A method for removing thermal fog from images based on gray-level weighted averaging
By performing grayscale weighted averaging on high-temperature DIC images and using ZNCC to calculate weights, the image distortion problem caused by thermal fog disturbance was solved, the measurement accuracy and reliability were improved, the experimental complexity and cost were reduced, and the application scope was expanded.
Patent Information
- Application Number
- CN202411075312.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-07
- Publication Date
- 2025-10-28
- Estimated Expiration
- 2044-08-07
AI Technical Summary
In digital image correlation (DIC) measurements at high temperatures, thermal fog disturbances cause image distortion, affecting measurement accuracy. Existing device optimization methods increase experimental complexity and cost.
By performing grayscale weighted averaging on multiple frames of images acquired under high-temperature conditions, and using zero-mean normalized cross-correlation coefficient (ZNCC) to calculate weights, a weighted average grayscale image is generated, reducing noise caused by thermal fog and improving image clarity and contrast.
It effectively smooths out random noise caused by thermal radiation and surface changes, improves the accuracy and reliability of DIC measurements, reduces measurement errors, and expands the application range to experimental research on more high-temperature materials and structures.
Smart Images
Figure CN118967541B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the fields of aerospace materials and solid mechanics, specifically relating to an image thermal fog removal method based on grayscale weighted averaging. Background Technology
[0002] In high-temperature digital image correlation (DIC) testing, the heating device heats not only the sample but also the air between the camera and the sample, resulting in a significant thermal gradient and severe air turbulence between them. These air turbulence and changing thermal gradients inevitably cause changes in the air's refractive index, further distorting the light; this phenomenon is known as thermal fogging. Thermal fogging turbulence leads to image distortion and introduces considerable error into DIC measurement results.
[0003] The impact of thermal fog on the accuracy of DIC measurements largely depends on the experimental setup and the method of sample heating. Lyons et al. first raised the issue of thermal fog during high-temperature furnace experiments. They found that changes in the refractive index of heated air caused significant image distortion. By using a fan to stabilize the airflow in front of the window, they achieved displacement and strain accuracies at temperatures up to 600°C comparable to room temperature. Novak and Zok et al. verified that using an air knife could eliminate the effect of thermal fog to some extent; at 800°C, the use of an air knife reduced the standard deviation of the test results. Furthermore, Grant et al. successfully eliminated the thermal fog effect by using a vacuum chamber. Su et al. employed an observation window with a liquid-cooled system, which effectively eliminated the thermal fog effect both inside and outside the vacuum chamber.
[0004] While optimizing the experimental setup can significantly reduce the impact of thermal fog disturbances on the accuracy of DIC measurements under high-temperature environments, this increases the complexity and cost of the experiment, thus limiting its feasibility for widespread application. Therefore, after acquiring high-temperature thermal fog images, algorithmic processing is used to eliminate thermal fog disturbances in the images. This method has broad applicability under various experimental conditions, making it a challenging and important direction in current thermal fog removal research. Summary of the Invention
[0005] The technical problem to be solved by the present invention is to provide a thermal fog removal method based on image gray-level weighted averaging to overcome the influence of thermal fog disturbance on the image in high-temperature DIC measurement, improve measurement accuracy, and the applicable temperature range is 50℃~3000℃.
[0006] The technical solution adopted by this invention to solve its technical problem is as follows:
[0007] This method reduces noise and improves DIC measurement accuracy by weighted averaging of grayscale values from multiple frames of images acquired under high-temperature conditions. In the high-temperature experiment, continuously acquired high-temperature images are used, and the ZNCC values of the high-temperature images and reference images are calculated. Weights are then designed based on these ZNCC values to perform a weighted average of the image grayscale values, generating a new weighted average grayscale image. This method effectively smooths out random noise caused by thermal radiation and surface variations, improves image clarity and contrast, and enables the DIC system to more accurately identify and track feature points, thereby enhancing measurement accuracy and reliability.
