distribution value of each pixel area.

[0044] The system performs mathematical synthesis of optical contributions. First, it performs an arithmetic multiplication of the base reflectance brightness value and the reflectance intensity coefficient to obtain a geometrically corrected reflectance modulation brightness value, which represents the contribution of direct reflected light to the brightness of the area. Then, the system performs an arithmetic addition of the reflectance modulation brightness value and the scattered light intensity to superimpose the effects of direct and indirect illumination, generating the expected comprehensive brightness distribution value for each pixel area. This step-by-step calculation and superposition method ensures the accurate quantification and reasonable combination of contributions from different optical mechanisms.

[0045] S246, combining the expected brightness distribution values ​​of each pixel region, generates the expected brightness distribution values ​​of each region in the detected image.

[0046] The system performs statistical analysis and region aggregation on the expected brightness distribution values ​​of all pixel regions, merging adjacent pixel regions with similar brightness characteristics into larger processing areas to generate a regional brightness distribution map suitable for actual brightness uniformity processing. This aggregation process employs region growing and mean clustering algorithms, which not only preserves the main characteristics of the brightness distribution but also simplifies the complexity of subsequent processing. The resulting expected brightness distribution values ​​provide a scientifically accurate theoretical basis for determining the overall brightness uniformity parameters, enabling the system to implement precise and personalized processing strategies for the optical characteristics of different regions.

[0047] S25, calculate the brightness difference between the expected brightness distribution value and the preset standard brightness value, and determine the brightness uniformity parameters of the detected image based on the brightness difference.

[0048] The system compares the expected brightness distribution of each region with a preset standard brightness value set for optimal detection results, calculating the brightness difference distribution. A positive brightness difference indicates that the region is expected to be too bright and requires reduced processing intensity, while a negative brightness difference indicates that the region is expected to be too dark and requires increased processing intensity. Based on the statistical distribution characteristics of these differences, including parameters such as the mean, standard deviation, and extreme values, the system uses an optimization algorithm to calculate the most suitable brightness homogenization parameters for the current environment, including adjustment coefficients and processing weights for each region. This achieves targeted brightness homogenization processing, ensuring that the final homogenized image eliminates brightness unevenness while preserving image detail and contrast characteristics.

[0049] S103, perform brightness homogenization processing on the detected image according to the brightness homogenization parameters to obtain a homogenized image.

[0050] In practice, the system first preprocesses the detection image, using a Gaussian filter to remove noise. The kernel size of this filter is typically set to 5×5 or 7×7 pixels, and the standard deviation parameter is adaptively adjusted according to the noise level of the image, generally between 0.8 and 1.5. Gaussian filtering effectively smooths random noise and electronic noise generated during the acquisition process while preserving the main structural information of the image. Simultaneously with noise removal, the system uses the Laplacian or Sobel operator for edge enhancement. This is because defects inside the engine often manifest as subtle edge changes, such as cracks, wear marks, and corrosion spots; edge enhancement can highlight these key detection target features.

[0051] After preprocessing, the system begins the core brightness uniformity adjustment process. The system divides the detected image into multiple pixel regions in the same manner as in step S102, and assigns a corresponding brightness uniformity coefficient to each region. For the first region where the brightness value exceeds a preset upper limit threshold, the system identifies these potentially overbright areas. The preset upper limit threshold is typically set to a grayscale value of 200-220; areas exceeding this threshold often exhibit strong reflections or overexposure. The system reduces the brightness of the first region according to the corresponding brightness uniformity coefficient. The specific adjustment formula is: Adjusted brightness value = Original brightness value × (1 - Brightness uniformity coefficient × Adjustment intensity factor), where the adjustment intensity factor is dynamically calculated based on the difference between the original brightness value and the preset upper limit threshold. The larger the difference, the higher the adjustment intensity, thus achieving a gradual brightness attenuation effect.

[0052] Simultaneously, the system identifies a second region whose brightness value is below a preset lower threshold. This preset lower threshold is typically set to a grayscale value of 30-50. These regions are often shadow areas or areas with insufficient lighting. The system then performs brightness enhancement processing on the second region according to the corresponding brightness uniformity coefficient. The adjustment formula is: Adjusted brightness value = Original brightness value + (Standard brightness value - Original brightness value) × Brightness uniformity coefficient × Enhancement intensity factor. Here, the standard brightness value is the target brightness level determined in step S102, and the enhancement intensity factor ensures that the brightness enhancement process does not produce excessive enhancement leading to noise amplification.

[0053] For other areas with brightness values ​​within the normal range, the system employs a more refined adjustment strategy, fine-tuning based on the brightness gradient relationship between this area and its surrounding areas to ensure a smooth overall brightness transition. This regionally differentiated processing approach can reduce the brightness of overly bright areas while moderately improving the visibility of overly dark areas, avoiding the overall brightness reduction or contrast loss problems that are easily caused by traditional global adjustment methods.

[0054] After pixel-level brightness adjustment, the system smooths the adjusted image using a bilateral filtering algorithm to generate the final homogenized image. Bilateral filtering is an edge-preserving smoothing algorithm that smooths image noise while maintaining important edge information. This is particularly important for engine defect detection, as the edge features of defects are key to determining their type and severity. The spatial filtering parameters of bilateral filtering are typically set to 9-15 pixels, and the color similarity parameter is set to 50-80. These parameters ensure an optimal balance between smoothing effect and edge preservation.

