Bridge crack sub-pixel identification method and system based on moire phase unwrapping

Abstract

The invention discloses a bridge crack sub-pixel identification method and system based on moire phase unwrapping, and relates to the field of structural health monitoring. The method comprises the steps that a bridge image is collected and uploaded to a system, and the image is processed to obtain a de-noised image; performing sub-pixel registration on the reference grating image of the denoised image to obtain a registered reference grating image; using the registered reference grating image to generate a moire image to calculate a global mass fraction, calculating a wrapped phase by the moire image greater than a threshold value, otherwise, requiring the unmanned aerial vehicle to perform re-sampling; calculating an unwrapping phase for the moire image greater than the threshold value based on the pixel weight, and deducing a displacement field of each pixel by using the unwrapping phase; and combining the displacement gradient amplitude and the level set function to identify a crack region and quantify crack information. Through automatic moire analysis and displacement field calculation, the bridge tiny crack is accurately detected, the size and depth of the crack are quantified, and reliable support is provided for bridge health monitoring and maintenance decision making.

Description

Technical Field

[0001] The present invention relates to the field of structural health monitoring, and specifically to a method and system for sub-pixel identification of bridge cracks based on moiré phase unwrapping. Background Art

[0002] As an important infrastructure, bridges undertake important functions of transportation and logistics distribution. Ensuring the safety and long-term stable operation of bridges is one of the keys to public safety management. With the advancement of urbanization and the continuous increase in traffic flow, the problems of bridge aging and damage have gradually emerged, and the importance of bridge health monitoring and assessment has become increasingly prominent. Traditional bridge detection methods, such as manual inspection, acoustic detection, and structural sensor monitoring, although they can provide certain detection capabilities, have problems such as low efficiency, high cost, inability to monitor in real time, and difficulty in obtaining comprehensive information. Therefore, in recent years, bridge detection methods based on image processing and unmanned aerial vehicle (UAV) technology have gradually become a research and application hotspot.

[0003] In the past few decades, bridge health monitoring has usually relied on manual inspection or ground inspection. These methods are often accompanied by the consumption of time and labor costs, and it is difficult to cover the entire structure of the bridge. The results of manual inspection not only depend on the experience and skills of inspectors, but are also vulnerable to weather and environmental conditions, with relatively large uncertainties. In addition, bridge health monitoring also includes technologies such as acoustic methods, vibration methods, and strain monitoring. Although these methods can provide dynamic response information of the bridge, there are still certain limitations because most of them cannot accurately obtain the minute deformations of surface cracks.

[0004] Although UAV technology and image processing methods have been widely used in bridge detection, there are still some challenges at present. First, the accuracy and robustness of image processing algorithms need to be further improved. Especially in harsh weather, complex lighting conditions, and low-contrast scenarios, the image quality is poor, which affects the effect of crack detection. Second, the moiré interference method has very high requirements for image registration. Especially in the case of large-area and high-resolution images, the registration accuracy directly affects the final detection result. In addition, existing crack detection methods mostly rely on static images, while there may be dynamic changes on the bridge surface. This requires image processing technology not only to analyze static images, but also to adapt to changes in dynamic scenarios. Therefore, new algorithms and technological innovations are urgently needed to break through. Summary of the Invention

[0005] Based on the above-mentioned disadvantages of the existing technology, the purpose of the present invention is to provide a method and system for sub-pixel identification of bridge cracks based on moiré phase unwrapping to solve the above-mentioned technical problems.

[0006] To achieve the above object, the present invention provides the following technical solutions: A method for sub-pixel recognition of bridge cracks based on moiré phase unwrapping, including: S1. A drone carries a high-resolution camera to collect images of the bridge surface and uploads them to the cloud system. Geometric correction and denoising processing are performed on the bridge surface images to obtain denoised images; S2. Generate a reference grating image from the denoised image according to the optimal period, and use sub-pixel registration to obtain the registered reference grating image; S3. Generate a moiré image using the registered reference grating image, calculate the global quality score for the generated moiré image, calculate the wrapped phase for the moiré image with a score greater than the scoring threshold, otherwise require the drone to re-collect; S4. Calculate the pixel weights for the moiré image with a score greater than the scoring threshold, calculate the unwrapped phase based on the pixel weights, and use the interferometer principle to derive the displacement field of each pixel for the unwrapped phase; S5. Analyze the displacement gradient amplitude in the displacement field, combine with the level set function to identify the crack area and quantify the width, length, and depth of the crack, providing quantitative information for structural damage assessment.

[0007] The present invention is further configured such that S2 specifically includes: Calculate the gradient amplitude of the denoised image, and optimize the grating period using the gradient amplitude to obtain the optimal period; Generate a reference grating image based on the optimal period; Calculate the displacement using the phase correlation method, and perform sub-pixel registration to obtain the registered reference grating image.

