Deep foundation pit crack detection method and equipment for deep foundation pit tunneling

Abstract

The invention relates to the technical field of image optimization recognition, in particular to a deep foundation pit crack detection method and device for deep foundation pit tunneling. According to the method, a deep foundation pit image and a motion vector of an image acquisition device are obtained; and blurred images in the deep foundation pit images are screened out, and rigid edges are identified. And screening out the smear edge by using the position relationship between other edges and the rigid edge, the pixel value change stability on the path and the correlation degree between the edge vector and the motion vector. And constructing a two-dimensional fuzzy kernel, and adaptively setting a fuzzy kernel value by using gradient information and position information of pixel values in the two-dimensional fuzzy kernel. And performing deblurring on the blurred image through the two-dimensional blurring kernel, and performing crack identification. The feature information in the blurred image is extracted and analyzed, the smear effect in the image is quantified, the blurred kernel in the image is determined in a self-adaptive mode, deblurring is carried out, the quality of the deep foundation pit image is improved, and the accuracy of a crack recognition result is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image optimization and recognition, and in particular to a deep foundation pit crack detection method and equipment used for deep foundation pit excavation. Background Art

[0002] Deep foundation pit engineering is a key link in infrastructure construction. During the excavation process, deep foundation pits are affected by factors such as soil stress changes, groundwater action, and construction technology, which can cause cracks in the foundation pit structure. The generation of cracks will not only reduce the stability and bearing capacity of the foundation pit, but may also cause serious safety accidents such as soil collapse and groundwater leakage, posing a huge threat to the safety of life and property of surrounding buildings, underground pipelines, and construction workers.

[0003] In the prior art, image processing methods can be used to monitor images in deep foundation pits and locate cracks. However, during the construction process, vibrations caused by large construction equipment such as excavator operations, pile driver vibrations, and earth-moving vehicles are inevitable environmental factors. The vibrations will also cause the image acquisition equipment to vibrate, resulting in blurred images. Due to the randomness of the vibrations, the degree of blur and the direction of the images at different times are random. If a fixed blur kernel is used to deblur the image, the image deblurring effect will be poor, affecting the accuracy of crack identification and detection. Summary of the invention

[0004] In order to solve the technical problem that the existing technology uses a fixed fuzzy kernel to deblur the deep foundation pit image, resulting in poor deblurring effect and affecting the accuracy of crack identification, the purpose of the present invention is to provide a deep foundation pit crack detection method and device for deep foundation pit excavation. The technical solution adopted is as follows: The present invention proposes a deep foundation pit crack detection method for deep foundation pit excavation, the method comprising: Obtaining a deep foundation pit image captured by an image acquisition device in the deep foundation pit; the image acquisition device is integrated with a sensor for detecting a motion vector generated by vibration; Obtain a blurred image in the deep foundation pit image when vibration occurs; select a rigid edge according to the position of each edge in the blurred image and the significance of the gradient information; For each edge, the first smear probability of each edge is obtained according to the positional relationship between the edge and the rigid edge and the stability of the change of the pixel value on the path; the second smear probability of each edge is obtained according to the degree of correlation between the edge vector from the rigid edge to the edge and the motion vector; the smear edge is screened out from all edges according to the first smear probability and the second smear probability; Obtain the smear distance and smear direction between the rigid edge and the smear edge; set the size of the two-dimensional blur kernel according to the smear distance, determine whether the pixel point is within the smear range within the two-dimensional blur kernel, and obtain the blur kernel value of the pixel point position according to the gradient of the pixel point and the size of the two-dimensional blur kernel; use the two-dimensional blur kernel to deblur the blurred image; The deblurred deep foundation pit image is used for crack identification.

[0005] Furthermore, the method for obtaining the blurred image includes: The deep foundation pit image is convolved using a Laplace operator to obtain a judgment image. If the energy value in the judgment image is lower than a preset energy threshold, the deep foundation pit image corresponding to the judgment image is used as a blurred image.

[0006] Furthermore, the rigid edge screening method includes: For each edge, the distance between the edge and the center point of the image is taken as the reference distance, and the ratio of the length of the edge to the reference distance is taken as the position morphological parameter of the edge; the ratio of the overall gradient amplitude on the edge to the grayscale variance is taken as the first gradient significance; the degree of perpendicularity between the overall gradient direction on the edge and the edge tangent direction is taken as the second gradient significance; the rigidity of each edge is obtained according to the position morphological parameters, the first gradient significance and the second gradient significance; the edge with the largest rigidity in the blurred image is selected as the rigid edge.

