A target defect recognition method and system based on underwater inspection system

Abstract

This invention provides a method and system for identifying target defects based on an underwater inspection system. Image information and location are collected; the collected data is transmitted via a base station. A data processing center uses an intelligent image recognition algorithm to pre-process the image, combining data from a historical image database and a target information database to identify defects and output the defect severity and target defect location. By integrating multiple sensors and intelligent algorithms, this method enables efficient and accurate detection and identification of defects in key underwater components, such as bridge piers and pile foundations, as well as ship bottoms and propellers.

Description

Technical Field

[0001] The present invention relates to the field of underwater detection technology, and in particular to a method and system for identifying target defects of underwater parts of bridges and ships based on an underwater inspection system. Background Art

[0002] Bridges and ships are critical transportation infrastructure, and the health of their underwater components is directly related to the safety and durability of their structures. With the advancement of underwater inspection technologies and equipment, the data sources for underwater defect detection have become more diverse. However, due to the complex and ever-changing underwater environment, traditional inspection methods often suffer from low efficiency and poor accuracy. Therefore, developing a method that can efficiently and accurately identify defects in the underwater components of bridges and ships is crucial. Summary of the Invention

[0003] The present invention aims to address the aforementioned challenges in the existing technology by providing a method and system for identifying target defects based on an underwater inspection system. By integrating multiple sensors and intelligent algorithms, this method enables efficient and accurate detection and identification of defects in key underwater components, such as bridge piers and pile foundations, as well as ship bottoms and propellers.

[0004] In order to achieve the above object, the present invention provides a target defect identification method and system based on an underwater inspection system, the method comprising the following steps:

[0005] Step S1: collecting image information and location;

[0006] Step S2: The collected data is transmitted via the base station;

[0007] Step S3: The data processing center uses an image intelligent recognition algorithm to pre-process the image, combine the historical image database and the target information database data, identify the defect, and output the defect severity and target defect location.

[0008] Preferably, the underwater inspection robot uses a high-definition camera to collect image information IM(i) of the target to be inspected, and uses a navigation and positioning system to record the inspection and its position P(i), shooting angle S(i) when each image is taken, where i represents the serial number of the captured image.

[0009] Preferably, the collected data is transmitted to a data processing center via a base station using a communication system.

[0010] Preferably, step S3.1: using the image preprocessing network NET_PRE to perform denoising, enhancement, and calibration;

[0011] The image preprocessing network inputs the image IM(i), shooting position P(i), and shooting angle S(i) collected by the underwater inspection robot's high-definition camera, preprocesses the collected image IM(i), realizes denoising, enhancement, calibration, etc. of the sensor-collected image, and stores the preprocessed image in the historical image database IMS for comparative analysis in image recognition.

[0012] Preferably, step S3.2: using the image recognition network NET_REC to identify the target defect type TY(i), severity DG(i), and defect location PO(i);

[0013] The input of the image recognition network NET_REC includes the preprocessed target image, position, angle, and historical images of the current area;

[0014] After processing, the image recognition network NET_REC outputs the target image defect type TY(i), severity DG(i), and defect location PO(i);

[0015] The defect type TY(i) and the severity DG(i) are directly output by the network, and the defect position PO(i) is calculated by outputting the pixel position of the defect relative to the image, superimposing the inspection robot's shooting position P(i), and the shooting angle S(i) information.

[0016] Preferably, the robot position P ( i ) = [ x i ( P ) , y i ( P ) , z i ( P ) ] , the relative pixel position of the defect position PO(i) obtained by image recognition is [ x i ( pix ) , y i ( pix ) ] , the camera field of view is , resolution is Pix_X Pix_Y, the distance between the camera and the target is L, the shooting angle S ( i ) = [ θ i , φ i ] ,in is the azimuth, is the pitch angle,

[0017] The center point of the image is:

[0018] ;

[0019] Since the diameter of the camera's field of view is equal to the diagonal length of the captured image, , the field of view angle corresponding to each pixel is , pix is ​​the field of view angle corresponding to one pixel;

[0020] Calculate the field angle offset of the defect position relative to the image center,

[0021] ;

[0022] The spatial coordinates of the defect position in the image are identified as:

[0023] .

[0024] Preferably, step S3.3: using the maintenance decision network NET_DEC to output maintenance suggestions; using the target image defect type TY(i), severity DG(i), and defect position PO(i) data output by the defect recognition network, combined with the target type and maintenance history data, to obtain maintenance suggestions.

