CN115375924A - A bridge health monitoring method and system based on image recognition - Google Patents

Abstract

The invention relates to the technical field of bridge health monitoring, and discloses a bridge health monitoring method and system based on image recognition. The method first identifies the image of the bridge to be monitored, and uses a discrete centroid search algorithm to extract the sub Pixel-level displacement time-history response, obtain the modal parameters of the micro-vibration of the bridge through the Hankel dynamic mode decomposition method, establish the modal vibration calculation model according to the bridge modal parameters, and identify the damage position of the bridge and the bridge according to the modal vibration calculation model degree of damage. In this way, the method based on image recognition can accurately identify the damage location and damage degree of the bridge, and timely discover the potential safety hazards of the bridge.

Description

A bridge health monitoring method and system based on image recognition

technical field

The invention relates to the technical field of bridge health monitoring, in particular to a bridge health monitoring method and system based on image recognition.

Background technique

Bridge structural vibration measurement is the key to bridge structural health monitoring (Structure Health Monitoring, SHM). Modal parameters (natural frequency, mode shape and damping ratio, etc.) are important indicators to reflect the health of the structure. The changes in the modal parameters can be used to identify bridge structure damage and state changes, and evaluate the performance of the bridge structure. During the operation of the bridge, vibrations will inevitably occur. How to quickly measure the micro-vibration response of the bridge is the premise to ensure the safe operation of the bridge, and it is also a powerful guarantee for real-time monitoring of the health of the bridge.

At present, the commonly used instrument for measuring the vibration characteristics of bridges is the acceleration sensor, but the acceleration sensor has the disadvantages of high cost, difficult installation, limited measuring points, low measurement accuracy and poor real-time performance, which makes it difficult to meet the real-time monitoring needs of bridge dynamic response. Other conventional measurement methods such as levels, dial gauges and total stations are difficult to perform dynamic measurements. Although GPS (Global Positioning System, Global Positioning System) can achieve dynamic measurements, the commissioning and installation are very cumbersome, and the complex bridge regional working environment and Factors such as satellites and weather will affect the accuracy of the measurement and the time of the measurement. Most of the existing non-contact vibration measurement methods based on computer vision technology are only suitable for scenes with large structural vibration displacement amplitudes, and are difficult to apply to the micro-vibration measurement of structures, and the micro-vibration signals of bridges just contain important information. Therefore, the accuracy of the existing bridge monitoring method is low, and the potential safety hazards of the bridge cannot be found in time.

Contents of the invention

The invention provides a bridge health monitoring method and system based on image recognition to solve the problem that the existing bridge monitoring method has low precision and cannot detect potential safety hazards in the bridge in time.

In order to achieve the above object, the present invention is achieved through the following technical solutions:

In a first aspect, the present invention provides a bridge health monitoring method based on image recognition, including:

Obtain a first image of the bridge to be monitored, and preprocess the first image to obtain a second image;

performing spatial decomposition on the second image to obtain a corresponding target image sequence;

calculating the centroid position sub-pixel coordinates of the continuous image frames in the target image sequence, and determining the real displacement time-history response of the bridge to be monitored according to the centroid position sub-pixel coordinates;

Utilize the Hankel dynamic mode decomposition method to extract bridge modal parameters from the real displacement time-history response, and the bridge modal parameters include natural frequency, mode shape and damping ratio;

A modal vibration calculation model is established according to the bridge modal parameters, and a bridge damage location and a bridge damage degree are identified according to the modal vibration calculation model.

In a second aspect, the present invention provides a bridge health monitoring system based on image recognition, which includes a memory, a processor, and a computer program stored on the memory and operable on the processor, and the computer program is implemented when the processor executes the computer program. The steps of the method described in the first aspect above.

Beneficial effect:

The bridge health monitoring method based on image recognition provided by the present invention firstly identifies the image of the bridge to be monitored, uses the discrete centroid search algorithm to extract the sub-pixel level displacement time-history response of the bridge micro-vibration from continuous image frames, and uses Hankel dynamic mode decomposition The modal parameters of the micro-vibration of the bridge are obtained by using the method, the modal vibration calculation model is established according to the bridge modal parameters, and the bridge damage location and bridge damage degree are identified according to the modal vibration calculation model. In this way, the image recognition-based method can accurately identify the damage location and damage degree of the bridge, and discover the potential safety hazards of the bridge in time.