[0008] The present invention is further configured such that the gradient amplitude calculation logic is: , where is the gradient amplitude, is the denoised image, is the continuous differential operator of the differential vector; The optimal period calculation logic is: , where is the optimal period, is the grating period, is the sine function, is the angular frequency of the sine function; The reference grating calculation logic is: , where is the reference grating image, and are respectively the base value and amplitude of the reference image, is the sine generation function of the grating waveform; The reference grating registration calculation logic is: , where is the registered reference grating image, and is the image translation displacement calculated by the phase correlation method, is the bicubic interpolation function.

[0009] The present invention is further configured such that S3 specifically includes: Multiplying the registered reference grating image and the denoised image to synthesize moiré to obtain a moiré image; Calculating the global quality score for the generated moiré image using the local contrast and frequency domain energy ratio; Calculating the wrapped phase for the moiré image greater than the scoring threshold, otherwise requiring the drone to re-acquire.

[0010] The present invention is further configured such that the moiré image generation logic: , where is the moiré image; The global quality score calculation logic: , where is the global quality score, and is the size of the image, is the local contrast, is the frequency domain energy ratio, is the noise compensation factor, is the noise level; The local contrast calculation logic: , where is the local contrast, is the maximum gray value within the local area, is the minimum gray value within the local area; The frequency domain energy ratio calculation logic: , where is the frequency domain energy ratio, is the frequency domain representation of the image, is the frequency domain representation of the desired image.

[0011] The present invention is further configured such that S4 specifically includes: Combining the global quality score, local contrast, and structural consistency to calculate the pixel weight; Based on the pixel weight, using the maximum spanning tree to calculate the path integral to obtain the unwrapped phase; Using the interferometer principle for the unwrapped phase to derive the displacement field of each pixel.

[0012] The present invention is further configured such that the pixel weight calculation logic: , where is the pixel weight, is the global quality score, is the local contrast, is the structural consistency, is the noise level; Path integral calculation logic: , where, is the unwrapped phase, is the wrapped phase, is the unwrapped phase gradient, is the wrapped phase gradient, is the rounding function, is the maximum spanning tree path, is the phase period; Displacement field calculation logic: , where, is the displacement field, is the laser wavelength. Specifically, the displacement field calculation uses the interferometer principle to convert the phase difference into physical displacement. Every phase change corresponds to displacement.

[0013] The present invention is further configured such that S5 specifically includes: Obtaining the sub-pixel displacement gradient amplitude using the displacement field calculation; Describing the crack profile based on the displacement gradient amplitude and the level set function, adding the coordinate sets of all detected crack boundaries to the crack profile point set, and calculating the length, width, and depth of the positions in the crack profile point set to obtain crack information.

[0014] The present invention is further configured such that the displacement gradient amplitude calculation logic: , where, is the displacement gradient amplitude; Crack profile description calculation logic: , where, is the level set function, is the curvature, is the gradient amplitude of the level set function; Crack profile point set construction logic: ; where, is the crack profile point set; Length calculation logic: , where, is the crack length, is the total number of profile points, is the point in the crack profile set, is the next point in the crack profile set, is the Euclidean distance between crack profile points; Depth calculation logic: , where, is the crack depth, is a constant factor, is the maximum value of the displacement field of all points in the set of contour points; Width calculation logic: , where is the standard deviation, which is obtained by Gaussian fitting of the gray value profile of the image sampled along the normal vector direction of each contour point.

[0015] The present invention also provides a bridge crack sub-pixel recognition system based on moiré phase unwrapping. The system includes: Image acquisition and preprocessing module: A high-resolution camera carried by a drone acquires images of the bridge surface and uploads them to the cloud system. Geometric correction and denoising processing are performed on the bridge surface images to obtain denoised images; Grating generation and registration module: Generates a reference grating image for the denoised image according to the optimal period, and uses sub-pixel registration to obtain the registered reference grating image; Moiré synthesis and quality assessment module: Generates a moiré image using the registered reference grating image, calculates the global quality score for the generated moiré image, calculates the wrapped phase for the moiré image with a score greater than the scoring threshold, otherwise requires the drone to re-acquire; Phase unwrapping and displacement calculation module: Calculates the pixel weights for the moiré image with a score greater than the scoring threshold, calculates the unwrapped phase based on the pixel weights, and uses the interferometer principle to derive the displacement field of each pixel for the unwrapped phase; Crack recognition and quantification module: Analyzes the magnitude of the displacement gradient in the displacement field, combines the level set function to identify the crack area and quantifies the width, length and depth of the crack, providing quantitative information for structural damage assessment.