[0055] Through the aforementioned brightness homogenization process, the homogenized image generated by the system exhibits significantly improved visual quality and analytical applicability. The brightness distribution in the homogenized image is more balanced, glare in overly bright areas is effectively suppressed, and detail information in overly dark areas is moderately enhanced, resulting in improved overall image contrast and clarity. This processing effect creates a more ideal analytical foundation for subsequent reflection detection, enabling the system to more accurately identify true reflective areas while avoiding misinterpreting normal brightness changes as reflections, thereby improving the accuracy and reliability of the entire reflection suppression process.

[0056] Based on the above embodiments, as an optional implementation, in S103, performing brightness homogenization processing on the detected image according to the brightness homogenization parameters to obtain a homogenized image specifically includes S31-S33: S31, preprocess the detected image to remove noise and enhance edge information.

[0057] The system first performs preprocessing on the detected image to optimize the foundation for subsequent processing. Noise removal employs an adaptive median filter, which dynamically adjusts the filter window size based on the statistical characteristics of local pixels. This effectively suppresses salt-and-pepper noise and Gaussian noise while avoiding excessive smoothing of image edges. Edge enhancement uses a Laplacian sharpening operator convolved with the original image. The intensity coefficient of the enhancement operator is adaptively adjusted based on the overall contrast of the image, typically set between 0.2 and 0.5. This processing highlights the boundary features of internal engine defects, providing clearer feature information for subsequent detection and analysis.

[0058] S32, adjust the brightness value of each region in the detected image according to the brightness uniformity coefficient in the brightness uniformity parameter; wherein, for the first region whose brightness value is higher than the preset upper limit threshold, reduce the brightness value of the first region according to the brightness uniformity coefficient; for the second region whose brightness value is lower than the preset lower limit threshold, increase the brightness value of the second region according to the brightness uniformity coefficient.

[0059] The system implements differentiated brightness adjustment strategies for each region based on the brightness uniformity coefficient determined in step S102. For the first region where the brightness value exceeds the preset upper threshold, these regions typically correspond to highly reflective surfaces or overexposed areas. The system uses a non-linear attenuation function to reduce brightness, with the adjustment formula being: Adjusted brightness = Original brightness × (1 - Brightness uniformity coefficient × Attenuation intensity factor), where the attenuation intensity factor is dynamically calculated based on the degree to which the original brightness exceeds the preset upper threshold; the greater the exceedance, the stronger the attenuation. For the second region where the brightness value is below the preset lower threshold, these regions are typically located in shadows or underlit areas. The system uses a progressive enhancement function to increase brightness, with the adjustment formula being: Adjusted brightness = Original brightness + (Target brightness - Original brightness) × Brightness uniformity coefficient × Enhancement intensity factor. This bidirectional adjustment mechanism ensures that the image brightness distribution converges towards the preset ideal level.

[0060] S33, smooth the adjusted detection image to generate a uniform image.

[0061] The system smooths the brightness-adjusted image to eliminate potential discontinuities and blockiness during the adjustment process. The smoothing process employs an edge-preserving filter, specifically a bilateral filtering algorithm. This algorithm considers pixel similarity while performing spatial smoothing. The spatial filtering parameters are set to a range of 11-15 pixels, and the intensity similarity parameter is set to 60-80. This parameter configuration effectively smooths brightness transition areas while preserving important edge information. The system also incorporates a gradient constraint mechanism in the smoothing process, applying additional gradient processing to areas with excessive brightness gradients to ensure a natural and smooth brightness transition between adjacent areas. The resulting uniform image exhibits good brightness consistency while preserving important details from the original image, providing a high-quality image foundation for subsequent reflection detection and defect identification.

[0062] S104, detect whether there is a reflection in the homogenized image. If there is a reflection in the homogenized image, adjust the exposure parameters of the industrial endoscope camera.

[0063] In practice, the system first calculates the brightness value of each pixel in the homogenized image. For RGB color images, the system uses a weighted average method to convert the RGB three-channel values ​​into grayscale brightness values. The conversion formula is: Brightness value = 0.299×R + 0.587×G + 0.114×B. This formula takes into account the differences in human eye sensitivity to different colors of light, with the green component having the highest weight and the blue component having the lowest weight. The system then sets a preset reflection detection threshold. This threshold is typically set to 1.8 to 2.2 times the average brightness of the overall image after brightness homogenization, or to a grayscale range of 230-245 in absolute value. The determination of this threshold must ensure that it can identify genuine reflection phenomena while avoiding misjudging normal bright areas as reflections.

[0064] The system scans the entire homogenized image, identifying all pixels whose brightness exceeds a preset reflectivity threshold and marking these pixels as candidate reflective pixels. Due to noise and isolated bright spots, the high brightness of a single pixel does not necessarily represent true reflection. Therefore, the system performs connectivity analysis on the candidate reflective pixels, employing an eight-connected region growing algorithm to aggregate spatially adjacent candidate reflective pixels into continuous regions, forming candidate reflective regions. Connectivity analysis is a region segmentation technique in image processing that groups spatially continuous pixels with similar features into the same region, thereby achieving effective separation of different objects in the image.

[0065] For each candidate reflective region, the system calculates its area and average brightness value. The area is determined by counting the total number of pixels contained in the region, and the average brightness value is the arithmetic mean of the brightness values ​​of all pixels in the region. The system sets preset area thresholds and preset brightness thresholds as the final criteria for determining reflective regions. The preset area threshold is typically set to 50-200 pixels because genuine reflections often cover a certain continuous area, while noise or isolated bright spots usually have a very small area. The preset brightness threshold is set to a grayscale value of 240-250 to ensure that only areas with truly strong reflective characteristics are identified as reflective regions.