[0025] Preferably, ultrasonic data, magnetic sensor data or sonar data may be added to the target defect identification method for identification.

[0026] Another aspect of the present invention provides a target defect recognition system based on an underwater inspection system, characterized in that the system is used to execute the above method.

[0027] Preferably, the underwater inspection system is designed for underwater partial defect detection and identification, and is mainly composed of an inspection device, a base station, a data processing center, and corresponding communication and energy supply systems.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] 1) Efficiency: Intelligent recognition algorithms significantly improve underwater defect detection speed. Furthermore, intelligent algorithms can be trained offline, decoupled from the system, and leverage open source data training to continuously iterate and optimize the algorithm.

[0030] 2) Accuracy: The use of multiple sensors and intelligent algorithms has significantly improved the accuracy of defect identification. While providing the defect type and severity, it can also output the spatial location of the defect, providing location information for subsequent maintenance.

[0031] 3) Scalability: The system can be expanded and upgraded according to actual needs to adapt to the detection requirements in different scenarios. The overall defect detection accuracy of the system can be improved by simply upgrading the image preprocessing network NET_PRE, the image recognition network NET_REC, and the maintenance decision network NET_DEC. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Schematic diagram of a target defect recognition method based on image data in a target defect recognition method of an underwater inspection system of the present invention;

[0033] Figure 2 A schematic diagram of the robot position and shooting in the target defect recognition method of the underwater inspection system of the present invention;

[0034] Figure 3 Schematic diagram of a target defect recognition method based on ultrasonic data in a target defect recognition method of an underwater inspection system of the present invention;

[0035] Figure 4 This is a schematic diagram of a target defect recognition method based on magnetic sensor data in a target defect recognition method of an underwater inspection system of the present invention. DETAILED DESCRIPTION

[0036] The present invention is further described in detail below with reference to the accompanying drawings:

[0037] In order to better understand the present invention, the embodiments of the present invention are explained in detail below with reference to the accompanying drawings.

[0038] A first embodiment of the present invention provides a system for underwater inspection.

[0039] The underwater inspection system is designed specifically for detecting and identifying defects in underwater components such as bridges and ships. It integrates a variety of advanced sensors and intelligent algorithms to achieve efficient and accurate inspections of key underwater areas. The system primarily consists of an inspection device (underwater robot), a base station, a data processing center, and corresponding communication and energy supply systems.

[0040] The underwater inspection system includes an inspection device (underwater robot).

[0041] The inspection device is the core of the entire system, responsible for autonomous navigation, positioning, data collection, and preliminary defect identification in underwater environments. Its main components include:

[0042] Carrier structure: Made of high-strength, corrosion-resistant materials to ensure stability and durability in underwater environments.

[0043] Navigation and positioning system: Integrates multiple technologies such as GPS, sonar, and inertial navigation to achieve precise underwater positioning and autonomous navigation.

[0044] Sensor array: including high-definition cameras, ultrasonic sensors, magnetic sensors, water quality sensors, sonar sensors, etc., used to collect images of underwater environments, target thickness, magnetic fields, water quality data, sonar images, etc.

[0045] Data processing unit: built-in high-performance processor to perform preliminary processing and analysis on the collected data and extract key information.

[0046] Energy supply system: Use high-efficiency battery packs or underwater wireless charging technology to ensure that the inspection device can work underwater for a long time.

[0047] The underwater inspection robot includes a base station.

[0048] The base station is set up at a fixed location on or near the water surface and is responsible for receiving data sent by the inspection device and performing preliminary processing and analysis. Its main functions include:

[0049] Data reception and forwarding: Receive the original data sent by the inspection device, perform preliminary processing and then forward it to the data processing center.

[0050] Status monitoring: Real-time monitoring of the working status of the inspection device, including power, position, speed, etc., to ensure the smooth progress of the inspection task.

[0051] Communication relay: Provide communication relay services between the inspection device and the data processing center to ensure the stability and reliability of data transmission.

[0052] The underwater inspection robot includes a data processing center.

[0053] The data processing center is the "brain" of the entire system, responsible for receiving data transmitted by the base station and using advanced image processing and machine learning algorithms to identify defects. Its main functions include:

[0054] Data preprocessing: Perform preprocessing operations such as denoising, enhancement, and calibration on the received raw data to improve the accuracy of subsequent analysis.