[0016] The present invention provides a bridge crack sub-pixel recognition method and system based on moiré phase unwrapping. The method acquires images of the bridge surface by a high-resolution camera carried by a drone and uploads them to the cloud system. Geometric correction and denoising processing are performed on the bridge surface images to obtain denoised images; Generates a reference grating image for the denoised image according to the optimal period, and uses sub-pixel registration to obtain the registered reference grating image; Generates a moiré image using the registered reference grating image, calculates the global quality score for the generated moiré image, calculates the wrapped phase for the moiré image with a score greater than the scoring threshold, and requires the drone to re-acquire for the moiré image with a score less than the scoring threshold; Calculates the pixel weights for the moiré image with a score greater than the scoring threshold, calculates the unwrapped phase based on the pixel weights, and uses the interferometer principle to derive the displacement field of each pixel for the unwrapped phase; Analyzes the magnitude of the displacement gradient in the displacement field, combines the level set function to identify the crack area and quantifies the width, length and depth of the crack, providing quantitative information for structural damage assessment. The beneficial effects produced include: Efficient and Accurate Bridge Surface Damage Detection: By using drones equipped with high-resolution cameras to collect bridge surface images and combining technologies such as image denoising, image registration, moiré pattern generation, and displacement field analysis, it is possible to accurately obtain information on minute cracks and other structural damages on the bridge surface. This method not only improves the detection accuracy but also effectively enhances the detection efficiency, providing more comprehensive and accurate technical support for the health monitoring of bridges.

[0017] Automated Crack Detection and Quantitative Evaluation: Through the generated moiré pattern images, combined with global quality scores such as local contrast and frequency domain energy ratio, automatically evaluate the quality of the moiré pattern images, and precisely detect cracks and perform wrapped phase unwrapping. Use the gradient calculation method to derive the displacement field from the unwrapped phase, further quantifying the geometric features of the cracks, such as width, length, depth, etc. The automation of this process significantly reduces the need for manual intervention and improves the accuracy of crack detection.

[0018] High-Precision Crack Size and Depth Analysis: Based on the sub-pixel displacement gradient amplitude analysis calculated from the displacement field and combined with the level set function for crack contour description, it is possible to accurately identify the crack area and quantify the width, length, and depth of the cracks. Through the set analysis of the crack contour points, accurately calculate the geometric features of the cracks, providing a quantitative basis for structural damage assessment. This technology is of great significance for the long-term monitoring and maintenance decision-making of bridges.

[0019] The above description is only an overview of the technical solution of this application. In order to be able to more clearly understand the technical means of this application, it can be implemented in accordance with the content of the specification. And in order to make the above and other purposes, features, and advantages of this application more obvious and understandable, the following specifically illustrates the specific embodiments of this application. Brief Description of the Drawings

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. In the drawings: Figure 1 It is a flowchart of a method for sub-pixel identification of bridge cracks based on moiré pattern phase unwrapping shown in an exemplary embodiment of the present invention; Figure 2 It is a schematic structural diagram of a system for sub-pixel identification of bridge cracks based on moiré pattern phase unwrapping shown in an exemplary embodiment of the present invention. Detailed Description of the Embodiments

[0021] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand the other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, rather than for limiting the protection scope of the present invention.

[0022] It should be noted that the diagrams provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner. Therefore, only the components related to the present invention are shown in the diagrams, rather than being drawn according to the number, shape, and size of the components in actual implementation. The type, quantity, and ratio of each component in actual implementation can be arbitrarily changed, and the component layout type may also be more complex.

[0023] In the following description, a large number of details are explored to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand. Embodiment 1

[0024] A sub-pixel recognition method for bridge cracks based on moiré phase unwrapping, as Figure 1 shown, includes: S1. A drone carries a high-resolution camera to collect images of the bridge surface and upload them to the cloud system. Geometric correction and denoising processing are performed on the bridge surface images to obtain denoised images; S2. Generate a reference grating image for the denoised image according to the optimal period, and use sub-pixel registration to obtain the registered reference grating image; S3. Generate a moiré image using the registered reference grating image, calculate the global quality score for the generated moiré image, calculate the wrapped phase for the moiré image with a score greater than the scoring threshold, otherwise require the drone to re-collect; S4. Calculate the pixel weights for the moiré image with a score greater than the scoring threshold, calculate the unwrapped phase based on the pixel weights, and use the interferometer principle to derive the displacement field of each pixel for the unwrapped phase; S5. Analyze the displacement gradient amplitude in the displacement field, combine the level set function to identify the crack area and quantify the width, length, and depth of the crack, providing quantitative information for structural damage assessment.

[0025] The present invention is further configured such that S2 specifically includes: Calculate the gradient amplitude for the denoised image, and optimize the grating period using the gradient amplitude to obtain the optimal period; Generate a reference grating image based on the optimal period; Use the phase correlation method to calculate the displacement and perform sub-pixel registration to obtain the registered reference grating image. Specifically, the gradient magnitude represents the edge intensity or saliency of the image at that position. The larger the gradient magnitude, the more drastic the change in the image at that position, usually indicating the presence of an edge or texture at that position; use these gradient magnitudes to optimize the grating period. The goal of optimization is to find a period such that the texture in the image matches the grating period, that is, find the optimal period by maximizing the weighted sine wave response; generate a reference grating image according to the optimal period. This reference image is an image with periodic sine texture, and the generated grating pattern can match the texture features of the original image; the registered reference grating is obtained by calculating the phase correlation between images in the frequency domain using the phase correlation method to calculate the translational displacement between images, and after obtaining the translational displacement, using bicubic interpolation to achieve sub-pixel image translation. The basic idea of the phase correlation method is to calculate the correlation between images through multiplication in the frequency domain. After Fourier transform, the phase part contains the spatial information of the image. Using the complex conjugate product can obtain the displacement between images, and this displacement corresponds to the offset that needs to be adjusted in image registration.