[0066] When a candidate reflective region simultaneously meets both conditions—an area greater than a preset area threshold and an average brightness value greater than a preset brightness threshold—the system officially identifies the candidate reflective region as a reflective region. This dual-determination mechanism effectively filters out false detections caused by noise or localized highlights, improving the accuracy and reliability of reflective detection. When the system detects at least one reflective region, it determines that there is indeed a reflective phenomenon in the homogenized image that requires further processing.

[0067] Once a reflection is detected, the system immediately initiates the camera exposure parameter adjustment procedure. The system first acquires the location information and reflection intensity of all reflective areas. Location information includes geometric parameters such as the center coordinates and bounding box range of the reflective area. Reflection intensity is quantified by the difference between the average brightness value of the area and a preset reflection threshold. Based on the magnitude of the reflection intensity, the system calculates the corresponding exposure parameter adjustments, including exposure time and gain adjustments. These adjustments are calculated using a non-linear function; for areas with higher reflection intensity, the adjustment amount is increased accordingly to ensure effective suppression of strong reflections.

[0068] The system adjusts the exposure time and gain settings of the industrial endoscope camera based on the calculated exposure parameters via the camera control interface. The exposure time adjustment is typically within 70%-90% of the original setting, and the gain adjustment is typically within 80%-95% of the original setting. This moderate adjustment reduces the brightness of reflective areas without significantly impacting the overall image quality. After adjustment, the system acquires a new detection image and repeats steps S102 and S103 of the brightness homogenization process on this new image to obtain a new homogenized image.

[0069] The system re-detects the original reflective areas in the new homogenized image and calculates the residual reflective intensity of these areas. Residual reflective intensity refers to the degree to which the brightness value of the area still exceeds a preset reflective threshold after exposure parameter adjustments. The system determines whether the residual reflective intensity is lower than a preset reflective suppression threshold, typically set to a grayscale difference of 5-15, representing an acceptable level of residual reflective intensity. If the residual reflective intensity is still not lower than the preset reflective suppression threshold, the system continues to calculate new exposure parameter adjustments based on the current residual reflective intensity, iteratively adjusting until the residual reflective intensity decreases to an acceptable level, or the overall image brightness decreases below a preset brightness threshold.

[0070] Through this precise reflection detection and adaptive exposure adjustment mechanism, the system can effectively suppress strong local reflections while maintaining the overall image quality, providing a clearer and more uniform image basis for subsequent engine defect detection and analysis, thereby significantly improving the accuracy and reliability of detection.

[0071] Based on the above embodiments, as an optional implementation, in S104, detecting whether there is a reflection in the homogenized image specifically includes S41-S46: S41, calculate the brightness value of each pixel in the homogenized image.

[0072] The system calculates the corresponding luminance value for each pixel in the homogenized image. For color images, it uses the standard luminance conversion formula Y=0.299R+0.587G+0.114B, while for grayscale images, it directly extracts the grayscale value of the pixel. The system also establishes an index mapping table between pixel coordinates and luminance values, providing a foundation for fast access in subsequent spatial analysis and region processing. This preprocessing ensures the efficiency and accuracy of subsequent calculations.

[0073] S42 identifies candidate reflective pixels whose brightness values ​​exceed a preset reflectivity threshold.

[0074] The system sets a preset reflectivity threshold as a preliminary screening criterion. This threshold is typically set to 2.0 to 2.5 times the average brightness of the overall image after brightness homogenization, or, depending on the specific application scenario, an absolute grayscale value of 220-240. The system scans the entire image and marks all pixels with brightness values ​​exceeding this threshold as candidate reflective pixels. This threshold screening method can quickly eliminate most pixels with normal brightness, significantly improving the efficiency of subsequent processing.

[0075] S43, perform connectivity analysis on candidate reflective pixels, and aggregate interconnected candidate reflective pixels into candidate reflective regions.

[0076] The system performs spatial connectivity analysis on scattered candidate reflective pixels and employs an eight-connected region growing algorithm to aggregate spatially adjacent candidate reflective pixels into contiguous regions. Connectivity analysis is a fundamental algorithm in image processing; it examines the connectivity between each pixel and its eight neighboring pixels, grouping spatially contiguous pixels with similar characteristics into the same region. This aggregation process can uniformly identify multiple bright pixels generated by the same reflective light source as a single candidate reflective region, while filtering out isolated noise points.

[0077] S44, calculate the area and average brightness value of each candidate reflective region.

[0078] The system calculates the geometric and optical characteristic parameters of each candidate reflective region. Area is calculated by counting the total number of pixels within the region, while the average brightness value is obtained by arithmetically averaging the brightness values ​​of all pixels within the region. The system also calculates auxiliary parameters such as the region's geometric center, boundary coordinates, and aspect ratio, which provide data support for subsequent shape analysis and reflective feature verification.

[0079] S45, when the area of ​​the candidate reflective region is greater than the preset area threshold and the average brightness value is greater than the preset brightness threshold, the candidate reflective region is determined as a reflective region.