In a preferred solution, the final target image sequence is obtained by denoising the initial target image sequence, which enhances measurement accuracy and facilitates obtaining accurate real displacement time-history responses and modal parameters.

In the preferred scheme, by dividing each image frame in the enlarged target image sequence into grids of multiple cropped regions, using the Otsu threshold segmentation algorithm in each grid to calculate the centroid of the discretized object, and calculate all continuous The sub-pixel-level coordinates of the centroid position of the image frame can obtain a more accurate displacement time-history response.

Description of drawings

Fig. 1 is one of the flowcharts of a bridge monitoring method based on image recognition in a preferred embodiment of the present invention;

Fig. 2 is the second flow chart of a bridge monitoring method based on image recognition in a preferred embodiment of the present invention;

Fig. 3 is a schematic diagram of centroid positions of two consecutive frames of images in a preferred embodiment of the present invention;

Fig. 4 is a schematic diagram of the acquisition principle of the real dynamic displacement time-history response of the micro-vibration of the bridge according to the preferred embodiment of the present invention.

Detailed ways

The technical solution of the present invention is described clearly and completely below, obviously, the described embodiments are only some embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

Unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the usual meanings understood by those skilled in the art to which the present invention belongs. "First", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Likewise, words like "a" or "one" do not denote a limitation in quantity, but indicate that there is at least one. Words such as "connected" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "Down", "Left", "Right" and so on are only used to indicate the relative positional relationship. When the absolute position of the described object changes, the relative positional relationship also changes accordingly.

Please refer to Figure 1-Figure 2, a bridge health monitoring method based on image recognition provided by this application, including:

Obtain a first image of the bridge to be monitored, and preprocess the first image to obtain a second image;

performing spatial decomposition on the second image to obtain a corresponding target image sequence;

Calculate the sub-pixel coordinates of the centroid position of the continuous image frames in the target image sequence, and determine the real displacement time-history response of the bridge to be monitored according to the sub-pixel coordinates of the centroid position;

The modal parameters of the bridge are extracted from the real displacement time-history response by using the Hankel dynamic mode decomposition method. The modal parameters of the bridge include natural frequency, mode shape and damping ratio;

The modal vibration calculation model is established according to the bridge modal parameters, and the bridge damage location and bridge damage degree are identified according to the modal vibration calculation model.

In this embodiment, the acquisition of the first image of the bridge to be monitored may be obtained by shooting with a high-speed camera.

The above-mentioned image recognition-based bridge health monitoring method first identifies the image of the bridge to be monitored, and uses the discrete centroid search algorithm to extract the sub-pixel level displacement time-history response of the bridge’s micro-vibration from continuous image frames, and obtains it through the Hankel dynamic mode decomposition method. The modal parameters of bridge micro-vibration, the modal vibration calculation model is established according to the bridge modal parameters, and the bridge damage location and bridge damage degree are identified according to the modal vibration calculation model. In this way, the image recognition-based method can accurately identify the damage location and damage degree of the bridge, and discover the potential safety hazards of the bridge in time.

Optionally, preprocessing the first image to obtain the second image includes:

The first image is rotated, cropped and scaled to obtain the second image.

In this optional implementation, digital image processing software can be used to perform preprocessing such as rotation, cropping, and scaling on the first image. In this way, the noise in the image sequence can be preliminarily eliminated through preprocessing, avoiding the use of broadband phase motion. When the magnification algorithm performs magnification processing, the noise is also amplified in equal proportions to eliminate the influence of noise on the image magnification result.

Optionally, performing spatial decomposition on the second image to obtain a corresponding target image sequence, including:

performing spatial decomposition on the second image to obtain an initial target image sequence, where the initial target image sequence includes a first residual part, different frequency baseband phase information and a second residual part;

Use the two-dimensional Gabor wavelet filter to extract the texture features of the initial target image sequence, and perform translation processing on the image information in the target image sequence except for the texture features, so as to complete the noise reduction process on the initial target image sequence and obtain the final target image sequence.