[0007] Furthermore, the method for obtaining the first smear probability includes: For each edge, the angle between the edge and the rigid edge is used as the position angle; each point on the edge is mapped to each point on the rigid edge, and the path between the two mapped points is used as the mapping path; the gradient amplitude sequence and the gradient direction angle sequence of the pixel points on the mapping path are obtained; the change stability is obtained according to the difference between the elements in the gradient amplitude sequence and the gradient direction angle sequence; the first ghosting probability is obtained according to the change stability and the position angle.

[0008] Furthermore, the method for obtaining the second smear probability includes: For each edge, the sum vector of the vectors formed by the mapping path pointing the edge in the direction of the rigid edge is used as the edge vector between the rigid edge and the edge; the sum of the moduli of the edge vector and the motion vector is used as the denominator, and the modulus of the difference vector between the edge vector and the motion vector is used as the numerator to obtain the second ghosting probability.

[0009] Furthermore, the method for screening the smear edge includes: The sum of the first smear probability and the second smear probability is used as the overall smear probability of each edge, and the edge with the largest overall smear probability is selected as the smear edge of the rigid edge.

[0010] Furthermore, the method for obtaining the smear distance and smear direction includes: The modulus of the edge vector between the rigid edge and the smear edge is used as the smear distance, and the direction of the edge vector between the rigid edge and the smear edge is used as the smear direction.

[0011] Furthermore, the method for obtaining the blur kernel value includes: For a pixel point within the two-dimensional blur kernel, the line segment from the pixel point to the center point of the two-dimensional blur kernel is used as the determination line segment. If the projection length of the determination line segment in the smear direction is less than or equal to the smear distance, the pixel point is judged to be within the smear range, the inverse of the smear distance is used as the distance weight, the gradient amplitude of the pixel point is negatively correlated and normalized, and then multiplied by the distance weight to obtain the blur kernel value at the pixel point position; if it is judged that the pixel point is no longer within the smear range, the blur kernel value at the pixel point position is set to 0.

[0012] Furthermore, the pre-trained crack detection neural network model is used to identify cracks in the deblurred deep foundation pit image.

[0013] The present invention also proposes a deep foundation pit crack detection device for deep foundation pit excavation, the device comprising: A deep foundation pit information acquisition module is used to obtain a deep foundation pit image acquired by an image acquisition device in the deep foundation pit; the image acquisition device is integrated with a sensor for detecting a motion vector generated by vibration; The first deep foundation pit image analysis module is used to obtain a fuzzy image in the deep foundation pit image when vibration occurs; a rigid edge is screened out according to the position of each edge in the fuzzy image and the significance of the gradient information; The second deep foundation pit image analysis module is used to obtain, for each edge, a first smear probability of each edge according to the positional relationship between the edge and the rigid edge and the stability of the pixel value on the path; obtain a second smear probability of each edge according to the degree of correlation between the edge vector from the rigid edge to the edge and the motion vector; and screen out the smear edge from all edges according to the first smear probability and the second smear probability; The image enhancement module is used to obtain, for each edge, a first smear probability of each edge according to the positional relationship between the edge and the rigid edge and the stability of the pixel value on the path; obtain a second smear probability of each edge according to the degree of correlation between the edge vector from the rigid edge to the edge and the motion vector; and screen out the smear edge from all edges according to the first smear probability and the second smear probability; The crack recognition module is used to identify cracks in the deblurred deep foundation pit image.

[0014] The present invention has the following beneficial effects: The present invention firstly considers that the rigid edge in the deep foundation pit image has the characteristics of high contrast and high significance information, which can be used as the degree of analyzing the smear. Therefore, the present invention selects a rigid edge in the blurred image according to the position and gradient information of each edge for smear analysis. The smear edge should have a positional relationship approximately parallel to the rigid edge, and the smear path can be regarded as the effect of rigid edge blur, so the pixel value information on the smear path should follow a certain rule, so the first smear probability can be obtained; further combined with the real motion vector of the device fed back by the sensor, the second smear probability is determined by the correlation between the edge vector and the motion vector, and then the smear edge of the rigid edge is determined. The two-dimensional blur kernel of each position in the blurred image can be adaptively determined by using the smear distance and the smear direction, and the deep foundation pit image with significant feature information can be obtained by deblurring the blurred image with the adaptive two-dimensional blur kernel, and then it is applied to crack recognition. The present invention extracts and analyzes the feature information in the blurred image, quantifies the smear effect in the image, and then adaptively determines the blur kernel in the image and deblurs it, thereby improving the quality of the deep foundation pit image and the accuracy of the crack recognition result. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 A flow chart of a deep foundation pit crack detection method for deep foundation pit excavation provided by one embodiment of the present invention; Figure 2 A schematic diagram of a mapping between a rigid edge and other edges provided by an embodiment of the present invention; Figure 3 A blurred image provided by an embodiment of the present invention; Figure 4 An embodiment of the present invention provides Figure 3 Deblurring effect diagram. DETAILED DESCRIPTION

