WO2022127311A1 - 一种自动化码头桥吊遗留锁垫检测方法及系统 - Google Patents

一种自动化码头桥吊遗留锁垫检测方法及系统 Download PDF

Info

Publication number
WO2022127311A1
WO2022127311A1 PCT/CN2021/123337 CN2021123337W WO2022127311A1 WO 2022127311 A1 WO2022127311 A1 WO 2022127311A1 CN 2021123337 W CN2021123337 W CN 2021123337W WO 2022127311 A1 WO2022127311 A1 WO 2022127311A1
Authority
WO
WIPO (PCT)
Prior art keywords
lock pad
lock
bottom corner
pad
box
Prior art date
Application number
PCT/CN2021/123337
Other languages
English (en)
French (fr)
Inventor
高仕博
唐波
张聪
张伯川
刘燕欣
郑智辉
徐安盛
魏小丹
闫涛
亓贺
Original Assignee
北京航天自动控制研究所
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京航天自动控制研究所 filed Critical 北京航天自动控制研究所
Publication of WO2022127311A1 publication Critical patent/WO2022127311A1/zh

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
    • B65G—TRANSPORT OR STORAGE DEVICES, e.g. CONVEYORS FOR LOADING OR TIPPING, SHOP CONVEYOR SYSTEMS OR PNEUMATIC TUBE CONVEYORS
    • B65G67/00—Loading or unloading vehicles
    • B65G67/60—Loading or unloading ships
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66C—CRANES; LOAD-ENGAGING ELEMENTS OR DEVICES FOR CRANES, CAPSTANS, WINCHES, OR TACKLES
    • B66C13/00—Other constructional features or details
    • B66C13/16—Applications of indicating, registering, or weighing devices
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66C—CRANES; LOAD-ENGAGING ELEMENTS OR DEVICES FOR CRANES, CAPSTANS, WINCHES, OR TACKLES
    • B66C13/00—Other constructional features or details
    • B66C13/18—Control systems or devices
    • B66C13/48—Automatic control of crane drives for producing a single or repeated working cycle; Programme control
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66C—CRANES; LOAD-ENGAGING ELEMENTS OR DEVICES FOR CRANES, CAPSTANS, WINCHES, OR TACKLES
    • B66C15/00—Safety gear
    • B66C15/06—Arrangements or use of warning devices