[0026] The present invention is further configured as follows, the gradient magnitude calculation logic: , where is the gradient magnitude, is the denoised image, is the continuous differential operator of the differential vector; The optimal period calculation logic: , where is the optimal period, is the grating period, is the sine function, is the angular frequency of the sine function; The reference grating calculation logic: , where is the reference grating image, and are respectively the base value and amplitude of the reference image, is the sine generation function of the grating waveform; The reference grating registration calculation logic: , where is the registered reference grating image, and is the image translation displacement calculated by the phase correlation method, is the bicubic interpolation function. Specifically, the gradient magnitude It is used to indicate the intensity of local changes in an image. The larger the gradient amplitude, the more sensitive it is to the selection of the grating period. Indicates that the corrected image is at position The gradient at the position is a measure of the rate of change of image intensity, indicating the direction and rate of change of the image at that position. It refers to the Euclidean norm of the image gradient vector, which is used here to indicate the magnitude of the gradient. The larger the gradient, the more drastic the image change, that is, the more obvious the edge. By adjusting the grating period To optimize the texture response of the image, the grating period The value range is This is a search interval, the purpose is to select the grating period that can best enhance the image features within this range, the sine function Indicates that the image position under the grating period The sine function is used here to generate a texture response with the grating period The corresponding fundamental frequency component expresses the texture characteristics of the grating in the form of fluctuations, matches the grating structure of different periods with the image content, and the angular frequency Indicates each position Grating frequency and grating period The relationship between is the basic frequency of the grating, which indicates the frequency of the grating fluctuation in one cycle and determines the periodicity of the sine wave. The frequency of a smaller grating will be very high and the fluctuations will be very dense, while if Larger fluctuations are more sparse; refer to the raster image Based on the optimized grating period and a sine wave function, is the base brightness of the image, ensuring that the generated image value is within In the range, 127 is the amplitude of the raster image, ensuring that the generated image pixel values ​​are from Changes between is the sinusoidal function that generates the grating pattern and defines the spatial periodicity of the pattern, where Indicates along The periodic variation of the axis, Indicates along The axis changes periodically, and its frequency is Half of the axis direction; reference raster image after registration The reference raster image is translated at the sub-pixel level through bicubic interpolation to align the target image. This translation adjusts the spatial position of the image so that the texture structure of the reference raster image and the target image are as consistent as possible. Perform a translation operation on the reference grating image to adjust the image position to align with the target image. The bicubic interpolation function is a high-precision interpolation method that calculates new pixel values by considering the 16 neighboring pixel values in the image. Compared with nearest-neighbor interpolation and bilinear interpolation, it can provide higher precision.

[0027] The present invention is further configured such that S3 specifically includes: Multiply the registered reference grating image and the denoised image to synthesize moiré patterns to obtain a moiré pattern image; Calculate the global quality score for the generated moiré pattern image using the local contrast and frequency domain energy ratio; Calculate the wrapped phase for the moiré pattern image with a quality score greater than the scoring threshold, otherwise require the drone to re-acquire the image. Specifically, during the moiré pattern synthesis process, the pixel values of the two images are multiplied, and the resulting new image can enhance the periodic structure in the image. This is equivalent to the texture information of the two images being superimposed on each other, producing the effect of moiré patterns; the global quality score combines the contrast, frequency domain purity, and noise level of the image. The higher the quality score of the image, the better the details, structure, and signal-to-noise ratio of the image. Among them: the higher the average contrast, the richer the details and clearer the structure of the image; the higher the frequency domain purity, the more obvious the periodic information and less noise in the image; the noise compensation is used to reduce the influence of noise and improve the quality score of the image; set the scoring threshold. If it is less than the threshold, it proves that the drone image corresponding to the moiré pattern is not clear enough and cannot be processed further, and the image at this position needs to be re-acquired. If it is greater than the threshold, it proves that the moiré pattern image corresponding to the global quality score belongs to a high-quality image and can be used for subsequent crack identification. Calculate the wrapped phase for this image and enter the next module. Calculating the phase wrap is a prior art and will not be elaborated here.