In this optional embodiment, the second image is decomposed in space using a complex steerable pyramid (CSP), and image sequences of different scales, directions, and positions are obtained, so that the bridge to be tested can be realized by the image sequence. Expression of local phase versus local motion in microvibration videos of structures. It should be pointed out that the initial target image sequence obtained after decomposition includes the first residual part, different frequency baseband phase information and the second residual part, wherein the first residual part refers to the high-pass residual part, and the second residual part refers to the low-pass residual part, and the baseband phase information of different frequencies refers to the baseband phase information of different frequencies in the middle.

In the embodiment of the present application, after obtaining the initial target image sequence, it is necessary to further perform noise reduction processing on the initial target image sequence to obtain the image sequence to be measured. Specifically, the initial target is extracted using a two-dimensional Gabor wavelet filter texture features of the image sequence, and perform translation processing on image information other than the texture features in the initial target image sequence to complete noise reduction processing on the initial target image sequence to obtain a final target image sequence. It should be pointed out that the two-dimensional Gabor wavelet filter is sensitive to the edge information of the image and can provide good direction selection and scale selection characteristics. In addition, the two-dimensional Gabor wavelet filter can be used for spatial filtering. It can be understood that using the two-dimensional Gabor wavelet filter for noise reduction processing effectively avoids artifacts generated by noise during motion amplification.

In this embodiment, the two-dimensional Gabor wavelet filter is a sine function modulated by a Gaussian function, and its complex expression is as follows:

；（1）

;(1)

The real part looks like this:

；（2）

;(2)

The imaginary part is as follows;

；（3）

;(3)

表示正弦函数的波长；

表示调谐函数的相位偏移量；

决定二维Gabor函数的空间长宽比；

为高斯函数标准差，决定二维Gabor滤波器核可接受区域的大小；

表示二维Gabor小波滤波器核的方向，且

；

和

为空间位置变量；exp表示指数计算；i表示虚数单位，

。In the formula,

Indicates the wavelength of the sine function;

Indicates the phase offset of the tuning function;

Determine the spatial aspect ratio of the two-dimensional Gabor function;

It is the standard deviation of the Gaussian function, which determines the size of the acceptable area of the two-dimensional Gabor filter kernel;

Indicates the direction of the two-dimensional Gabor wavelet filter kernel, and

;

and

is the space position variable; exp means exponent calculation; i means imaginary number unit,

.

和

分别表示视频图像序列的方向信息和空间信息，其满足如下关系式：It should be pointed out that,

and

respectively represent the direction information and spatial information of the video image sequence, which satisfy the following relationship:

。（4）

. (4)

Optionally, before calculating the sub-pixel coordinates of the centroid positions of consecutive image frames in the target image sequence, the above method further includes:

The target image sequence is amplified based on different frequency baseband phase information and a preset magnification factor;

Calculate the sub-pixel coordinates of the centroid position of the continuous image frames in the target image sequence, and determine the real displacement time-history response of the bridge to be monitored according to the sub-pixel coordinates of the centroid position, including:

Divide each image frame in the enlarged target image sequence into grids of multiple cropping regions, use the Otsu threshold segmentation algorithm in each grid to calculate the centroid of the discretized object, and calculate the centroid position sub-pixel of all consecutive image frames Level coordinates, the centroid coordinates of the subsequent frame images starting from the second frame image are subtracted from the centroid coordinates of the first frame image to obtain the sub-pixel level displacement time-history response of the bridge to be monitored, and the bridge to be monitored is calculated according to the scale factor The time-history response of sub-pixel displacement is transformed into the real displacement time-history response.

In this embodiment, the specific steps of the amplification process are as follows:

，并设置合适的放大因子

作为预设的放大因子，采用预设的放大因子对选取的宽带频率基带进行放大处理，实现感兴趣频率基带微小振动幅值的放大处理。将进行放大处理后的宽带频率基带加回高通残差部分和低通残差部分图像序列中， 利用复向可操控金字塔对放大后的图像序列进行重建输出放大后的视频。其中，

为所选宽带频率基带的低频截止频率；

为所选宽带频率基带的高频截止频率。Select the broadband frequency baseband of the region of interest (ROI) in the baseband phase information of different frequencies in the middle

, and set an appropriate magnification factor

As a preset amplification factor, the selected broadband frequency baseband is amplified by using the preset amplification factor, so as to realize the amplification processing of the tiny vibration amplitude of the frequency baseband of interest. Add the amplified broadband frequency baseband back to the image sequence of the high-pass residual part and the low-pass residual part, and use the reversible controllable pyramid to reconstruct the amplified image sequence and output the amplified video. in,

is the low-frequency cut-off frequency of the selected broadband frequency baseband;

is the high-frequency cutoff frequency of the selected wideband frequency baseband.