[0017] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of a deep foundation pit crack detection method and device for deep foundation pit excavation proposed by the present invention, its specific implementation method, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0018] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0019] A specific scheme of a deep foundation pit crack detection method and device for deep foundation pit excavation provided by the present invention will be described in detail below in conjunction with the accompanying drawings.

[0020] See also Figure 1 , which shows a flow chart of a deep foundation pit crack detection method for deep foundation pit excavation provided by an embodiment of the present invention, the method comprising: Step S1: obtaining a deep foundation pit image captured by an image acquisition device in the deep foundation pit; the image acquisition device is integrated with a sensor for detecting a motion vector generated by vibration.

[0021] In an embodiment of the present invention, the image acquisition device is integrated with sensors such as an acceleration sensor and a gyroscope for detecting the intensity and direction of vibration. After the sensor collects the high-frequency data of the vibration, the acceleration data can be denoised and integrated through the Kalman filter algorithm, and the actual spatial position and posture of the image acquisition device at each moment can be determined in combination with the rotation angle information measured by the gyroscope. By comparing the position and posture of the image acquisition device at adjacent moments, it is possible to determine whether a vibration has occurred and determine the motion vector. The specific data processing algorithm is a technical means well known to those skilled in the art and will not be described in detail here. It should be noted that in order to facilitate subsequent image processing operations, after obtaining the motion vector of the device in reality, it is necessary to convert it into the image coordinate system to facilitate the subsequent correlation analysis of the edge vector.

[0022] The deep foundation pit image can be collected by using the image acquisition device, and the motion vector at each moment can be collected during vibration, wherein the time sequence corresponding to the motion vector and the deep foundation pit image is the same.

[0023] Step S2: Obtain a blurred image in the deep foundation pit image when vibration occurs; and select a rigid edge according to the position of each edge in the blurred image and the significance of the gradient information.

[0024] During the vibration process, the image acquisition device will move due to the vibration, resulting in the presence of smear blur in the captured image. Smear blur is manifested as the attenuation of the overall gradient in the image. Smear weakens the significant edge information in the image, resulting in unclear gradients and smoothed edge details, that is, the high-frequency components in the image will be reduced. Therefore, the blurred image is screened out from the deep foundation pit image, that is, the time corresponding to the blurred image is the time when the deep foundation pit has obvious vibration.

[0025] Preferably, in an embodiment of the present invention, blurred images in deep foundation pit images can be automatically screened out, because the high-frequency information in blurred images will be significantly reduced. Therefore, an embodiment of the present invention uses a Laplace operator to convolve the deep foundation pit image to obtain a judgment image. For the judgment image, if the deep foundation pit image corresponding to the judgment image is a blurred image, its high-frequency components are relatively small, and the pixel value changes smoothly, and the obtained judgment image will have a smaller energy value. Therefore, an energy threshold is set in an embodiment of the present invention. If the energy value in the judgment image is lower than a preset energy threshold, the deep foundation pit image corresponding to the judgment image is used as a blurred image. In an embodiment of the present invention, the energy threshold is set to 1200. The specific Laplace operator convolution and the method for obtaining image energy are technical means well known to those skilled in the art, and will not be elaborated here.

[0026] It should be noted that, in the embodiment of the present invention, if it is determined that the deep foundation pit image at a certain moment is a blurred image, the corresponding motion vector collected by the image acquisition device can be used for subsequent step analysis.