[0028] The present invention is further configured such that the moiré pattern image generation logic: , where is the moiré pattern image; The global quality score calculation logic: , where is the global quality score, and are the dimensions of the image, is the local contrast, is the frequency domain energy ratio, is the noise compensation factor, is the noise level; The local contrast calculation logic: , where is the local contrast, is the maximum gray value within the local region, is the minimum gray value within a local area; Frequency domain energy ratio calculation logic: , where is the frequency domain energy ratio, is the frequency domain representation of the image, is the frequency domain representation of the expected image. Specifically, in the moiré pattern image calculation, by multiplying the values at each pixel position of two images, the features of the two images can be combined. For example, if the two images contain different periodic structures, the product will highlight these structures. This operation can enhance the periodic interference in the image and can also be used to enhance certain specific texture features of the image; the global quality score quantifies the quality of the moiré pattern image and judges the quality of the moiré pattern image. The global quality score calculation combines the contrast, frequency domain purity, and noise level of the image. The higher the quality score of the image, the better the details, structure, and signal-to-noise ratio of the image. is the input value, and the global quality score of this moiré pattern image is calculated through a formula. The size of the image and represent the width and height of the image respectively, is the total number of pixels in the image, represents the average contrast of the image, which is the average of the contrast values at all pixel positions; the local contrast at position is the contrast at that position, usually a measure of the local change in the gray values of the image. Contrast is used to measure the detail and structural clarity of the image. A higher contrast usually indicates more details in the image. and are the maximum and minimum gray values within the local window, which reflect the degree of gray difference within this local area. The local contrast calculation is usually local, and the contrast of each pixel value is calculated based on the gray values of its surrounding pixels; the frequency domain energy ratio is the energy ratio of the image in the frequency domain, representing the purity of the image frequency domain or the concentration of frequency domain information, and is used to measure the ratio of the frequency components of the image to the noise or irrelevant components. The frequency domain representation of the image is obtained through the Fourier transform of the image and contains information about the image at various frequencies. The frequency domain representation of the expected image is also obtained through the Fourier transform of the image and contains information about the image at various frequencies. This expected image is a preset template image, which provides a benchmark for the frequency domain energy ratio and is used to quantify the quality of the target image; the noise compensation factor functions to compensate for the noise in the image through this factor and reduce the impact of noise on the quality score. The noise level It is calculated through the variance of the moiré pattern image, representing the noise intensity of the moiré pattern image. A higher variance value indicates stronger noise in the image. The specific calculation formula is: , where represents the pixel variance of the moiré pattern image at position . Variance is a measure of the change in pixel gray values within a local area of the image. A larger variance means greater pixel variation in that area, usually associated with noise or texture.

[0029] The present invention is further configured such that S4 specifically includes: Calculating the pixel weight by combining the global quality score, local contrast, and structural consistency; Based on the pixel weight, using the maximum spanning tree to calculate the path integral to obtain the unwrapped phase; Deriving the displacement field of each pixel from the unwrapped phase using the interferometer principle. Specifically, the global quality score in the pixel weight is for the signal-to-noise ratio of the overall tibia moiré pattern image, the local contrast is for reflecting the stripe clarity of the moiré pattern image, and the structural consistency is for quantifying the consistency of the stripe direction. The pixel weight calculated by coupling these three physical features is more reliable than a single weight; constructing the maximum spanning tree enables the integration path to preferentially pass through high-weight paths, reducing error propagation and making the obtained unwrapped phase more reliable; in the fields of image registration, object deformation, or motion analysis, the calculation of the displacement field is very important for understanding the spatial relationships and changes in the image. Using the interferometer principle to derive the displacement field of each pixel, especially in optical image processing, can help accurately capture local changes in the image, especially small deformations or displacements.

[0030] The present invention is further configured with the pixel weight calculation logic: , where is the pixel weight, is the global quality score, is the local contrast, is the structural consistency, is the noise level; Path integral calculation logic: , where is the unwrapped phase, is the wrapped phase, is the unwrapped phase gradient, is the wrapped phase gradient, is the rounding function, is the maximum spanning tree path, is the phase period; Displacement field calculation logic: , where is the displacement field, is the laser wavelength. Specifically, the pixel weight By combining the global quality score, local contrast, and structural consistency, a weight is assigned to each pixel according to the different characteristics of the image, thereby generating a high-quality image synthesis result. The compensation of the noise level further ensures that the noise area does not affect the final image quality; structural consistency is used to detect the stability of the texture and shape of the local area. If the local structure of the image is more consistent, the weight of this area is higher, otherwise it is lower. The structural consistency can be calculated by methods such as edge detection and texture analysis. The value range of the structural consistency is between, and the larger the value, the more stable the structure; noise level is used to represent the noise level of the image in the local area. The area with larger noise will affect the image quality, so it needs to be compensated. In the formula, the noise level is used as a factor to adjust the final weight to ensure that the weight of the noise area is lower. The formula is: , where represents the pixel variance of the moiré pattern image at the position . The value range is greater than or equal to , and the larger the value, the stronger the noise and the worse the image quality. In the calculation of pixel weights: high-quality areas will obtain higher pixel weights, low-quality areas will obtain lower pixel weights, and the noise area is automatically suppressed through the introduction of , avoiding the interference of noise effects on image processing; path integration In, the edge weights of the maximum tree are defined according to the pixel weights. The edges with larger pixel weights are preferentially passed to ensure that the high-reliability areas are integrated first. After the phase jump correction calculation, the correction values are accumulated along the path, and the true phase is gradually restored. Among them, by calculating the difference between the unwrapped phase gradient and the wrapped phase gradient and then using integer multiples of to compensate for the jump to obtain the correction value. The wrapped phase is calculated by calculating the global quality score for the moiré pattern image in S3 and then generating it after judging that it is greater than the threshold. This is the prior art and will not be elaborated here. The maximum spanning tree path is the integration path based on the pixel weights and is generated by an algorithm. The path order of directly determines the accumulation order that ultimately affects the phase compensation. The phase period is the period length of the interference fringes and is the benchmark for jump compensation; in the displacement field calculation, the displacement field uses the interferometer principle to convert the phase difference into physical displacement. The conversion principle is that every phase change corresponds to displacement. The laser wavelength is a laser physical constant, and its value is usually , is the interference optical path coefficient, and its value is usually .