，

为

在时域内发生的微小振动位移，通过预设的放大因子

，希望得到放大后桥梁竖向振动空域信号为

。对于宽带相位运动放大，首先将桥梁振动的空域信号

表示为一系列复正弦信号的叠加，如式（5）所示。Among them, the principle of the broadband phase motion amplification algorithm is introduced as follows, assuming that the airspace signal of the bridge vibration is

,

for

Small vibration displacements that occur in the time domain, by a preset amplification factor

, it is hoped that the amplified vertical vibration airspace signal of the bridge is

. For broadband phase motion amplification, the spatial domain signal of the bridge vibration is first

Expressed as the superposition of a series of complex sinusoidal signals, as shown in formula (5).

； （5）

; (5)

为某个子正弦信号的圆频率；

为某个子正弦信号的幅值；

为空间位置变量。In the formula,

is the circular frequency of a certain sub-sine signal;

is the amplitude of a sub-sine signal;

is the spatial position variable.

的某个复正弦信号有：For a circular frequency of

A complex sinusoidal signal of is:

； （6）

;(6)

表示复正弦信号，其相位为

，对该信号进行时域滤波，滤除

即可得到：

Represents a complex sinusoidal signal whose phase is

, time-domain filtering is performed on the signal, and the

You can get:

； （7）

;(7)

表示相位差。

Indicates the phase difference.

放大

倍后加回

，得到式（8）的结果Using broadband phase motion amplification algorithm to

enlarge

add back after doubling

, get the result of formula (8)

； （8）

; (8)

表示放大后的复正弦信号，将放大信号

加回原信号，即得到了宽带频率基带

下的宽带相位运动放大结果。从而得到放大后的目标图像序列。

Represents the amplified complex sinusoidal signal, the amplified signal

Adding back the original signal, the broadband frequency baseband is obtained

The broadband phase motion amplification results below. Thus, the enlarged target image sequence is obtained.

Further, using the discrete centroid search algorithm to extract the dynamic displacement time-history response of the micro-vibration of the bridge structure from the enlarged target image sequence, the steps are as follows:

的质心。计算所有连续图像帧的质心位置亚像素级坐标，将从第2帧图像开始的后续帧图像的质心坐标与第1帧图像的质心坐标相减，即可得到桥梁振动的亚像素级位移时程响应，连续两帧图像的质心位置如图3所示。Divide the zoomed-in bridge vibration image frame into multiple grids of cropped regions, and in each grid discretize objects using the Otsu threshold segmentation algorithm

centroid. Calculate the sub-pixel coordinates of the centroid position of all consecutive image frames, and subtract the centroid coordinates of the subsequent frame images starting from the second frame image from the centroid coordinates of the first frame image to obtain the sub-pixel displacement time history of bridge vibration In response, the centroid positions of two consecutive frames of images are shown in Figure 3.

It is worth pointing out that compared with Digital Image Correlation (DIC), optical flow method, edge detection algorithm and target tracking algorithm based on deep learning, the discrete centroid search algorithm in this application has less interference from image background noise, With the advantages of strong real-time performance and high displacement precision, in this embodiment, a more accurate displacement time-history response can be obtained through the discrete centroid search algorithm.

Furthermore, after obtaining the sub-pixel level dynamic displacement time-history response of the micro-vibration of the bridge, the sub-pixel level dynamic displacement time-history can be converted into a physical displacement time-history response according to the scale factor.

Specifically, the scale factor is calculated in different ways in different cases. When the optical axis of the camera is perpendicular to the bridge structure plane, that is, the optical axis is collinear with the normal of the structure plane, the s scale factor is shown in formula (9) or formula (10) .

； (9)

; (9)

or:

； (10)

; (10)

为像素尺寸。In the formula, D is the size of the selected object in the structure plane; d is the corresponding pixel number in the image plane; f is the focal length of the lens; Z is the distance from the camera to the structure plane;

is the pixel size.