[0008] The specific technical solution adopted in this invention is as follows: The quality of thermal fog images acquired under high temperature is evaluated and weighted. Then, a weighted average of corresponding pixel positions in a set of images acquired under the same condition is calculated. Thermal fog removal is performed by defining the value of each pixel in the synthesized image as the weighted average of pixels at corresponding positions in multiple images. The grayscale value of image pixels (x, y) affected by thermal fog under high temperature conditions can be defined by the following formula:
[0009] g i (x,y)=f(x,y)+r i (x,y)
[0010] Where g i (x,y) represents the grayscale value of the image at these coordinates, r i (x,y) represents the grayscale error value caused by thermal fog, and f(x,y) represents the grayscale value unaffected by thermal fog. The weight design is based on the zero-mean normalized cross-correlation criterion (ZNCC). The ZNCC value can characterize the matching quality between the high-temperature image and the reference image. The larger the ZNCC cross-correlation coefficient and the closer it is to 1, the higher the correlation, which can be regarded as the optimal match. The ZNCC calculation method is as follows:
[0011]
[0012] Where f(x,y) is the gray intensity of the reference image at coordinates (x,y), and g(x′) is the gray intensity of the reference image at coordinates (x,y). i ,y′ j ) is the deformed image at coordinates (x′) i ,y′ j The grayscale intensity at () is given. The size of the subset is k×k pixels. and These are the average grayscale intensity values of the reference image and the deformed image subset, respectively. The weights of the images are obtained by calculating the ZNCC (Zero-Negative Coefficient of Gravity) between a set of images affected by high-temperature thermal fog and the reference image, and then normalizing and performing non-linear transformation on the obtained ZNCC values. The specific calculation method is as follows:
[0013]
[0014]
[0015] Where w i Let W be the weight of each image, W be the sum of image weights, and M be the number of image processes.
[0016] Gray values processed by gray-scale weighted average technique Calculated using the following expression:
[0017]
[0018] The value of each pixel in the synthesized image is the weighted average of pixels at the corresponding location in multiple images. Using grayscale weighted averaging reduces random noise caused by thermal fog, indicating that the algorithm can effectively improve the signal-to-noise ratio and DIC calculation accuracy of high-temperature images within the temperature range of 50℃ to 3000℃.
[0019] The present invention also claims the use of the image thermal fog removal method based on gray-scale weighted averaging for removing thermal fog from DIC images of aerospace materials in high-temperature environments.
[0020] The beneficial effects are as follows:
[0021] 1) Reduce thermal fog interference
[0022] Gray-scale weighted averaging technology filters out high-frequency noise by weighted averaging of gray values from multiple frames of an image, thus preserving the main features of the image. After noise reduction, the image clarity and contrast are significantly improved, allowing the DIC system to more accurately identify and track feature points and reduce measurement errors caused by thermal fog.
[0023] 2) Improve measurement accuracy
[0024] By reducing noise and enhancing image quality, gray-level weighted averaging helps improve the correlation between images, making it easier for the DIC algorithm to find and match feature points. This reduces data fluctuations and errors, improving the overall accuracy of DIC analysis. DIC can more accurately measure the deformation, displacement, and strain of materials. This improvement is crucial for high-precision material property analysis and the reliability of experimental results.
[0025] 3) Expanding the scope of application
[0026] Using algorithms to remove thermal fog not only improves the measurement accuracy of DIC at high temperatures, but also reduces the complexity and cost of experiments, expanding its application to experimental research on more high-temperature materials and structures. Attached Figure Description
[0027] Figure 1This is a flowchart of an image thermal fog removal method based on gray-level weighted averaging in Example 1.
[0028] Figure 2 This refers to the DIC image acquisition system in Example 1.
[0029] Figure 3 This is a comparison of the displacement field cloud maps calculated by DIC from the thermal fog image and the image after algorithm processing in Example 1.
[0030] In the figure, 1 is the DIC calculation result of the original image, and 2 is the DIC calculation result of the processed image.
[0031] Figure 4 The displacement fluctuation curves are calculated by DIC from the thermal fog image and the image after algorithm processing in Example 1. Detailed Implementation
[0032] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.
[0033] The purpose of this invention is to improve the accuracy of DIC (Digital Conversion Index) measurement by using a grayscale weighted averaging method for image thermal fog removal. To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments.
[0034] Example 1
[0035] This embodiment is a gray-level weighted averaging method for removing thermal fog from images, such as... Figure 1 The flowchart is shown below, and the specific steps are as follows:
[0036] 1) Data Acquisition System Setup
[0037] A digital image correlation method image acquisition system is constructed. The acquisition system consists of a camera, lens, and light source, such as... Figure 2 As shown;
[0038] 2) Image Acquisition
[0039] By adjusting the light source intensity and camera parameters such as focal length and exposure, image clarity was ensured. Based on this, the DIC image acquisition system built in step one was used to acquire a reference image without thermal fog under room temperature conditions and an image with thermal fog interference under high temperature conditions.