[0031] The present invention is further configured such that S5 specifically includes: Calculating the sub-pixel displacement gradient amplitude using the displacement field; Describing the crack profile based on the displacement gradient amplitude and the level set function, adding the coordinate sets of all detected crack boundaries to the crack profile point set, calculating the length, width and depth of the positions in the crack profile point set to obtain crack information. Specifically, when calculating the displacement field, with sub-pixel accuracy, finer displacement changes can be captured. The displacement gradient refers to the rate of change between adjacent pixels in the displacement field, which can reflect the local change trend in the image. Especially at the edges of cracks or defects, the displacement gradient usually shows obvious changes. The sub-pixel displacement gradient amplitude can be accurate to a higher precision than the pixel scale, thus identifying small cracks or damages in the image; the level set method is a mathematical tool for dealing with the evolution of curves or surfaces. Through the level set function, the crack boundaries in the image can be described and tracked. Its function is that by setting a specific value range, the crack region and the non-crack region can be effectively distinguished, and more accurate modeling and description of the cracks can be carried out; combining the displacement gradient information of the cracks in the displacement field and the level set function, the shape and profile of the cracks can be accurately described. The crack profile point set is composed of the detected crack boundary coordinates. Through these point sets, more refined geometric analysis of the cracks can be carried out; based on the crack profile point set, by calculating the Euclidean distance between the points in the crack profile point set, the length and width of the crack can be obtained. The Euclidean distance is used to measure the straight-line distance between two points. By analyzing the distribution of the crack boundary, the geometric dimensions of the crack can be accurately calculated; the depth of the crack is usually estimated by the maximum value of the displacement field or the gray level profile sampled vertically from the crack boundary. The crack depth is an important parameter affecting the structural safety. By quantifying the depth, the threat of the crack to the bridge structure can be evaluated.

[0032] The present invention is further configured such that the calculation logic of the displacement gradient amplitude: , where is the displacement gradient amplitude; The calculation logic of the crack profile description: , where is the level set function, is the curvature, is the gradient amplitude of the level set function; The construction logic of the crack profile point set: ; where is the crack profile point set; The length calculation logic: , where is the crack length, is the total number of contour points, is the point concentrated in the crack contour, is the next point concentrated in the crack contour, is the Euclidean distance between the crack contour points; Depth calculation logic: , where is the crack depth, is the constant factor, is the maximum value of the displacement field of all points in the contour point set; Width calculation logic: , where is the standard deviation, which is obtained by Gaussian fitting of the gray value profile of the image sampled along the normal vector direction of each contour point. Specifically, the displacement gradient amplitude measures the intensity of local deformation in the image, which varies greatly in the crack area. By calculating the displacement difference between adjacent pixels of each pixel point in the image, the gradient amplitude of the displacement field is obtained. The larger this value, the more it indicates that the image corresponds to the edge area of the crack or damage. represents the gradient of the displacement field in direction, that is, the displacement change rate of adjacent pixels of each pixel point in the horizontal direction. represents the gradient of the displacement field in direction, that is, the displacement change rate of adjacent pixels of each pixel point in the vertical direction; The level set function represents the "implicit representation" of the crack boundary in the image. It is a function used to describe and track the dynamic changes of the crack boundary, where: , represents the crack edge represents the area outside the crack, and the curvature represents the degree of bending of the crack boundary, and the calculation formula is: A higher curvature value corresponds to a more curved crack. The gradient amplitude of the level set function represents the change rate of the crack boundary, and the gradient amplitude at the boundary is larger. This formula describes the evolution process of the level set function over time, considering the curvature and displacement field of the crack. Among them, the first term promotes the crack boundary to tend to be smooth, while the second term promotes the crack boundary to expand along the displacement gradient direction, helping to identify the crack area. By solving this equation, the crack boundary can be dynamically tracked and the contour representation of the crack can be gradually optimized; The set of crack contour points represents all points belonging to the crack boundary in the image, which is obtained by screening the point set of the level set function ; The crack length By calculating the distances between consecutive points in the set of crack contour points and accumulating them, the length of the entire crack is obtained. The Euclidean distance is used to measure the straight-line distance between two points; the crack depth is obtained by finding the maximum displacement value in the set of crack contour points and then multiplying it by a constant factor. The constant factor is mainly used to adjust the quantization range of the depth, and its value usually depends on experiments or material properties the maximum displacement value of all points in the set of crack contour points; the crack width is obtained by performing Gaussian fitting on the gray-scale profile of the crack boundary to obtain the standard deviation and then multiplying it by a constant factor. This method is an existing technology and will not be elaborated here Example Two