[0027] In a blurred image, the essence of the existing smear is the linear diffusion of the original edge in the direction of vibration. The intensity of the diffusion is positively correlated with the vibration amplitude and frequency, and the direction of the diffusion is also close to the vibration direction. In order to accurately analyze the degree of smear in a blurred image, it is necessary to use an edge with significant information as a representative edge for analysis, that is, the rigid edge extracted by the embodiment of the present invention. The rigid edge is an edge with prominent information under the original engineering structure in a deep foundation pit. It has an obvious high contrast. Before and after the vibration occurs, the rigid edge is not a clear and significant edge in the image. The smear will cause a blurring effect on the edge. The originally insignificant edge will be more blurred due to the influence of the smear. The edge formed by the smear is more distorted than the original edge, and it is not easy to perform reference analysis. The rigid edge has a strong information reference in the image because of its significant edge feature, and the blurring and distortion effects caused by the smear are easier to observe. Therefore, the rigid edge can be used as a reference edge for analyzing the smear.

[0028] For each edge in the blurred image, the closer it is to the middle area and the longer the edge length is, the more conducive the position and the size of the edge are to analyzing the degree of smearing; and the greater the significance of the gradient information on the edge, the more significant the edge information is, and the more likely it is a rigid edge. Therefore, the embodiment of the present invention selects a rigid edge for subsequent analysis based on the position of each edge in the blurred image and the significance of the gradient information.

[0029] Preferably, in an embodiment of the present invention, the method for screening rigid edges includes: For each edge, the distance between the edge and the center of the image is taken as the reference distance, and the ratio of the edge length to the reference distance is taken as the position morphology parameter of the edge. The smaller the reference distance, the closer the edge is to the center of the image. The larger the edge length, the easier it is to observe the smear. The larger the position morphology parameter, the more likely the edge is a rigid edge.

[0030] The ratio of the overall gradient amplitude to the grayscale variance on the edge is taken as the first gradient significance. That is, the smaller the grayscale variance, the more stable the grayscale information on the edge. At the same time, the larger the overall gradient amplitude, the clearer the edge. The larger the first gradient significance, the more likely the edge is a rigid edge.

[0031] The second gradient significance is determined by the degree of perpendicularity between the overall gradient direction on the edge and the edge tangent direction. That is, the second gradient significance is analyzed based on the gradient direction. Since a rigid edge is an edge with significant gradient information, its gradient direction should be obviously perpendicular to the edge tangent, indicating that the edge pixel at this position has a significant gradient change. Therefore, the greater the degree of perpendicularity, the greater the second gradient significance, and the more likely the edge is a rigid edge.

[0032] The rigidity of each edge is obtained according to the position morphological parameter, the first gradient significance and the second gradient significance; the edge with the largest rigidity in the blurred image is selected as the rigid edge. In the embodiment of the present invention, after the position morphological parameter, the first gradient significance and the second gradient significance are quantified respectively, the product of the three is taken as the rigidity.

[0033] In the embodiment of the present invention, the overall gradient direction is the direction corresponding to the sum of the unit vectors corresponding to the gradient direction of each point on the edge, and similarly, the overall gradient amplitude is the average value of the gradient amplitude of each point on the edge. The acute angle between the overall gradient direction and the edge tangent direction is obtained, and the sine value of the angle is used as the degree of perpendicularity, that is, the closer the angle is to 90 degrees, the larger the sine value is, and the greater the degree of perpendicularity is, indicating that the gradient direction is more perpendicular to the edge tangent.

[0034] Step S3: For each edge, the first trailing probability of each edge is obtained according to the positional relationship between the edge and the rigid edge, and the stability of the change of pixel values ​​on the path; the second trailing probability of each edge is obtained according to the degree of correlation between the edge vector from the rigid edge to the edge and the motion vector; the trailing edges are screened out from all edges according to the first trailing probability and the second trailing probability.

[0035] After the rigid edge is obtained in step S2, its corresponding trailing edge can be further determined. The trailing edge is the edge after the trailing produced by the rigid edge after vibration, that is, the trailing edge is the boundary of the rigid edge trailing. The boundary is generated by the trailing of the rigid edge, so the boundary has similar edge trends and pixel value characteristics as the rigid edge. And between the trailing edge and the rigid edge is the trailing path, which can be regarded as generated by the rigid edge moving in the direction of vibration, and the pixel value changes on the path should be relatively similar and stable.

[0036] Therefore, in order to filter out the smear edge of the rigid edge in the blurred image, for each edge, the first smear probability of each edge is first obtained based on the positional relationship between the edge and the rigid edge and the stability of the change of the pixel value on the path. That is, the more parallel an edge is to the rigid edge, the more likely it is that the edge is a smear edge generated by the rigid edge moving in the vibration direction; the greater the stability of the change of the pixel value on the path, the more likely it is that the edge is the boundary of the rigid edge smear.