[0040] 3) Calculate the zero-mean normalized cross-correlation coefficient.
[0041] Zero-mean normalized cross-correlation (ZNCC) was calculated for each high-temperature image to characterize the matching quality between the thermal fog image and the reference image. A higher ZNCC cross-correlation coefficient, closer to 1, indicates a higher correlation and can be considered an optimal match. The specific formula is as follows:
[0042]
[0043] Here, f(x,y) is the gray intensity of the reference image at coordinates (x,y), and g(x′,y′) is the gray intensity of the deformed image at coordinates (x′,y′). The size of the subset is k×k pixels. and These are the average grayscale intensity values of the reference image and the deformed image subset, respectively.
[0044] 4) Calculate the weights
[0045] The calculated Normalized Cross-Correlation Coefficient (ZNCC) values are further normalized and subjected to nonlinear transformation to determine the weight of each image. First, the ZNCC values are normalized to ensure they fall within a specific range. Then, a nonlinear transformation is applied to enhance image differences within this specific range, thereby better reflecting the relative importance of the images. The specific formula is as follows:
[0046]
[0047] Where w i W represents the weight of each image, M represents the sum of image weights, and M represents the number of image processing operations. To obtain a better effect in removing thermal fog, the value of M should be 20-30.
[0048] 5) Gray-scale weighted average processing
[0049] The calculated weights are used to perform a weighted average of the grayscale values at each pixel location in each image, ultimately generating a composite image. The grayscale values after processing using the grayscale weighted averaging technique are shown below. Calculated using the following expression:
[0050]
[0051] The value of each pixel in the synthesized image is a weighted average of the pixels at the corresponding location in multiple images. Using a grayscale weighted averaging technique can reduce random noise caused by thermal fog.
[0052] 6) Displacement calculation
[0053] The displacement field is calculated using the DIC method on the image processed in step five. The gray-level weighted algorithm can reduce the failure of DIC contour map calculations and displacement fluctuations caused by thermal fog, such as... Figure 3-4As shown, this method achieves thermal fog removal from DIC images under high-temperature conditions.
[0054] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for removing thermal fog from images based on gray-level weighted averaging, comprising the following steps: 1) Image Acquisition A set of images was acquired under high-temperature conditions, including images of thermal fog interference under high-temperature conditions; and a reference image without thermal fog was acquired under room temperature conditions. 2) Calculate ZNCC Zero-mean normalized cross-correlation (ZNCC) was performed on each high-temperature image to characterize the matching quality between the thermal fog image and the reference image. The ZNCC calculation method is as follows: in, f(x,y) is the gray intensity of the reference image at coordinates (x,y), while g(x′,y′) is the gray intensity of the deformed image at coordinates (x′,y′). The size of the subset is k×k pixels. and These are the average grayscale intensity values of the reference image and the deformed image subset, respectively. 3) Calculate the weights Based on the calculated ZNCC values, normalization and nonlinear transformation processes are performed to determine the weight of each image. First, the ZNCC values are normalized to ensure they fall within a specific range. Then, nonlinear transformations are applied to enhance image differences within that specific range. The calculation method is as follows: Where w i Let W be the weight of each image, W be the sum of image weights, and M be the number of image processes. 4) Gray-scale weighted average processing Using the calculated weights, a weighted average is applied to the grayscale values at each pixel location in each image to generate a composite image. The grayscale values of pixel points (x, y) in an image affected by thermal fog in a high-temperature environment are defined by the following formula: g i (x,y)=f(x,y)+r i (x,y) Where g i (x,y) represents the grayscale value of the image at coordinates (x,y), r i (x,y) represents the grayscale error value caused by thermal fog, and f(x,y) represents the grayscale value unaffected by thermal fog, processed by grayscale weighted averaging. Calculated using the following expression: 5) Displacement calculation The displacement field is calculated using the DIC method on the image processed in step 4).
2. The image thermal fog removal method based on gray-level weighted averaging according to claim 1, characterized in that, In step 4), the value of each pixel in the synthesized image is the weighted average of the pixels at the corresponding positions in multiple images.
3. The image thermal fog removal method based on gray-level weighted averaging according to claim 1, characterized in that, Gray-scale weighted averaging is used to reduce random noise caused by thermal fog.
4. The image thermal fog removal method based on gray-level weighted averaging according to any one of claims 1-3, characterized in that, Used for removing thermal fog from DIC images of aerospace materials in high-temperature environments.