[0033] Please refer to Figure 2 , the exemplary moiré phase unwrapping-based sub-pixel crack recognition system for bridges includes: Image acquisition and preprocessing module: A drone equipped with a high-resolution camera acquires images of the bridge surface and uploads them to the cloud system. Geometric correction and denoising processing are performed on the bridge surface images to obtain denoised images; Grating generation and registration module: According to the optimal period, a reference grating image is generated from the denoised image, and sub-pixel registration is used to obtain the registered reference grating image; Moiré synthesis and quality assessment module: Using the registered reference grating image to generate a moiré image, calculating the global quality score for the generated moiré image. For moiré images with a score greater than the scoring threshold, calculate the wrapped phase, otherwise require the drone to re-acquire; Phase unwrapping and displacement calculation module: Calculate the pixel weights for moiré images with a score greater than the scoring threshold, calculate the unwrapped phase based on the pixel weights, and use the interferometer principle to derive the displacement field of each pixel from the unwrapped phase; Crack recognition and quantification module: Analyze the magnitude of the displacement gradient in the displacement field, combine with the level set function to identify the crack area and quantify the width, length, and depth of the crack, providing quantitative information for structural damage assessment.

[0034] It should be noted that the moiré phase unwrapping-based sub-pixel crack recognition system for bridges provided in the above embodiments and the moiré phase unwrapping-based sub-pixel crack recognition method for bridges provided in the above embodiments belong to the same concept. The specific ways in which each module and unit perform operations have been described in detail in the method embodiments and will not be elaborated here. In practical applications, the moiré phase unwrapping-based sub-pixel crack recognition system for bridges provided in the above embodiments can, according to needs, allocate the above functions to different functional modules, that is, divide the internal structure of the system into different functional modules to complete all or part of the functions described above. This will not be limited here either.

[0035] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0036] It should be understood that the term "and / or" in this document is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood by referring to the context before and after.

[0037] In this application, "at least one" means one or more, and "a plurality" means two or more. "At least one of the following" or its similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or multiple.

[0038] It should be understood that in various embodiments of the present application, the magnitudes of the sequence numbers of the above processes do not imply the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0039] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Skilled professionals can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of this application.

[0040] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0041] In several embodiments provided in this application, it should be understood that the disclosed systems can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0042] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0043] In addition, the functional units in each embodiment of this application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0044] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0045] As described above, the above are only specific implementation manners of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. A method for sub-pixel recognition of bridge cracks based on moiré phase unwrapping, characterized in that Including: S1. The drone carries a high-resolution camera to collect bridge surface images and uploads them to the cloud system, and geometric correction and denoising processing are performed on the bridge surface images to obtain denoised images; S2. Generate a reference grating image from the denoised image according to the optimal period, and use sub-pixel registration to obtain the registered reference grating image; S3. Generate a moiré image using the registered reference grating image, calculate the global quality score for the generated moiré image, calculate the wrapped phase for the moiré image with a score greater than the scoring threshold, otherwise require the drone to re-collect; S4. Calculate the pixel weights for the moiré image with a score greater than the scoring threshold, calculate the unwrapped phase based on the pixel weights, and use the interferometer principle to derive the displacement field of each pixel for the unwrapped phase; S5. Analyze the displacement gradient amplitude in the displacement field, and combine the level set function to identify the crack area and quantify the width, length and depth of the crack, providing quantitative information for structural damage assessment.

2. The method for sub-pixel recognition of bridge cracks based on moiré phase unwrapping according to claim 1, characterized in that S2 specifically includes: Calculate the gradient amplitude of the denoised image, and optimize the grating period using the gradient amplitude to obtain the optimal period; Generate a reference grating image based on the optimal period; Calculate the displacement using the phase correlation method, and perform sub-pixel registration to obtain the registered reference grating image.