[0037] Furthermore, in combination with the motion vector obtained by the image acquisition device, since the motion vector has a clear direction and modulus, it represents the actual motion of the image acquisition device after being affected by vibration, and the image field of view moves with the movement of the device, thereby generating a smear in the image. That is, the direction of the smear in the image is related to the movement direction of the device, so the embodiment of the present invention further obtains the edge vector between the rigid edge and each edge, and determines the degree of correlation between the edge vector and the motion vector. The greater the degree of correlation, the greater the probability of the second smear of the edge, and the more likely the edge is a smear edge.

[0038] By combining the first smear probability and the second smear probability, the smear edge can be screened out from all the edges.

[0039] Preferably, in an embodiment of the present invention, the method for obtaining the first smear probability includes: For each edge, the angle between the edge and the rigid edge is used as the position angle. That is, the smaller the position angle is, the more parallel the two edges are, and the more likely the edge is a smear edge of the rigid edge. It should be noted that the method for obtaining the angle of the edge trend is to fit the two edges separately with a straight line, obtain the angle of the fitting line relative to the horizontal direction, and use the angle difference between the two fitting lines as the position angle.

[0040] To further analyze whether the area between the edge and the rigid edge is a smear area, each point on the edge is mapped to each point on the rigid edge, and the path between the two mapped points is used as the mapping path. Figure 2 , which shows a schematic diagram of a mapping of a rigid edge to other edges provided by an embodiment of the present invention. Figure 2 The edge i1 on the left is relatively parallel to the rigid edge and has a higher probability of being a trailing edge; the edge i2 on the right is an edge with a larger angle to the rigid edge and has a lower probability of being a trailing edge. By mapping, we can analyze the information changes on the path and confirm whether it is a path in the trailing area.

[0041] The gradient amplitude sequence and gradient direction angle sequence of the pixel points on the mapping path are obtained. If the mapping path is a path of the smear area, the gradient information on the path is generated by a rigid edge, and the gradient amplitude and gradient direction are relatively stable, so the change stability is obtained according to the difference between the elements in the gradient amplitude sequence and the gradient direction angle sequence. In an embodiment of the present invention, the standard deviations of the two sequences are multiplied, negative correlation mapping is performed, and normalization is performed to obtain the change stability.

[0042] The first smear probability is obtained according to the change stability and the position angle. In the embodiment of the present invention, the greater the change stability and the smaller the position angle, the greater the first smear probability of the edge, so the value of the position angle is negatively correlated and normalized, and then multiplied by the change stability to obtain the first smear probability.

[0043] It should be noted that negative correlation mapping and normalization are mathematical means well known to those skilled in the art, for example, they can be implemented by an exponential function with a natural constant as the base, and the opposite number of the data is used as the power of the exponential function, and the function mapping result is the result after negative correlation mapping and normalization. Those skilled in the art can flexibly select other algorithms for negative correlation mapping and normalization, which are not limited or elaborated here, and the contents of the related algorithms involved in the subsequent embodiments of the present invention will not be elaborated.

[0044] Preferably, in an embodiment of the present invention, the method for obtaining the second smear probability includes: For each edge, since there are multiple mapping paths, the sum vector of the vectors formed by the mapping paths that point to the direction of the rigid edge is taken as the edge vector between the rigid edges.

[0045] If a significant vibration occurs at a certain moment, there will be a significant distance between the rigid edge and the trailing edge in the blurred image, and the trailing direction will be significantly correlated with the vibration direction. In order to quantify this correlation, the edge vector in the embodiment of the present invention is the direction in which the edge points to the rigid edge, that is, if the edge is a trailing edge, the direction of the edge vector should be opposite to the direction of the motion vector. Therefore, the sum of the modulus lengths of the edge vector and the motion vector is used as the denominator, and the modulus length of the difference vector between the edge vector and the motion vector is used as the numerator to obtain the second trailing probability. If the edge is a trailing edge, the denominator and the numerator are equal, that is, the second trailing probability is 1, otherwise if the edge is not a trailing edge, the direction between the edge vector and the motion vector is not opposite, and the obtained difference vector modulus length will be less than the sum of the two adjacent modulus lengths, that is, the numerator is less than the denominator, and the second trailing probability becomes smaller. That is, the greater the second trailing probability, the greater the correlation between the edge vector and the adjacent motion, and the more likely the edge is a trailing edge.

