Container bottom lock detection method based on computer vision

Abstract

The invention discloses a container bottom lock detection method based on computer vision, and relates to the technical field of container body lock detection, and the method comprises the following steps: 1, starting a transfer platform lock detection system; 2, a camera is used for capturing and recognizing the lock head, whether the lock head exists or not is detected, and if yes, a server controls the lock head to send out a dry contact signal; if the existence of the lock head is not detected, the server judges whether a dry contact signal needs to be sent out or not, and then the path of signal is closed; 3, the PLC industrial personal computer judges whether a dry contact signal is received or not, and if the dry contact signal is received, the step 4 is executed; the method follows a doubtful subordinate principle, focuses on no missing detection on algorithm and software business levels, combines two modes of computer vision detection and manual inspection detection, solves extreme situations which cannot be identified by some algorithms by utilizing a manual inspection mode, further reduces missing report rate and false alarm rate, and improves the detection accuracy. And therefore, the safety in actual business work can be ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of container bottom lock detection, and in particular to a container bottom lock detection method based on computer vision. Background Art

[0002] The container bottom lock is a safety device used to fix the connection between the container and the transport vehicle. It prevents the container from sliding or falling off during transportation by locking the corner fittings at the bottom of the container, ensuring transportation safety. There are many designs and operating methods for container locks, including fixed type, lifting type, etc., to adapt to different types of containers and transport vehicles. When selecting and using a container lock, you should ensure that the lock matches the specifications of the container and install and lock it according to the correct operating procedures to achieve the best safety effect.

[0003] After entering the dock, the ship will load and unload containers on the seaside shore crane transfer platform (hereinafter referred to as the platform). After the container is placed on the platform, the on-site dock workers will remove the locks at the four corners of the container, and then ensure that all the locks are removed before proceeding to the next step. If the next step is carried out without removing the locks due to negligence, a production safety incident may occur. The prior art publication number CN116343125A is a computer vision-based container bottom lock detection method. Although computer vision technology is applied to the detection process of container bottom locks, there are still cases of missed detection and false alarms because the system cannot distinguish whether there is a lock or the judgment is wrong, which affects the overall detection accuracy of the system. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a method for detecting the lock on the bottom of a container based on computer vision. It follows the principle of avoiding missed detections at the algorithm and software business levels, and combines computer vision detection with manual inspection. Manual inspection is used to solve some extreme situations that cannot be identified by the algorithm, further reducing the missed alarm rate and false alarm rate to ensure safety in actual business work.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is: a container bottom lock detection method based on computer vision, comprising the following steps:

[0006] Step 1: Start the transfer platform lock detection system;

[0007] Step 2: Use the camera to capture and identify the lock to detect whether the lock exists. If the lock exists, the server controls to send a dry contact signal;

[0008] If the lock is not detected, the server determines whether a dry contact signal needs to be sent and then turns off the signal.

[0009] Step 3: The PLC industrial computer determines whether a dry contact signal is received. If a dry contact signal is received, the process proceeds to step 4.

[0010] If no dry contact signal is received, proceed to step five;

[0011] Step 4: The PLC industrial computer controls the signal light to remain on, and then manually determines whether the lock is cleaned. If the lock is not cleaned, return to step 2;

[0012] If the lock is clean, manually tap the abnormal button and then go to step 6;

[0013] Step 5: The PLC industrial computer controls the signal light to remain off;

[0014] Step 6. After the lock detection is completed, proceed to the next step.

[0015] Preferably, the number of the cameras is set to twenty, and the twenty cameras are installed at the lock landing points of all different container lifting schemes to ensure that the lock appears near the vertical center line of the camera screen;

[0016] Twenty cameras are installed below the calibration platform on both sides of the container.

[0017] Preferably, the camera is fixed under the calibration platform by means of a multi-point and multi-end fixing method of an integrated cast bracket. After the fixing screws of the camera are locked, they are welded and fixed between the screws and the fixing parts, and cable ties are used to assist in the position of the camera as a whole and the fixing parts to prevent the camera from moving after multiple shocks.

[0018] Preferably, the server is deployed in a computer room, and is used to run an image recognition algorithm and business software to perform digital image processing on continuous images captured by the camera, and utilize the image recognition algorithm to detect whether there is any residual lock at the bottom of the container when the shore crane is operating in real time.