3. The method for sub-pixel identification of bridge cracks based on moiré phase unwrapping according to claim 2, wherein Gradient magnitude calculation logic: , where is the gradient magnitude, is the denoised image, is the continuous differential operator of the differential vector; Optimal period calculation logic: , where is the optimal period, is the grating period, is the sine function, is the angular frequency of the sine function; Reference grating calculation logic: , where is the reference grating image, and are respectively the base value and amplitude of the reference image, is the grating waveform sine generation function; Reference grating registration calculation logic: , where is the registered reference grating image, and are the image translation displacements calculated by the phase correlation method, is the bicubic interpolation function.

4. The method for sub-pixel recognition of bridge cracks based on moiré phase unwrapping according to claim 1, characterized in that S3 specifically includes: Multiply the registered reference grating image with the denoised image to synthesize moiré to obtain the moiré image; Calculate the global quality score for the generated moiré image using the local contrast and frequency domain energy ratio; Calculate the wrapped phase for the moiré image with a score greater than the scoring threshold, otherwise require the drone to re-collect.

5. The method for sub-pixel recognition of bridge cracks based on moiré phase unwrapping according to claim 4, characterized in that, Moiré pattern image generation logic: , where is the Moiré pattern image; Global mass fraction calculation logic: , where is the global mass fraction, and are the dimensions of the image, is the local contrast, is the frequency domain energy ratio, is the noise compensation factor, is the noise level; Local contrast calculation logic: , where is the local contrast, is the maximum gray value within the local area, is the minimum gray value within the local area; Frequency domain energy ratio calculation logic: , where is the frequency domain energy ratio, is the frequency domain representation of the image, is the frequency domain representation of the expected image.

6. The method for sub-pixel recognition of bridge cracks based on moiré phase unwrapping according to claim 1, characterized in that S4 specifically includes: Calculate the pixel weights by combining the global quality score, local contrast, and structural consistency; Based on the pixel weights, use the maximum spanning tree to calculate the path integral to obtain the unwrapped phase; Use the interferometer principle to derive the displacement field of each pixel for the unwrapped phase.

7. The method for sub-pixel identification of bridge cracks based on moiré phase unwrapping according to claim 6, wherein Pixel weight calculation logic: , where is the pixel weight, is the global quality score, is the local contrast, is the structural consistency, is the noise level; Path integral calculation logic: , where is the unwrapped phase, is the wrapped phase, is the unwrapped phase gradient, is the wrapped phase gradient, is the rounding function, is the maximum spanning tree path, is the phase period; Displacement field calculation logic: , where is the displacement field, is the laser wavelength. Specifically, the displacement field calculation uses the interferometer principle to convert the phase difference into physical displacement. For every phase change, it corresponds to displacement.

8. The method for sub-pixel recognition of bridge cracks based on moiré phase unwrapping according to claim 1, characterized in that S5 specifically includes: Calculate the sub-pixel displacement gradient amplitude using the displacement field; Describe the crack contour based on the displacement gradient amplitude and the level set function, add the coordinate sets of all detected crack boundaries to the crack contour point set, and calculate the length, width and depth of the positions in the crack contour point set to obtain crack information.

9. The method for sub-pixel identification of bridge cracks based on moiré phase unwrapping according to claim 8, wherein Displacement gradient magnitude calculation logic: , where is the displacement gradient magnitude; Calculation logic for crack profile description: , where is the level set function, is the curvature, is the gradient magnitude of the level set function; Crack contour point set construction logic: ; among which, is the crack contour point set; Length calculation logic: , where is the crack length, is the total number of contour points, is the point in the crack contour concentration, is the next point in the crack contour concentration, is the Euclidean distance between the crack contour points; Depth calculation logic: , where is the crack depth, is a constant factor, is the maximum value of the displacement field of all points in the contour point set; Width calculation logic: , where is the standard deviation, which is obtained by Gaussian fitting of the gray value profile of the image sampled from each contour point along the direction of the normal vector.

10. A bridge crack sub-pixel recognition system based on moiré phase unwrapping, which is used to implement the bridge crack sub-pixel recognition method based on moiré phase unwrapping according to any one of claims 1-9, characterized in that, Including: Image acquisition and preprocessing module: The drone carries a high-resolution camera to collect bridge surface images and uploads them to the cloud system, and geometric correction and denoising processing are performed on the bridge surface images to obtain denoised images; Grating generation and registration module: Generate a reference grating image from the denoised image according to the optimal period, and use sub-pixel registration to obtain the registered reference grating image; Moiré synthesis and quality assessment module: Generate a moiré image using the registered reference grating image, calculate the global quality score for the generated moiré image, calculate the wrapped phase for the moiré image with a score greater than the scoring threshold, otherwise require the drone to re-collect; Phase unwrapping and displacement calculation module: Calculate the pixel weights for moiré images with scores greater than the scoring threshold, obtain the unwrapped phase based on the pixel weights, and use the interferometer principle on the unwrapped phase to derive the displacement field of each pixel; Crack identification and quantification module: Analyze the magnitude of the displacement gradient in the displacement field, combine with the level set function to identify the crack area and quantify the width, length, and depth of the crack, providing quantitative information for structural damage assessment.

Images