[0080] The system employs a dual-criteria approach to determine the true reflective area. The preset area threshold is typically set between 80 and 200 pixels, based on the statistical characteristics of real reflections. Areas that are too small are often noise or isolated bright spots, while true reflections usually cover a continuous area. The preset brightness threshold is set to a grayscale value of 235-250, ensuring that only areas with strong optical characteristics are identified as reflective. When a candidate reflective area simultaneously meets both conditions—an area greater than the preset area threshold and an average brightness value greater than the preset brightness threshold—the system officially classifies it as a reflective area. This rigorous dual-verification mechanism effectively improves the accuracy and reliability of reflection detection.

[0081] S46, when at least one reflective area is detected, it is determined that there is a reflective phenomenon in the homogenized image.

[0082] The system counts the number of verified reflective areas. When at least one reflective area that meets the criteria is detected, the system determines that there is indeed a reflective phenomenon in the current homogenized image that requires further processing. The system simultaneously records detailed information such as the location, size, and intensity of all reflective areas, providing precise target location and intensity assessment data for subsequent exposure parameter adjustments. This comprehensive reflective phenomenon detection and quantitative analysis lays a solid foundation for the intelligent control of the entire reflective suppression process.

[0083] Based on the above embodiments, as an optional implementation, in S104, adjusting the exposure parameters of the camera of the industrial endoscope specifically includes S401-S406: S401, obtain the location information and reflection intensity of the reflective area in the homogenized image.

[0084] The system extracts detailed quantitative information from the reflective area detection results obtained in step S46. Location information includes the geometric center coordinates, boundary range, and spatial distribution characteristics of each reflective area. Reflective intensity is quantified by calculating the difference between the pixel brightness value within the reflective area and the average background brightness. The system establishes a reflective intensity grading system, dividing reflective intensity into three levels: mild (difference 50-100), moderate (difference 100-180), and severe (difference above 180). Different intensity levels correspond to different processing strategies. This quantitative analysis provides a precise data foundation for subsequent parameter adjustments.

[0085] S402, calculate the exposure parameter adjustment amount based on the reflection intensity. The exposure parameter adjustment amount includes the exposure time adjustment amount and the gain adjustment amount.

[0086] The system calculates exposure parameter adjustments using a piecewise linear function based on the reflectivity level. For mild reflectivity, the exposure time adjustment is typically 5%-8% of the current setting, and the gain adjustment is 3%-6% of the current setting. For moderate reflectivity, the exposure time adjustment increases to 8%-12%, and the gain adjustment increases to 6%-10%. For severe reflectivity, the exposure time adjustment reaches 12%-18%, and the gain adjustment reaches 10%-15%. The system also considers the spatial distribution characteristics of reflective areas. When reflective areas are too concentrated, the adjustment range is appropriately increased; when reflective areas are dispersed, the adjustment range is correspondingly decreased. This adaptive adjustment strategy ensures the targeted and effective processing results.

[0087] S403 adjusts the exposure parameters to reduce the camera's exposure time and gain.

[0088] The system applies the calculated adjustments to the camera's exposure control system. By reducing the exposure time to decrease the accumulation time of light and by reducing the gain to decrease the signal amplification, the combined effect of these two measures effectively reduces the brightness of overly bright areas in the image. The system employs a gradual adjustment method to avoid abrupt changes in parameters that could impact image quality. After each adjustment, the system waits for the camera parameters to stabilize before proceeding to the next step, ensuring the stability and controllability of the adjustment results.

[0089] S404: Obtain the new detection image after adjusting the exposure parameters, perform brightness homogenization processing on the new detection image, and obtain a new homogenized image.

[0090] The system re-acquires the detection image using the adjusted exposure parameters and performs complete brightness homogenization processing on the new detection image according to the procedure in step S103. The new homogenized image reflects the actual effect of the exposure parameter adjustment. The system evaluates the effectiveness of the adjustment by comparing the differences between the old and new images, and at the same time provides an updated data basis for the next round of reflection detection and parameter optimization.

[0091] S405, calculate the residual reflectivity of the reflective regions in the new homogenized image.

[0092] The system repeats steps S41 to S45 of the reflection detection process to calculate the residual reflection intensity of each reflective region in the new homogenized image. Residual reflection intensity refers to the intensity level of reflection that still exists after exposure parameter adjustments; it directly reflects the sufficiency of the current adjustment effect. The system evaluates the effect of a single adjustment by comparing the changes in reflection intensity before and after the adjustment, and optimizes subsequent adjustment strategies accordingly.

[0093] S406, determine whether the residual reflective intensity is lower than the preset reflective suppression threshold. If the residual reflective intensity is not lower than the preset reflective suppression threshold, then continue to adjust the exposure parameters based on the residual reflective intensity until the residual reflective intensity is lower than the preset reflective suppression threshold or the overall brightness is lower than the preset brightness threshold.

[0094] The system employs dual termination conditions to control the iterative adjustment process. A preset glare suppression threshold is typically set at 10%-20% of the original glare intensity. When the residual glare intensity drops below this threshold, it indicates that the glare suppression effect has reached the expected target. Simultaneously, the system continuously monitors the overall brightness level of the image. When the overall brightness drops below a preset brightness threshold, the system will stop adjusting even if glare suppression has not yet fully met the target to ensure basic image usability. This dual control mechanism establishes a dynamic balance between glare suppression effectiveness and image quality, ensuring that the final output image possesses both good glare suppression and sufficient brightness and detail, providing a high-quality image foundation for the accurate inspection of aircraft engines.