[0019] Preferably, a hard disk recorder is installed in the machine room for storing the images of the front-end camera.

[0020] Preferably, the PLC industrial computer is installed below the calibration platform to integrate all information from the acquisition equipment and feed back the corresponding information through other equipment.

[0021] Preferably, the signal light is installed under the calibration platform to receive identification information from the server and provide feedback.

[0022] Preferably, the equipment in the machine room communicates with the equipment below the calibration platform via a switch.

[0023] Compared with the prior art, the present invention provides a container bottom lock detection method based on computer vision, which has the following beneficial effects:

[0024] 1. The overall process of the present invention follows the principle of "if there is doubt, follow the existing one". At the algorithm and software business levels, priority is given to not missing any detections. Computer vision detection is combined with manual inspection. Manual inspection is used to solve some extreme situations that cannot be identified by the algorithm, further reducing the missed alarm rate and false alarm rate to ensure the safety of actual container lifting work.

[0025] 2. The response time of the present invention from camera capturing the picture to signal light indicating switching is within 1s, which is sensitive and fast, helping to shorten the lock detection time and reduce the impact of lock detection on container lifting efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a flow chart of the present invention;

[0027] Figure 2 It is a structural diagram of the transfer platform lock detection system of the present invention. DETAILED DESCRIPTION

[0028] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0029] like Figure 1 As shown, the present invention provides a container bottom lock detection method based on computer vision, comprising the following steps:

[0030] Step 1: Start the transfer platform lock detection system;

[0031] The structure of the transfer platform lock detection system is as follows: Figure 2 As shown, it includes signal lights, PLC industrial computers, cameras, switches, hard disk recorders and servers;

[0032] There are 20 cameras, which are installed at the lock landing points of all different container lifting schemes to ensure that the lock appears near the vertical center line of the camera screen, and the 20 cameras are installed below the calibration platform on both sides of the container;

[0033] Because the calibration platform will vibrate violently when the container is lowered, which may cause the camera screen to vibrate and the installation position to move, in order to fix the position of the camera, the camera in the present invention is fixed under the calibration platform by means of a multi-point multi-end fixing method of an integrated casting bracket. After the fixing screws of the camera are locked, they are welded and fixed between the screws and the fixing parts, and a cable tie is added to assist the position of the camera as a whole and the fixing parts to prevent the camera from moving after multiple vibrations;

[0034] Taking into account that containers are mostly used in port operations, that is to say, cameras are mostly used in water environments such as the seaside, where the air humidity is high, and outdoor equipment is also faced with problems such as salt spray corrosion, dew erosion, and dust adhesion. To solve this problem, the present invention gives priority to IP67 triple-proof devices in camera selection, and at the same time has a shell with certain corrosion resistance, and the customized welding parts and fixing parts also have excellent corrosion resistance and aging resistance.

[0035] The PLC industrial computer and signal light in the present invention are also installed under the calibration platform. The PLC industrial computer is used to integrate all the information of the acquisition equipment and feed back the corresponding information through other devices; the signal light is used to receive the identification information of the server and give feedback;

[0036] The server is deployed in the computer room and is used to run the image recognition algorithm and business software to perform digital image processing on the continuous images collected by the camera. The image recognition algorithm is used to detect whether there is a residual lock at the bottom of the container when the shore crane is operating in real time. In addition, a hard disk recorder is installed in the computer room to store the images of the front-end camera;

[0037] In order to realize the communication between the equipment on the machine room side and the equipment on the platform side, the number of switches in the present invention is set to two, which are installed in the calibration platform and the machine room respectively, and are mainly used to realize the communication between the equipment in the machine room and the equipment under the calibration platform.

[0038] Step 2: Use the camera to capture and identify the lock to collect the image information of the lock. The collected image information is uploaded to the server through the switch. The server runs the image recognition algorithm to perform digital image processing on the image and then identifies whether there is a lock;

[0039] If the server's identification result is that there is a lock, the server controls the sending of a dry contact signal;

[0040] If the server's identification result is that there is no lock, the server determines whether a dry contact signal needs to be sent and then closes the signal.

[0041] Step 3: The PLC industrial computer determines whether a dry contact signal is received. If a dry contact signal is received, the process proceeds to step 4.

[0042] If no dry contact signal is received, proceed to step five.