[0046] In the embodiment of the present invention, after obtaining the first and second smear probabilities, the sum of the first and second smear probabilities is used as the overall smear probability of each edge, and the edge with the largest overall smear probability is selected as the smear edge of the rigid edge.

[0047] Step S4: Obtain the smear distance and smear direction between the rigid edge and the smear edge; set the size of the two-dimensional blur kernel according to the smear distance, determine whether the pixel point is within the smear range within the two-dimensional blur kernel, and obtain the blur kernel value of the pixel point position according to the gradient of the pixel point and the size of the two-dimensional blur kernel; use the two-dimensional blur kernel to deblur the blurred image.

[0048] In a blurred image, the rigid edge and the corresponding smear edge are determined, the distance between the two is the smear distance, and the direction from the rigid edge to the smear edge is the smear direction. The smear direction and smear distance can be used to characterize the degree of smear in the blurred image, so that a two-dimensional blur kernel can be adaptively constructed. The size of the two-dimensional blur kernel can be set according to the smear distance, that is, if the smear length is L, the size of the two-dimensional blur kernel is L×L.

[0049] Furthermore, the blur kernel value of the two-dimensional blur kernel at each position in the blurred image can be set. In the two-dimensional blur kernel, it is first necessary to determine whether the pixel is within the smear range. If it is within the smear range, the blur kernel value needs to be determined based on its corresponding gradient information combined with the size of the two-dimensional blur kernel. If the gradient amplitude of the pixel position is large, the blur kernel weight at this position should be reduced accordingly, thereby reducing the smoothing of the feature during the deblurring process and avoiding information distortion. After setting each pixel position, an effective two-dimensional blur kernel can be determined to deblur the blurred image.

[0050] Specifically, in the embodiment of the present invention, the modulus of the edge vector between the rigid edge and the smear edge is used as the smear distance, and the direction of the edge vector between the rigid edge and the smear edge is used as the smear direction.

[0051] In one embodiment of the present invention, a method for obtaining a blur kernel value includes: For a pixel point in the two-dimensional blur kernel, the line segment from the pixel point to the center point of the two-dimensional blur kernel is used as the judgment line segment. If the projection length of the judgment line segment in the smear direction is less than or equal to the smear distance, the pixel point is judged to be within the smear range. Under the linear change caused by the smear, the edge diffusion within the smear range is uniform, so the reciprocal of the smear distance is used as the distance weight. The gradient amplitude of the pixel point is negatively correlated and normalized, and then multiplied by the distance weight to obtain the blur kernel value at the pixel point position. That is, the gradient information weight of the pixel value position is obtained after the gradient amplitude is negatively correlated and normalized. The larger the gradient amplitude, the less smoothing is required at this position. The blur kernel value at the pixel point position can be obtained by multiplying it with the distance weight.

[0052] If it is determined that the pixel point is no longer within the smear range, the blur kernel value at the pixel point position is set to 0.

[0053] In the embodiment of the present invention, after the two-dimensional blur kernel at each position in the blurred image is determined, a non-blind-area convolution operation can be performed on the blurred image according to the blur kernel to remove blur information in the image, thereby obtaining a deblurred image. It should be noted that the core of deblurring is to use the blur kernel and the blurred image to perform an inverse convolution operation to restore the original clear appearance of the image. Specifically, for the blurred image, the inverse process of convolving it with the blur kernel is to perform the inverse process of the convolution operation with the blur kernel. Since the blurring of the image is essentially the diffusion of each pixel point of the image in space according to the pattern of the blur kernel, deblurring is to re-aggregate the diffused pixel information according to the reverse pattern of the blur kernel. Please refer to Figure 3 and Figure 4 , Figure 3 A blurred image provided by an embodiment of the present invention, Figure 4 An embodiment of the present invention provides Figure 3 The deblurred effect diagram. After deblurring, a deep foundation pit image with significant feature information can be obtained.

[0054] Step S5: performing crack identification on the deblurred deep foundation pit image.