时，尺度因子s如式（11）所示。When the optical axis of the camera is not perpendicular to the plane of the bridge structure, that is, there is an angle between the optical axis and the normal of the structural plane

When , the scale factor s is shown in formula (11).

； (11)

; (11)

倍， 利用离散质心搜索算法提取的物理位移时程响应并非桥梁微小振动的真实位移时程响应。假定

为桥梁微小振动的真实位移，

为桥梁微小振动的位移幅值，

为由视频光照变化噪声引起的位移识别误差。则未进行放大处理时结构的位移为

，对桥梁的微小振动进行放大处理后的位移为

。对放大后的位移

进行运动归一化处理即可得到桥梁微小振动的真实动态位移时程响应，桥梁微小振动的真实动态位移时程响应获取原理如图4所示。从式（12）可以看出基于宽带相位运动放大处理可以有效地减小视频中噪声对桥梁微小振动位移识别的影响。When using the broadband phase motion amplification algorithm to amplify the micro-vibration of the bridge structure, the amplitude of the micro-vibration displacement of the bridge is amplified

times, the physical displacement time-history response extracted by the discrete centroid search algorithm is not the real displacement time-history response of the micro-vibration of the bridge. assumed

is the real displacement of the small vibration of the bridge,

is the displacement amplitude of the small vibration of the bridge,

is the displacement recognition error caused by the video illumination change noise. Then the displacement of the structure without zooming in is

, the displacement after amplifying the tiny vibration of the bridge is

. For the displacement after magnification

The real dynamic displacement time-history response of bridge micro-vibration can be obtained by performing motion normalization processing. The principle of obtaining the real dynamic displacement time-history response of bridge micro-vibration is shown in Figure 4. It can be seen from formula (12) that the amplification process based on broadband phase motion can effectively reduce the influence of noise in the video on the recognition of bridge micro-vibration displacement.

； （12）

;(12)

构成Hankel矩阵

，如式（13）所示。Further, the bridge modal parameters (natural frequency, mode shape and damping ratio) are extracted from the real dynamic displacement time-history response of the bridge by using the Hankel dynamic mode decomposition method. Specifically, the Hankel dynamic mode decomposition method can be used to obtain the Multipoint Dynamic Displacement Time History Response

Constitute the Hankel matrix

, as shown in formula (13).

； （13）

;(13)

表示第k个点在

时刻的动位移。In the formula, m , n , p and M are all positive integers,

Indicates that the kth point is in

Momentary displacement.

The specific steps of extracting the modal parameters of the bridge by using the Hankel dynamic mode decomposition method are as follows:

，其计算公式满足如下关系式：(1) Calculate the scaling factor

, and its calculation formula satisfies the following relationship:

； （14）

;(14)

是矩阵

的最后一列，

是

的最后一列的第一个子块。In the formula, where

is the matrix

the last column of the ,

Yes

The first subblock of the last column of .

(2) The Hankel matrix consists of formula (15).

；（15）

;(15)

为同一个Hankel矩阵在时间上向前的移动；

和

表示两个相邻时间数据矩阵。In the formula,

For the same Hankel matrix moving forward in time;

and

Represents two matrices of adjacent time data.

(3) The truncated singular value decomposition (SVD) calculation of X is shown in formula (16).

； （16）

;(16)

和

为奇异值分解得到的两种酉矩阵，

为

的伴随矩阵；

为对角矩阵,对角线元素为

个奇异值，

为正整数。In the formula,

and

Two kinds of unitary matrices obtained by singular value decomposition,

for

The adjoint matrix;

is a diagonal matrix, and the diagonal elements are

a singular value,

is a positive integer.

如式（17）所示。(4) Operator matrix

As shown in formula (17).

； （17）

;(17)

为

的伴随矩阵；

为

的共轭矩阵；

为

的逆矩阵。In the formula,

for

The adjoint matrix;

for

the conjugate matrix;

for

the inverse matrix of .

的特征值和特征向量为

和

，且

。(5)

The eigenvalues and eigenvectors of are

and

,and

.

(6) The natural frequency and damping ratio of the bridge are calculated by formula (18).