[0043] Step 4: After the PLC receives the dry contact signal, it controls the signal light to remain on. The process from the image being captured by the camera to the recognition feedback to the signal light takes no more than 1 second. Then the on-site staff manually checks to determine whether the lock on the bottom of the container is clean;

[0044] If the manual inspection result shows that the lock is not clean, the process returns to step 2, where the camera captures the container again and repeats the above process;

[0045] If the manual inspection result shows that the lock is clean, the abnormal button is manually tapped and then the process goes to step six.

[0046] Step 5: When the PLC industrial computer does not receive the dry contact signal, the control signal light remains off.

[0047] Step 6: After checking that the lock on the bottom of the container is clean, you can proceed to the next step.

[0048] The above overall process follows the principle of "if in doubt, follow the existing ones". At the algorithm and software business levels, priority is given to avoiding missed detections, achieving a missed alarm rate of <3‰ and a false alarm rate of <5% (the false alarm rate refers to the situation where the system cannot distinguish whether it is a lock or issues an alarm due to an incorrect judgment, and needs to notify staff to intervene for further judgment; the missed alarm rate refers to the frequency with which the container lock is not detected) to ensure safety in actual business work.

[0049] All cameras continuously capture and identify the padlock. As long as any camera recognizes the padlock, the PLC device controls the signal light to light up, reminding the worker to check all positions. The overall system response time from camera capture to signal light switching is within 1 second, which is sensitive and fast, helping to shorten the padlock detection time. Moreover, to further ensure that no padlock is missed, in some extreme cases, the padlock may be cleaned but the algorithm still recognizes the padlock. In this case, the worker only needs to patrol manually to determine that it has been cleaned, then tap the abnormal button and turn off the signal light.

[0050] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. A container bottom lock detection method based on computer vision, characterized in that: The steps include: Step 1: Start the transfer platform lock detection system; Step 2: Use the camera to capture and identify the lock to detect whether the lock exists. If the lock exists, the server controls to send a dry contact signal; If the lock is not detected, the server determines whether a dry contact signal needs to be sent and then turns off the signal. Step 3: The PLC industrial computer determines whether a dry contact signal is received. If a dry contact signal is received, the process proceeds to step 4. If no dry contact signal is received, proceed to step five; Step 4: The PLC industrial computer controls the signal light to remain on, and then manually determines whether the lock is cleaned. If the lock is not cleaned, return to step 2; If the lock is clean, manually tap the abnormal button and then go to step 6; Step 5: The PLC industrial computer controls the signal light to remain off; Step 6. After the lock detection is completed, proceed to the next step.

2. The method for detecting the bottom lock of a container based on computer vision according to claim 1 is characterized in that: The number of the cameras is set to be twenty, and the twenty cameras are installed at the lock landing points of all different container lifting schemes to ensure that the lock appears near the vertical center line of the camera screen; Twenty cameras are installed below the calibration platform on both sides of the container.

3. The method for detecting the bottom lock of a container based on computer vision according to claim 2 is characterized in that: The camera is fixed under the calibration platform by means of a multi-point and multi-end fixing method of an integrated cast bracket. After the fixing screws of the camera are locked, they are welded and fixed between the screws and the fixing parts, and cable ties are used to assist in the position of the camera as a whole and the fixing parts to prevent the camera from moving after multiple shocks.

4. The method for detecting the bottom lock of a container based on computer vision according to claim 2 is characterized in that: The server is deployed in the computer room and is used to run the image recognition algorithm and business software to perform digital image processing on the continuous images collected by the camera. The image recognition algorithm is used to detect whether there is any residual lock at the bottom of the container when the shore crane is operating in real time.

5. The method for detecting the bottom lock of a container based on computer vision according to claim 4 is characterized in that: A hard disk recorder is installed in the machine room to store the images of the front-end camera.

6. The method for detecting the bottom lock of a container based on computer vision according to claim 2 is characterized in that: The PLC industrial computer is installed under the calibration platform and is used to integrate all the information collected by the equipment and then feed back the corresponding information through other equipment.

7. The method for detecting the bottom lock of a container based on computer vision according to claim 2 is characterized in that: The signal light is installed under the calibration platform and is used to receive identification information from the server and provide feedback.

8. The method for detecting the bottom lock of a container based on computer vision according to claim 4 is characterized in that: The equipment in the machine room communicates with the equipment under the calibration platform via a switch.

Images