[0055] After obtaining a deep foundation pit image with significant feature information, the crack detection technology in the prior art can be used to identify cracks. In the embodiment of the present invention, a pre-trained crack detection neural network model is used to identify cracks in the deblurred deep foundation pit image. The crack detection neural network is a convolutional neural network, which uses its multi-layer convolutional layer and pooling layer to automatically extract rich texture, shape and other features in the image. Next, a large amount of annotated image data containing normal areas of deep foundation pits and various crack conditions is collected, and the neural network is trained as a training set. During the training process, the neural network continuously adjusts internal parameters to learn the characteristic patterns of cracks in the image, so that the prediction results output by the model are as close as possible to the annotated real crack conditions. When the model training is completed and reaches a certain accuracy, the deblurred deep foundation pit image is input into the neural network model. Finally, the prediction results of whether there are cracks in the image and the location, type and other information of the cracks are output. It can quickly and accurately judge the actual condition of cracks in deep foundation pits, provide key basis for engineering safety assessment and subsequent processing, and effectively improve the efficiency and reliability of deep foundation pit crack monitoring.

[0056] In summary, the present invention obtains a deep foundation pit image and a motion vector of an image acquisition device. The blurred image in the deep foundation pit image is screened out, and the rigid edge is identified. The trailing edge is screened out by utilizing the positional relationship between other edges and the rigid edge, the stability of the pixel value change on the path, and the degree of correlation between the edge vector and the motion vector. The trailing distance and trailing direction can be determined based on the rigid edge and the corresponding trailing edge, and then a two-dimensional blur kernel is constructed, and the gradient information and position information of the pixel value in the two-dimensional blur kernel are used to adaptively set the blur kernel value. The blurred image is deblurred by the two-dimensional blur kernel, and then crack identification is performed. The present invention quantifies the trailing effect in the image by extracting and analyzing the feature information in the blurred image, and then adaptively determines the blur kernel in the image and deblurs it, thereby improving the quality of the deep foundation pit image and the accuracy of the crack identification results.

[0057] Based on the same invention, the present invention also proposes a deep foundation pit crack detection device for deep foundation pit excavation, the device comprising: A deep foundation pit information acquisition module is used to obtain a deep foundation pit image acquired by an image acquisition device in the deep foundation pit; the image acquisition device is integrated with a sensor for detecting a motion vector generated by vibration; The first deep foundation pit image analysis module is used to obtain a fuzzy image in the deep foundation pit image when vibration occurs; a rigid edge is screened out according to the position of each edge in the fuzzy image and the significance of the gradient information; The second deep foundation pit image analysis module is used to obtain, for each edge, a first smear probability of each edge according to the positional relationship between the edge and the rigid edge and the stability of the pixel value on the path; obtain a second smear probability of each edge according to the degree of correlation between the edge vector from the rigid edge to the edge and the motion vector; and screen out the smear edge from all edges according to the first smear probability and the second smear probability; The image enhancement module is used to obtain, for each edge, a first smear probability of each edge according to the positional relationship between the edge and the rigid edge and the stability of the pixel value on the path; obtain a second smear probability of each edge according to the degree of correlation between the edge vector from the rigid edge to the edge and the motion vector; and screen out the smear edge from all edges according to the first smear probability and the second smear probability; The crack recognition module is used to identify cracks in the deblurred deep foundation pit image.

[0058] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0059] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A deep foundation pit crack detection method for deep foundation pit excavation, characterized in that: The method comprises: Obtaining a deep foundation pit image captured by an image acquisition device in the deep foundation pit; the image acquisition device is integrated with a sensor for detecting a motion vector generated by vibration; Obtain a blurred image in the deep foundation pit image when vibration occurs; select a rigid edge according to the position of each edge in the blurred image and the significance of the gradient information; For each edge, the first smear probability of each edge is obtained according to the positional relationship between the edge and the rigid edge and the stability of the change of the pixel value on the path; the second smear probability of each edge is obtained according to the degree of correlation between the edge vector from the rigid edge to the edge and the motion vector; the smear edge is screened out from all edges according to the first smear probability and the second smear probability; Obtain the smear distance and smear direction between the rigid edge and the smear edge; set the size of the two-dimensional blur kernel according to the smear distance, determine whether the pixel point is within the smear range within the two-dimensional blur kernel, and obtain the blur kernel value of the pixel point position according to the gradient of the pixel point and the size of the two-dimensional blur kernel; use the two-dimensional blur kernel to deblur the blurred image; The deblurred deep foundation pit image is used for crack identification.

2. A deep foundation pit crack detection method for deep foundation pit excavation according to claim 1, characterized in that: The method for obtaining the blurred image includes: The deep foundation pit image is convolved using a Laplace operator to obtain a judgment image. If the energy value in the judgment image is lower than a preset energy threshold, the deep foundation pit image corresponding to the judgment image is used as a blurred image.