；（18）

;(18)

表示桥梁动态位移时程响应的采样时间，

表示桥梁的各阶固有频率；

表示桥梁的各阶阻尼比；

表示模态阶数，为正整数。In the formula,

represents the sampling time of the bridge dynamic displacement time-history response,

Indicates the natural frequency of each order of the bridge;

Indicates the damping ratio of each order of the bridge;

Indicates the modal order, which is a positive integer.

(7) Then, the mode shape of the bridge can be identified by formula (19).

； （19）

;(19)

Optionally, the bridge damage location and bridge damage degree are identified according to the mode shape calculation model, including:

Normalize the modal shapes of each order of the bridge, and superimpose their absolute values, and calculate the average modal energy when the bridge is not damaged and when it is damaged, respectively. According to the average modal energy Calculate the average energy difference of each point on the bridge, and identify the bridge damage location and bridge damage degree according to the average energy.

Optionally, identifying the bridge damage location and bridge damage degree according to the average energy includes:

Generate a fitted curve based on the average energy;

The position of the sudden change of the fitting curve is determined as the damage position of the bridge, and the damage degree of the bridge structure is determined by judging the amplitude of the singular point of the fitting curve.

In this optional implementation, the modal shape superposition energy algorithm is used to identify the damage location and damage degree of the bridge. First, the modal shape of each order of the bridge is normalized, and the absolute Values are superimposed, as shown in formula (20).

； （20）

;(20)

为桥梁无损伤状态下前N阶模态振型的叠加，为避免因奇异点幅值的不同，造成归一化混乱，对桥梁损伤状态下各阶模态振型进行振型归一化时，使用桥梁无损伤状态下的同阶模态振型的最大值，如式（21）所示。In the formula,

It is the superposition of the first N- order mode shapes in the bridge without damage. In order to avoid the normalization confusion due to the difference in the amplitude of the singular point, when the mode shapes of each order in the bridge damage state are normalized , using the maximum value of the mode shape of the same order under the bridge without damage, as shown in equation (21).

； （21）

; (twenty one)

为桥梁损伤状态下的前N阶模态振型叠加；

为桥梁损伤状态下的各阶模态振型。In the formula,

is the mode shape superposition of the first N orders in the damaged state of the bridge;

are the mode shapes of each order in the damaged state of the bridge.

Secondly, the average modal energy of the bridge without damage and with damage is calculated respectively, as shown in equations (22)~(23).

； （22）

; (twenty two)

；（23）

;(twenty three)

为桥梁无损伤时第n点的平均模态振型能量值，

为桥梁有损伤时第n点的平均模态振型能量值。In the formula,

is the average mode shape energy value of the nth point when the bridge is undamaged,

is the average mode shape energy value of the nth point when the bridge is damaged.

Finally, the average energy difference of each point on the bridge is calculated, as shown in equation (24).

； （24）

; (twenty four)

的拟合曲线，则拟合曲线的突变的位置即为桥梁的损伤位置。通过判断奇异点幅值大小确定桥梁结构的损伤程度。这样，通过拟合曲线可以快速地识别桥梁的损伤位置和损伤程度。By observing

The fitting curve, then the position of the sudden change of the fitting curve is the damage position of the bridge. The damage degree of the bridge structure is determined by judging the magnitude of the singular point. In this way, the damage location and damage degree of the bridge can be quickly identified by fitting the curve.

Optionally, after the bridge damage location and bridge damage degree are identified according to the modal shape calculation model, the above method further includes:

Carry out a safety assessment of the bridge to be monitored according to the bridge damage location and bridge damage degree, and generate an alarm message when the damage location of the bridge is determined, or when the bridge damage degree exceeds the damage threshold.

In this optional implementation manner, the bridge damage location and damage degree acquired by the above image processing technology are combined with the historical data of the bridge health status to evaluate the operation safety of the bridge in real time. Based on the broadband phase motion amplification algorithm and the discrete centroid search algorithm, the micro-vibration of the bridge can be monitored in real time. Through the real-time monitoring data, the operation safety, degradation behavior and remaining life of all bridges in the entire area can be evaluated, which can contribute to the health of the bridge. Monitoring, maintenance and daily maintenance provide data support.

The embodiment of the present application also provides a bridge monitoring system based on image recognition, including a memory, a processor, and a computer program stored on the memory and operable on the processor, and the above method is implemented when the processor executes the computer program A step of. The bridge monitoring system based on image recognition can implement various embodiments of the above-mentioned method, and can achieve the same beneficial effect, and will not be repeated here.