[0095] S105 monitors the overall brightness of the homogenized image in real time during the process of adjusting the exposure parameters. When the overall brightness is lower than the preset brightness threshold, it stops adjusting the exposure parameters and outputs the detection image after reflection suppression.

[0096] In practice, the system establishes a real-time image brightness monitoring mechanism. After each adjustment of exposure parameters, the system immediately calculates the overall brightness index of the newly homogenized image. The overall brightness calculation employs a weighted statistical method. The system divides the image into three distinct weighted regions: the central region, the edge region, and the corner region. Since key information in engine detection is typically located in the central region, the weighting coefficient for the central region is set to 0.6, for the edge region to 0.3, and for the corner region to 0.1. The system calculates the average brightness value for each region separately, and then performs a weighted sum according to the weighting coefficients to obtain the overall brightness value that truly reflects the brightness level of the effective information areas in the image.

[0097] The preset brightness threshold set by the system is a critical brightness level determined based on the actual needs of engine inspection. It is usually set to a grayscale value range of 80-100. The determination of this threshold takes into account several factors: First, the visual perception ability of the human eye. Images with a brightness level below this level will make it difficult for inspectors to observe. Second, the performance requirements of image processing algorithms. Images that are too dark will cause a significant decrease in the effectiveness of edge detection, texture analysis and other algorithms. Finally, the contrast requirements of defect features. Defects such as cracks and wear inside the engine often manifest as subtle grayscale changes, and sufficient background brightness is required to ensure that these changes can be clearly identified.

[0098] During the iterative adjustment of exposure parameters, the system employs a gradual adjustment strategy to avoid drastic changes in brightness. The magnitude of each adjustment is controlled within a reasonable range. The single adjustment of exposure time typically does not exceed 10%-15% of the current setting, and the single adjustment of gain typically does not exceed 8%-12% of the current setting. This small-amplitude gradual adjustment ensures both the continuity and stability of reflection suppression and provides sufficient reaction time for overall brightness monitoring.

[0099] After each adjustment, the system re-executes the complete image processing flow, including reacquiring the detection image, recalculating the brightness uniformity parameters, re-performing the brightness uniformity process, and recalculating the overall brightness value. This complete processing loop ensures the accuracy and timeliness of brightness monitoring, avoiding the risk of making judgments based on outdated data. Simultaneously, the system establishes a brightness change trend analysis mechanism, monitoring not only the current absolute value of the overall brightness but also the rate and direction of brightness change. When it detects an excessively rapid rate of brightness decrease, the system proactively reduces the magnitude of subsequent adjustments, achieving a smoother control process.

[0100] When the system detects that the overall brightness has dropped below the preset brightness threshold, it immediately initiates a stop adjustment procedure. This procedure locks the current exposure parameter settings to prevent further automatic adjustments. The system also records the reflection suppression status at this time, including the number, location, and intensity of remaining reflective areas, providing data support for subsequent quality assessment and parameter optimization. In some cases, even when the overall image brightness just reaches the preset brightness threshold, some weak residual reflections may still exist. The system will decide whether to accept this state based on a preset priority strategy. Typically, the system prioritizes ensuring the overall usability of the image while accepting slight residual reflections.

[0101] The final output image from the system, after reflection suppression, is a high-quality image processed through a complete optimization workflow. This image effectively suppresses the strong reflections in the original image while maintaining sufficient overall brightness to ensure the effectiveness of defect detection. The quality indicators of the output image include reflective area coverage (typically controlled below 5% of the original image), overall brightness level (maintained above a preset brightness threshold), and contrast retention (maintained above 85% of the original image). These indicators comprehensively ensure that the output image has both good visual effects and excellent analytical applicability.

[0102] Through this real-time monitoring and dynamic balancing control mechanism, the system achieves an intelligent trade-off between reflection suppression and image quality, avoiding the over-processing or under-processing problems common in traditional methods. This provides a stable and reliable image foundation for the accurate detection of aircraft engines, significantly improving the efficiency and accuracy of the entire detection process.

[0103] Based on the above embodiments, as an optional implementation method, the method further includes S201-S207: S201, The detection image is input into a pre-trained deep learning model, which is trained based on industrial endoscope detection images with reflective labels.

[0104] The pre-trained deep learning model is built upon a large-scale industrial endoscope inspection image dataset, containing over 100,000 images of engine interiors collected under various environmental conditions. Each image is precisely annotated by professional technicians, including pixel-level boundaries of reflective areas, reflectivity levels, and reflectivity type classifications. The model employs an improved U-Net architecture as its base network structure, which is particularly suitable for pixel-level image segmentation tasks. The encoder uses ResNet-50 as the backbone to extract multi-scale features, while the decoder fuses feature information from different resolutions through skip connections, achieving accurate pixel-level reflective area localization. The training process uses a combination of cross-entropy loss and Dice loss; the former optimizes classification accuracy, while the latter improves boundary segmentation precision. The learning rate is gradually reduced from 0.001 to 0.0001 using a cosine annealing strategy. After 200 training epochs, the model achieves a reflectivity detection accuracy of 95.8% and a recall of 94.2% on the validation set, demonstrating highly reliable reflectivity recognition capabilities.

[0105] The system inputs the detection image to be processed into a pre-trained deep learning model. The image first undergoes standardization preprocessing, normalizing pixel values ​​to the range of 0-1 and adjusting the image size to the model's required 512×512 pixel resolution. The model's input layer accepts three-channel color image data and achieves feature extraction and spatial dimensionality reduction through a combination of multiple convolutional and pooling layers. This end-to-end processing avoids the complex manual feature design of traditional methods and can automatically learn deep feature representations of reflective phenomena.