3. A deep foundation pit crack detection method for deep foundation pit excavation according to claim 1, characterized in that: The rigid edge screening method comprises: For each edge, the distance between the edge and the center point of the image is taken as the reference distance, and the ratio of the length of the edge to the reference distance is taken as the position morphological parameter of the edge; the ratio of the overall gradient amplitude on the edge to the grayscale variance is taken as the first gradient significance; the degree of perpendicularity between the overall gradient direction on the edge and the edge tangent direction is taken as the second gradient significance; the rigidity of each edge is obtained according to the position morphological parameters, the first gradient significance and the second gradient significance; the edge with the largest rigidity in the blurred image is selected as the rigid edge.

4. A deep foundation pit crack detection method for deep foundation pit excavation according to claim 1, characterized in that: The method for obtaining the first smear probability includes: For each edge, the angle between the edge and the rigid edge is used as the position angle; each point on the edge is mapped to each point on the rigid edge, and the path between the two mapped points is used as the mapping path; the gradient amplitude sequence and the gradient direction angle sequence of the pixel points on the mapping path are obtained; the change stability is obtained according to the difference between the elements in the gradient amplitude sequence and the gradient direction angle sequence; the first ghosting probability is obtained according to the change stability and the position angle.

5. A deep foundation pit crack detection method for deep foundation pit excavation according to claim 4, characterized in that: The method for obtaining the second smear probability includes: For each edge, the sum vector of the vectors formed by the mapping path pointing the edge in the direction of the rigid edge is used as the edge vector between the rigid edge and the edge; the sum of the moduli of the edge vector and the motion vector is used as the denominator, and the modulus of the difference vector between the edge vector and the motion vector is used as the numerator to obtain the second ghosting probability.

6. A deep foundation pit crack detection method for deep foundation pit excavation according to claim 1, characterized in that: The method for screening the smear edge comprises: The sum of the first smear probability and the second smear probability is used as the overall smear probability of each edge, and the edge with the largest overall smear probability is selected as the smear edge of the rigid edge.

7. A deep foundation pit crack detection method for deep foundation pit excavation according to claim 5, characterized in that: The method for obtaining the smear distance and the smear direction includes: The modulus of the edge vector between the rigid edge and the smear edge is used as the smear distance, and the direction of the edge vector between the rigid edge and the smear edge is used as the smear direction.

8. A deep foundation pit crack detection method for deep foundation pit excavation according to claim 1, characterized in that: The method for obtaining the fuzzy kernel value includes: For a pixel point within the two-dimensional blur kernel, the line segment from the pixel point to the center point of the two-dimensional blur kernel is used as the determination line segment. If the projection length of the determination line segment in the smear direction is less than or equal to the smear distance, the pixel point is judged to be within the smear range, the inverse of the smear distance is used as the distance weight, the gradient amplitude of the pixel point is negatively correlated and normalized, and then multiplied by the distance weight to obtain the blur kernel value at the pixel point position; if it is judged that the pixel point is no longer within the smear range, the blur kernel value at the pixel point position is set to 0.

9. A deep foundation pit crack detection method for deep foundation pit excavation according to claim 1, characterized in that: A pre-trained crack detection neural network model is used to identify cracks in the deblurred deep foundation pit images.

10. A deep foundation pit crack detection device for deep foundation pit excavation, characterized in that: The device comprises: A deep foundation pit information acquisition module is used to obtain a deep foundation pit image acquired by an image acquisition device in the deep foundation pit; the image acquisition device is integrated with a sensor for detecting a motion vector generated by vibration; The first deep foundation pit image analysis module is used to obtain a fuzzy image in the deep foundation pit image when vibration occurs; a rigid edge is screened out according to the position of each edge in the fuzzy image and the significance of the gradient information; The second deep foundation pit image analysis module is used to obtain, for each edge, a first smear probability of each edge according to the positional relationship between the edge and the rigid edge and the stability of the pixel value on the path; obtain a second smear probability of each edge according to the degree of correlation between the edge vector from the rigid edge to the edge and the motion vector; and screen out the smear edge from all edges according to the first smear probability and the second smear probability; The image enhancement module is used to obtain, for each edge, a first smear probability of each edge according to the positional relationship between the edge and the rigid edge and the stability of the pixel value on the path; obtain a second smear probability of each edge according to the degree of correlation between the edge vector from the rigid edge to the edge and the motion vector; and screen out the smear edge from all edges according to the first smear probability and the second smear probability; The crack recognition module is used to identify cracks in the deblurred deep foundation pit image.

Images