[0055] Defect recognition algorithms: Develop deep learning-based image recognition algorithms, ultrasonic data analysis algorithms, and other algorithms to identify defects in bridge piers, pile foundations, ship bottoms, propellers, and other parts.

[0056] Result output and feedback: The recognition results are output in the form of images, reports, etc., including information such as the location, type, severity, etc. of the defect, and targeted maintenance suggestions are provided.

[0057] Communications and energy supply systems:

[0058] The communication and energy supply system is an important part of the entire system, ensuring data transmission and energy supply between the inspection device and the base station, and between the base station and the data processing center. Its main functions include:

[0059] Wireless communication: Underwater wireless communication technology is used to realize data transmission between the inspection device and the base station.

[0060] Wired communication: When necessary, wired communication methods such as underwater optical cables are used to ensure the stability and reliability of data transmission.

[0061] Energy supply: Provide continuous and stable energy supply for inspection devices and base stations to ensure long-term operation of the system.

[0062] Another aspect of the present invention provides a target defect recognition method based on an underwater inspection system.

[0063] In one embodiment of the present invention, a method for target defect recognition based on image data is used to recognize target defects in an underwater inspection system.

[0064] Image recognition algorithm based on deep learning is used to identify cracks, corrosion, cavities and other defects in bridge piers and pile foundations. Figure 1 The target defect recognition method based on image data of the present invention specifically includes the following steps:

[0065] Step S1: The underwater inspection robot uses a high-definition camera to collect image information IM(i) of the target to be inspected, and uses the navigation and positioning system to record the inspection and its position P(i) and shooting angle S(i) when each image is taken, where i represents the serial number of the captured image.

[0066] Step S2: Utilize the communication system to transmit the collected data to the data processing center via the base station.

[0067] Step S3: The data processing center uses the image intelligent recognition algorithm to identify cracks, corrosion, voids and other defects in bridge piers and pile foundations after pre-processing the image, with the support of the historical image database IMS and the target information database TARS, and can output the defects, severity and target defect locations.

[0068] Step S3.1: Use the image preprocessing network NET_PRE for denoising, enhancement, and calibration.

[0069] The network takes as input the image IM(i) captured by the underwater inspection robot's high-definition camera, along with the camera's position P(i) and angle S(i). It then preprocesses the image IM(i), performing denoising, enhancement, and calibration on the sensor-captured image. The preprocessed image is then stored in the historical image database IMS for comparative analysis during image recognition. The image preprocessing network can be trained offline, utilizing existing open-source data, fully leveraging the support provided by general-purpose data.

[0070] Step S3.2: Use the image recognition network NET_REC to identify the target defect type TY(i), severity DG(i), and defect location PO(i).

[0071] The input to the image recognition network NET_REC includes the preprocessed target image, its location, and its angle, as well as historical images of the current area. After processing, the image recognition network NET_REC outputs the target image's defect type TY(i), severity DG(i), and defect location PO(i). The defect type TY(i) and severity DG(i) are directly output by the network, while the defect location PO(i) is calculated by combining the network's output pixel position relative to the image with the inspection robot's camera position P(i) and camera angle S(i).

[0072] Set robot position P ( i ) = [ x i ( P ) , y i ( P ) , z i ( P ) ] , the relative pixel position of the defect position PO(i) obtained by image recognition is [ x i ( pix ) , y i ( pix ) ] , the camera field of view is , resolution is Pix_X Pix_Y, the distance between the camera and the target is L. Shooting angle S ( i ) = [ θ i , φ i ] ,in is the azimuth, is the pitch angle.

[0073] Specific reference Figure 2 .

[0074] The center point of the image is:

[0075]

[0076] Since the diameter of the camera's field of view is equal to the diagonal length of the captured image, , the field of view angle corresponding to each pixel is

[0077]

[0078] Where pix is ​​the field of view angle corresponding to one pixel.

[0079] Then calculate the field angle offset of the defect position relative to the image center,

[0080]

[0081] The spatial coordinates of the defect position in the image are identified as:

[0082]

[0083] Step S3.3: Output maintenance recommendations using the maintenance decision network NET_DEC

[0084] The target image defect type TY(i), severity DG(i), and defect location PO(i) data output by the defect recognition network are combined with the target type and maintenance history data to obtain maintenance recommendations such as the repair method and time requirements for the current defect.

[0085] In another embodiment of the present invention, target defect recognition of ultrasonic data can be added. Figure 3 .