The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the above method steps are implemented. The readable storage medium can implement various embodiments of the above-mentioned method, and can achieve the same beneficial effect, which will not be repeated here.

The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make many modifications and changes according to the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning or limited experiments on the basis of the prior art shall be within the scope of protection defined by the claims.

Claims (8)

1. A bridge health monitoring method based on image recognition is characterized by comprising the following steps:

acquiring a first image of a bridge to be monitored, and preprocessing the first image to obtain a second image;

performing spatial domain decomposition on the second image to obtain a corresponding target image sequence;

calculating sub-pixel coordinates of the centroid positions of continuous image frames in the target image sequence, and determining real displacement time course response of the bridge to be monitored according to the sub-pixel coordinates of the centroid positions;

extracting bridge modal parameters from the real displacement time-course response by using a Hankel dynamic modal decomposition method, wherein the bridge modal parameters comprise natural frequency, vibration mode and damping ratio;

and establishing a modal shape calculation model according to the bridge modal parameters, and identifying the damage position and the damage degree of the bridge according to the modal shape calculation model.

2. The bridge health monitoring method based on image recognition of claim 1, wherein the preprocessing the first image to obtain a second image comprises:

and rotating, cutting and scaling the first image to obtain a second image.

3. The bridge health monitoring method based on image recognition according to claim 1, wherein the performing spatial decomposition on the second image to obtain a corresponding target image sequence comprises:

performing spatial domain decomposition on the second image to obtain an initial target image sequence, wherein the initial target image sequence comprises a first residual error part, different-frequency baseband phase information and a second residual error part;

and extracting the texture features of the initial target image sequence by using a two-dimensional Gabor wavelet filter, and performing translation processing on the image information except the texture features in the target image sequence to finish the noise reduction processing on the initial target image sequence to obtain a final target image sequence.

4. The image recognition-based bridge health monitoring method of claim 3, wherein prior to the calculating the centroid position sub-pixel level coordinates of successive image frames in the sequence of target images, the method further comprises:

amplifying the target image sequence based on the different-frequency baseband phase information and a preset amplification factor;

the calculating the sub-pixel-level coordinates of the centroid positions of the continuous image frames in the target image sequence, and determining the real displacement time-course response of the bridge to be monitored according to the sub-pixel-level coordinates of the centroid positions comprises the following steps:

dividing each image frame in the amplified target image sequence into a plurality of grids of cutting areas, calculating the centroid of a discretized object in each grid by using an Otsu threshold segmentation algorithm, calculating the subpixel level coordinates of the centroid positions of all continuous image frames, subtracting the centroid coordinates of a subsequent image frame starting from a 2 nd image frame from the centroid coordinates of a 1 st image frame to obtain the subpixel level displacement time course response of the bridge to be monitored, and converting the subpixel level displacement time course response of the bridge to be monitored into a real displacement time course response according to a scale factor.

5. The bridge health monitoring method based on image recognition according to claim 1, wherein the recognizing a bridge damage position and a bridge damage degree according to the modal shape calculation model comprises:

and (3) performing mode normalization processing on each order of modal shape of the bridge, superposing absolute values of the modal shape, calculating average modal shape energy of the bridge when the bridge is not damaged and damaged respectively, calculating average energy difference of each point on the bridge according to the average modal shape energy, and identifying the damage position and the damage degree of the bridge according to the average energy.

6. The bridge health monitoring method based on image recognition of claim 5, wherein the identifying the bridge damage location and the bridge damage level according to the average energy comprises:

generating a fitting curve according to the average energy;

and determining the mutation position of the fitting curve as the damage position of the bridge, and determining the damage degree of the bridge structure by judging the singular point amplitude of the fitting curve.

7. The bridge health monitoring method based on image recognition according to claim 1, wherein after the bridge damage position and the bridge damage degree are recognized according to the modal shape calculation model, the method further comprises:

and performing safety assessment on the bridge to be monitored according to the bridge damage position and the bridge damage degree, and generating alarm information when the damage position of the bridge is determined or the bridge damage degree exceeds a damage threshold value.

8. A bridge health monitoring system based on image recognition, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the computer program implements the steps of the method according to any one of the preceding claims 1 to 7.

Images