[0106] S202 extracts features from the detected image using a deep learning model. These features include brightness gradient features, texture features, and edge features.

[0107] Deep learning models automatically extract multi-dimensional feature information from detected images through their internal convolutional neural network structure. Brightness gradient features are extracted through shallow convolutional kernels, similar to the Sobel operator, which can detect areas of drastic brightness changes in the image. Reflective areas typically exhibit a significant brightness jump compared to their surroundings. Texture features are extracted by mid-level convolutional layers, learning combinations of convolutional kernels at different scales to identify surface texture patterns. Reflections often disrupt the original texture continuity, forming unique texture discontinuity features. Edge features are extracted by deep networks, identifying the geometric boundary features of reflective areas through cascaded processing of multiple convolutional layers. The combination of these features provides a rich information foundation for the accurate localization of reflective areas.

[0108] S203. Based on the features, generate a reflectivity probability map, where each pixel in the reflectivity probability map corresponds to a reflectivity probability value.

[0109] The model inputs the extracted multi-level features into the final classification layer, generating a reflectivity probability map using the Sigmoid activation function. The reflectivity probability map is a single-channel image of the same size as the original image, where each pixel's value ranges from 0 to 1, representing the probability of reflection at that location. Values ​​closer to 1 indicate a higher probability of reflection, while values ​​closer to 0 indicate a lower probability. By learning from a large number of reflectivity samples in the training data, the model can accurately predict the reflectivity probability of each pixel. This pixel-level probability prediction provides detailed guidance for subsequent precise processing.

[0110] S204, aggregate pixels with a reflectivity probability value greater than a preset probability threshold in the reflectivity probability map into potential reflectivity areas.

[0111] The system sets a preset probability threshold, typically between 0.7 and 0.8. Pixels with probability values ​​exceeding this threshold in the reflectivity probability map are identified as high-probability reflective pixels. A connected component analysis algorithm is then used to aggregate spatially adjacent high-probability pixels into continuous potential reflective regions. This probability-based aggregation method is more intelligent than traditional brightness thresholding methods, capable of identifying areas with typical reflective characteristics that are not particularly bright, while avoiding misidentifying normal bright areas as reflective.

[0112] S205, calculate the average brightness value and area percentage of the potential reflective area.

[0113] The system calculates quantitative characteristic parameters for each potential reflective area. The average brightness value is obtained by arithmetically averaging the brightness values ​​of all pixels within the area, while the area ratio is calculated as the ratio of the number of pixels in that area to the total number of pixels in the entire image. These parameters reflect the intensity and spatial distribution characteristics of the reflective area, providing a quantitative basis for subsequent differential processing.

[0114] S206, determine the local brightness uniformity coefficient of the potential reflective area based on the average brightness value and area ratio.

[0115] The system establishes a calculation model for the local brightness uniformity coefficient based on the characteristic parameters of potential reflective areas. This coefficient calculation comprehensively considers the influence of average brightness value and area proportion. For areas with high average brightness values, the coefficient is set relatively large to achieve a stronger brightness reduction effect; for areas with large area proportions, the coefficient is appropriately reduced to avoid over-processing and impacting the overall image quality. The specific calculation formula is: Local Brightness Uniformity Coefficient = Baseline Coefficient × (1 + Brightness Influence Factor × Normalized Average Brightness Value) × (1 - Area Influence Factor × Normalized Area Proportion), where the baseline coefficient is typically set to 0.3-0.5, and the brightness influence factor and area influence factor are optimized according to the actual application scenario.

[0116] S207, when determining the brightness uniformity parameters of the detected image, the local brightness uniformity coefficient is used to reduce the brightness of the potential reflective area, and the global brightness uniformity coefficient is used to adjust the brightness of the non-reflective area outside the potential reflective area; wherein, the local brightness uniformity coefficient is greater than the global brightness uniformity coefficient.

[0117] The system implements a dual-coefficient differentiated brightness uniformity processing strategy, which fully leverages the intelligent processing advantages of deep learning-based reflection detection results. For potential reflective areas identified by the deep learning model, the system employs a relatively large local brightness uniformity coefficient for targeted enhancement. This effectively suppresses reflections without affecting surrounding non-reflective areas. For non-reflective areas outside the potential reflective areas, the system uses a relatively small global brightness uniformity coefficient for conventional brightness adjustment, maintaining the natural brightness distribution characteristics of the image. By setting a constraint that the local brightness uniformity coefficient is greater than the global brightness uniformity coefficient, the system ensures priority processing and enhancement of reflective areas. This precise localization significantly improves the reflection suppression effect while preserving the original characteristics of non-reflective areas in the image to the greatest extent, achieving an optimal balance between reflection suppression and image quality preservation.