[0086] Ultrasonic sensor data, combined with intelligent recognition algorithms, can be used to identify defects such as wear and tear on the ship's bottom and propellers. Target defect recognition based on ultrasonic data is handled similarly to image recognition defects and will not be analyzed in detail.

[0087] In another embodiment of the present invention, target defect recognition based on magnetic sensor data can be added. Figure 4 .

[0088] Magnetic sensors are used to detect abnormal changes in metal structures and assist in identifying potential structural damage. Target defect identification based on magnetic sensor data is handled similarly to image-based defect identification and will not be discussed in detail.

[0089] In the description of the present invention, it should be noted that, unless otherwise specified or limited, the terms "connected" and "connection" should be understood in a broad sense. For example, they can refer to fixed connection, detachable connection, or integral connection; mechanical connection, electrical connection; direct connection, or indirect connection through an intermediary. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0090] In the description of the present invention, unless otherwise specified, the terms "upper", "lower", "left", "right", "inside", "outside", etc., indicating directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, they cannot be understood as limiting the present invention.

[0091] Finally, it should be noted that the above technical solution is only one embodiment of the present invention. For those skilled in the art, it is easy to make various types of improvements or modifications based on the application methods and principles disclosed in the present invention, and it is not limited to the method described in the above specific embodiment of the present invention. Therefore, the method described above is only preferred and does not have a restrictive meaning.

Claims

1. A target defect identification method based on an underwater inspection system, characterized by: Step S1: Collect underwater image information and location; Step S2: The collected data is transmitted via the base station; Step S3: The data processing center uses an image intelligent recognition algorithm to pre-process the image, combine the historical image database and the target information database data, identify the defect, and output the defect severity and target defect location; The step S3 specifically includes: Step S3.1: Use the image preprocessing network NET_PRE for denoising, enhancement, and calibration; The image preprocessing network inputs the image IM(i), shooting position P(i), and shooting angle S(i) collected by the underwater inspection robot's high-definition camera, preprocesses the collected image IM(i), realizes denoising, enhancement, and calibration of the sensor-collected image, and stores the preprocessed image in the historical image database IMS for comparative analysis in image recognition; Step S3.2: Use the image recognition network NET_REC to identify the target defect type TY(i), severity DG(i), and defect location PO(i); The input of the image recognition network NET_REC includes the preprocessed target image, position, angle, and historical images of the current area; After processing, the image recognition network NET_REC outputs the target image defect type TY(i), severity DG(i), and defect location PO(i); The defect type TY(i) and the severity DG(i) are directly output by the network, and the defect position PO(i) is calculated by superimposing the pixel position of the defect relative to the image output by the network, the shooting position P(i) of the inspection robot, and the shooting angle S(i); The step S3.2 further includes: setting the robot position The relative pixel position of the defect position PO(i) obtained by image recognition is The camera's field of view is α, the resolution is Pix_X×Pix_Y, the distance between the camera and the target is L, and the shooting angle is where θ i is the azimuth, is the pitch angle, The center point of the image is: Since the diameter of the camera's field of view is equal to the diagonal length of the captured image, The field of view angle corresponding to each pixel is Pix is ​​the field of view angle corresponding to one pixel; Calculate the field angle offset of the defect position relative to the image center, The spatial coordinates of the defect position in the image are identified as: Step S3.3: Use the maintenance decision network NET_DEC to output maintenance recommendations; use the target image defect type TY(i), severity DG(i), and defect location PO(i) data output by the defect recognition network, combined with the target type and maintenance history data, to obtain maintenance recommendations.

2. The method according to claim 1, wherein The step S1 specifically includes: The underwater inspection robot uses a high-definition camera to collect the image information IM(i) of the target to be inspected, and uses the navigation and positioning system to record the inspection and its position g(i) and shooting angle S(i) when each image is taken, where i represents the sequence number of the captured image.

3. The method according to claim 2, wherein The step S2 specifically includes: The communication system is used to transmit the collected data to the data processing center through the base station.

4. The method according to claim 2, wherein: Ultrasonic wave data, magnetic sensor data, and sonar sensor data can be added to the target defect recognition method for recognition.

5. A target defect recognition system based on an underwater inspection system, characterized by: The system is used to execute the method according to any one of claims 1 to 4.

6. The system according to claim 5, characterized in that: The underwater inspection system is designed for underwater defect detection and identification, and is mainly composed of inspection devices, base stations, data processing centers, and corresponding communication and energy supply systems.

Images