[0118] Figure 2This is a schematic diagram of an intelligent reflection suppression system architecture for engine interior inspection provided in this application embodiment. Its core logic is presented as a star-shaped topology with a "reflection suppression processing host (electronic device e)" as the central hub. The entire workflow begins with the "industrial endoscope (electronic device a)" on the left and the "environmental perception unit (electronic device b)" above. The former is responsible for collecting raw image data of the engine interior, while the latter synchronously acquires and transmits internal environmental feature data. After these two sources of information converge to the central processing host, the algorithm unit inside the host performs core processing steps such as brightness uniformity calculation, reflection area detection, and threshold judgment. Based on the processing results, the system differentiates into two output paths: one points to the "exposure controller (electronic device c)" on the right, forming a feedback loop, which corrects the front-end acquisition parameters in real time by sending exposure adjustment commands to physically reduce reflection; the other points to the "detection display terminal (electronic device d)" below, which finally presents the high-quality, reflection-free, and uniformly bright image processed by the algorithm to the operator, thereby realizing a complete automated inspection process from environmental perception, image acquisition, feedback control to final display.

[0119] Based on the above method, this application also discloses a reflection suppression system for aircraft engine inspection based on an industrial endoscope, such as... Figure 3 As shown, Figure 3 This is a schematic diagram of a reflection suppression system for aircraft engine inspection based on an industrial endoscope, provided in an embodiment of this application. The system includes: an acquisition module, a determination module, a processing module, a detection module, and an adjustment module; wherein, The system comprises the following modules: an acquisition module for acquiring images of the interior of an aircraft engine through an industrial endoscope and for obtaining environmental characteristics of the engine's interior; a determination module for determining brightness uniformity parameters of the acquired image based on these environmental characteristics; a processing module for performing brightness uniformity processing on the acquired image according to the brightness uniformity parameters to obtain a uniform image; a detection module for detecting the presence of reflections in the uniform image; if reflections are present, adjusting the exposure parameters of the industrial endoscope's camera; and an adjustment module for monitoring the overall brightness of the uniform image in real time during exposure parameter adjustment; stopping exposure parameter adjustment and outputting the reflection-suppressed acquired image when the overall brightness falls below a preset brightness threshold.

[0120] It should be noted that the system provided in the above embodiments is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0121] Please see Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 4 As shown, the electronic device 1000 may include: at least one processor 1001, at least one network interface 1004, a user interface 1003, a memory 1005, and at least one communication bus 1002.

[0122] The communication bus 1002 is used to realize the connection and communication between these components.

[0123] The user interface 1003 may include a display screen and a camera. Optionally, the user interface 1003 may also include a standard wired interface and a wireless interface.

[0124] The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0125] The processor 1001 may include one or more processing cores. The processor 1001 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 1005, and by calling data stored in the memory 1005. Optionally, the processor 1001 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 1001 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content to be displayed on the screen; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 1001 and may be implemented as a separate chip.

[0126] The memory 1005 may include random access memory (RAM) or read-only memory. Optionally, the memory 1005 may include a non-transitory computer-readable storage medium. The memory 1005 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 1005 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 1005 may also be at least one storage device located remotely from the aforementioned processor 1001. Figure 4 As shown, the memory 1005, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for a method of suppressing reflections during aircraft engine inspection based on an industrial endoscope.

[0127] exist Figure 4 In the electronic device 1000 shown, the user interface 1003 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 1001 can be used to call the application program stored in the memory 1005, which is a method for suppressing reflection of aircraft engines based on industrial endoscopes. When executed by one or more processors, the electronic device performs one or more of the methods described in the above embodiments.

[0128] An electronic device readable storage medium stores instructions that, when executed by one or more processors, cause the electronic device to perform one or more of the methods described in the above embodiments.

[0129] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0130] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0131] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some service interfaces; indirect couplings or communication connections between devices or units may be electrical or other forms.

[0132] 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 units can be selected to achieve the purpose of this embodiment according to actual needs.

[0133] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0134] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0135] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will be readily apparent to those skilled in the art upon consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described herein. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for suppressing reflection during aircraft engine inspection based on an industrial endoscope, characterized in that, The method includes: Acquire images of the interior of an aircraft engine using an industrial endoscope, and acquire environmental characteristics of the interior of the aircraft engine; Based on the environmental characteristics, determine the brightness uniformity parameters of the detected image; The detected image is subjected to brightness homogenization processing according to the brightness homogenization parameters to obtain a homogenized image; The system detects whether there is glare in the homogenized image. If there is glare in the homogenized image, the exposure parameters of the camera of the industrial endoscope are adjusted. During the adjustment of the exposure parameters, the overall brightness of the homogenized image is monitored in real time. When the overall brightness is lower than the preset brightness threshold, the adjustment of the exposure parameters is stopped, and the detection image after reflection suppression is output.

2. The method for suppressing reflection during aircraft engine inspection based on an industrial endoscope according to claim 1, characterized in that, The environmental characteristics include component material, light source illumination angle, and chamber structure. Determining the brightness uniformity parameters of the detected image based on these environmental characteristics includes: Obtain the reflectance coefficient of the component material; Based on the illumination angle of the light source, calculate the incident angle and reflection angle of the light on the surface of the component material; Based on the chamber structure, the light propagation path and scattering characteristics are determined; Based on the light propagation path, combined with the reflection coefficient, the incident angle, the reflection angle, and the scattering characteristics, the expected value of the brightness distribution in each region of the detected image is calculated; Calculate the brightness difference between the expected brightness distribution value and the preset standard brightness value, and determine the brightness uniformity parameter of the detected image based on the brightness difference.

3. The method for suppressing reflection during aircraft engine inspection based on an industrial endoscope according to claim 2, characterized in that, The step of calculating the expected brightness distribution of each region in the detected image based on the light propagation path, combined with the reflection coefficient, the incident angle, the reflection angle, and the scattering characteristics, includes: The detected image is divided into multiple pixel regions, and the component surface position corresponding to each pixel region is determined according to the light propagation path. Based on the reflection coefficient at the surface position of the component, calculate the basic reflectance value of each pixel region; Calculate the reflection intensity coefficient of each pixel region by combining the incident angle and the reflection angle; Based on the scattering characteristics, the intensity of scattered light received by each pixel region is calculated; The basic reflective brightness value is arithmetically multiplied by the reflective intensity coefficient to generate a reflective modulation brightness value. The reflective modulation brightness value is arithmetically added by the scattered light intensity to generate the expected brightness distribution value for each pixel region. By combining the expected brightness distribution values ​​of each pixel region, the expected brightness distribution values ​​of each region in the detected image are generated.

4. The method for suppressing reflection during aircraft engine inspection based on an industrial endoscope according to claim 1, characterized in that, The step of performing brightness homogenization processing on the detected image according to the brightness homogenization parameters to obtain a homogenized image includes: The detected image is preprocessed to remove noise and enhance edge information; The brightness values ​​of each region in the detected image are adjusted according to the brightness uniformity coefficient in the brightness uniformity parameters; wherein, for a first region whose brightness value is higher than a preset upper threshold, the brightness value of the first region is reduced according to the brightness uniformity coefficient; and for a second region whose brightness value is lower than a preset lower threshold, the brightness value of the second region is increased according to the brightness uniformity coefficient. The adjusted detection image is smoothed to generate a uniform image.

5. The method for suppressing reflection during aircraft engine inspection based on an industrial endoscope according to claim 1, characterized in that, The detection of whether there is reflection in the homogenized image includes: Calculate the brightness value of each pixel in the homogenized image; Identify candidate reflective pixels whose brightness values ​​exceed a preset reflectivity threshold; Connectivity analysis is performed on the candidate reflective pixels, and interconnected candidate reflective pixels are aggregated into candidate reflective regions; Calculate the area and average brightness value of each candidate reflective region; When the area of ​​the candidate reflective region is greater than a preset area threshold and the average brightness value is greater than a preset brightness threshold, the candidate reflective region is determined as a reflective region. When at least one of the reflective areas is detected, it is determined that there is a reflective phenomenon in the homogenized image.

6. The method for suppressing reflection during aircraft engine inspection based on an industrial endoscope according to claim 1, characterized in that, Adjusting the exposure parameters of the camera in the industrial endoscope includes: Obtain the location information and reflection intensity of the reflective areas in the homogenized image; Based on the reflected light intensity, the exposure parameter adjustment amount is calculated, which includes the exposure time adjustment amount and the gain adjustment amount; The exposure time and gain of the camera are reduced based on the adjustment of the exposure parameters. A new detection image is obtained after adjusting the exposure parameters, and the brightness is uniformized on the new detection image to obtain a new uniform image; Calculate the residual reflectivity of the reflective region in the new homogenized image; Determine whether the residual reflective intensity is lower than a preset reflective suppression threshold. If the residual reflective intensity is not lower than the preset reflective suppression threshold, then continue to adjust the exposure parameters based on the residual reflective intensity until the residual reflective intensity is lower than the preset reflective suppression threshold or the overall brightness is lower than the preset brightness threshold.

7. The method for suppressing reflection during aircraft engine inspection based on an industrial endoscope according to claim 1, characterized in that, The method further includes: The detected image is input into a pre-trained deep learning model, which is trained based on industrial endoscope detection images with reflective labels. The deep learning model extracts features from the detected image, including brightness gradient features, texture features, and edge features. Based on the aforementioned features, a reflectivity probability map is generated, wherein each pixel in the reflectivity probability map corresponds to a reflectivity probability value; Pixels with a reflectivity probability value greater than a preset probability threshold in the reflectivity probability map are aggregated into potential reflectivity areas; Calculate the average brightness value and area percentage of the potential reflective areas; Based on the average brightness value and area ratio, the local brightness uniformity coefficient of the potential reflective area is determined; When determining the brightness uniformity parameters of the detected image, the local brightness uniformity coefficient is used to reduce the brightness of the potential reflective area, and the global brightness uniformity coefficient is used to adjust the brightness of the non-reflective area outside the potential reflective area; wherein, the local brightness uniformity coefficient is greater than the global brightness uniformity coefficient.

8. A reflection suppression system for aircraft engine inspection based on an industrial endoscope, characterized in that, The system includes: an acquisition module, a determination module, a processing module, a detection module, and an adjustment module; wherein, The acquisition module is used to acquire images of the interior of an aircraft engine obtained by an industrial endoscope, and to acquire environmental characteristics of the interior of the aircraft engine. The determining module is used to determine the brightness uniformity parameters of the detected image based on the environmental characteristics. The processing module is used to perform brightness uniformization processing on the detected image according to the brightness uniformization parameters to obtain a uniform image; The detection module is used to detect whether there is a reflection in the homogenized image. If there is a reflection in the homogenized image, the exposure parameters of the camera of the industrial endoscope are adjusted. The adjustment module is used to monitor the overall brightness of the homogenized image in real time during the process of adjusting the exposure parameters. When the overall brightness is lower than a preset brightness threshold, the adjustment of the exposure parameters is stopped and the detection image after reflection suppression is output.

9. An electronic device, characterized in that, The device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer program is stored that can be loaded by a processor and executed as described in any one of claims 1-7.

Images