Definitions

  • the invention belongs to the technical field of target detection, and more particularly, relates to a method and system for detecting leftover lock pads of an automated dock bridge crane.
  • the unloading operation process of the automated container terminal is: the main trolley on the sea side of the bridge crane grabs the container from the container ship on the shore, and the main trolley on the sea side of the bridge crane moves and places the container on the pedestal of the transfer platform of the bridge crane; The personnel manually remove the lock pad at the bottom corner of the container; then, the bridge crane land side grabs the container from the pedestal of the transfer platform and moves it, and places it on the ground AGV automatic navigation truck, and the container is transported by the AGV automatic navigation truck to the designated stack. After the yard, the rail crane on the sea side of the yard grabs the container and puts the container in the designated yard position.
  • the lock pad control of the container entering the yard from the sea side is an important concern of the port industry. At present, the removal and confirmation of the lock pad depends on the on-site operators. , which will bring great safety hazards to the operations in the yard. Therefore, there is an urgent need for an automatic detection system for left lock pads of dock bridge cranes to detect and warn the missing lock pads, so as to achieve the goal of zero lock pad entry and avoid accidents in the yard.
  • the purpose of the present invention is to provide an automated method and system for detecting the left lock pad of a wharf bridge crane, aiming at solving the safety problem of the missing lock pad entering the yard.
  • An automated method for detecting leftover lock pads of a dock bridge crane comprising:
  • the lock pad detection result signal includes: a signal with a left lock pad and no left lock pad Signal;
  • the building a box bottom corner image library includes:
  • a container is placed on the platform of the bridge crane transfer platform; the container includes a 45-foot container, a 40-foot container, a double 20-foot container and a single 20-foot container;
  • Lock pads are respectively attached to the four bottom corners of the container; the lock pads include a large pad, a middle pad and a small pad;
  • a plurality of container bottom images are collected to establish the container bottom corner image library.
  • generating the lock pad target training sample data and the background training sample data according to the box bottom corner image library includes:
  • the lock pad target training sample data includes large lock pad samples , middle lock pad samples and small lock pad samples;
  • the generation of the legacy lock pad detection network model according to the lock pad target training sample data and the background training sample data includes:
  • the SSD target detection and classification network is trained;
  • L loc (x, l, g) is the position error
  • L conf (x, c) is the confidence error
  • N is the number of positive samples in the prior box
  • c is the category confidence prediction value
  • g is the position of the positive sample parameters
  • l is the bounding box prediction value
  • the category of the positive sample is p
  • ⁇ is the weight coefficient
  • the legacy lock pad detection network model is obtained by optimizing the training loss function through the lock pad target training sample data and the background training sample data.
  • the real-time video image of the box bottom corner in the operation is input into the legacy lock pad detection network model for detection, and the generation of the lock pad detection result signal includes:
  • a lock pad detection result signal is generated according to the prediction frame and the confidence level.
  • the generating the lock pad detection result signal according to the prediction frame and the confidence level includes:
  • the type of the lock pad includes a large lock pad, a middle lock pad and a small lock pad ;
  • An automated dock bridge crane left lock pad detection system comprising:
  • the real-time video image acquisition device of the bottom corner of the container is used to acquire the real-time video image of the bottom corner of the container on the platform of the bridge crane transfer platform;
  • a real-time processing system for leftover lock pad information connected with the real-time acquisition device for video images of the bottom corner of the box, for generating a lock pad detection result signal according to the real-time video image of the box bottom corner, and sending the lock pad detection result signal to the bridge crane
  • the landside slave trolley electronic control system; the lock pad detection result signal includes: a signal with a left lock pad and a signal without a left lock pad;
  • the bridge suspension land side is controlled by the trolley electronic control system to stop and lift from the trolley;
  • the bridge suspension land side is controlled by the trolley electronic control system to continue to lift from the trolley.
  • the lock pad legacy information real-time processing system includes:
  • the communication interface module is connected with the electronic control system of the slave trolley on the land side of the bridge crane, and is used to transmit the operation information of the slave trolley on the land side of the bridge crane; the operation information includes: the landing signal and the operation box type;
  • An information processing module connected with the communication interface module, is used for receiving the job information, and detecting the real-time video image of the bottom corner of the box according to the job information to generate a lock pad detection result signal, and converting the lock pad detection result
  • the signal is sent to the bridge crane land side from the trolley electronic control system.
  • the beneficial effect of the method and system for detecting leftover lock pads of an automated wharf bridge crane provided by the present invention is that compared with the prior art, the present invention obtains the bridge through the real-time acquisition device of the video image of the box bottom angle during the unloading operation of the automated wharf.
  • the real-time video image of the box bottom corner of the container on the pedestal of the crane transfer platform, the real-time processing system of the lock pad leftover information is used to generate the lock pad detection result signal according to the video image of the box bottom corner, and the lock pad detection result signal is sent to the bridge crane land side from the trolley Electronic control system; when there is a left lock pad, the electronic control system of the trolley on the land side of the bridge crane will control the land side of the bridge crane to stop lifting from the trolley, and warn the field operators to remove the lock pad in time.
  • the system detects the container in real time, which can effectively prevent the lock pads that are neglected by the field operators from entering the yard due to negligence, and prevent the lock pads from entering the yard and causing safety accidents.
  • Fig. 1 is a flow chart of a method for detecting leftover lock pads of an automated dock bridge crane according to an embodiment of the present invention
  • FIG. 2 is a schematic diagram of a network for providing SSD target detection and classification according to an embodiment of the present invention
  • FIG. 3 is a schematic diagram of a camera installation position according to an embodiment of the present invention.
  • FIG. 4 is a working flow chart of the automatic dock bridge crane left lock pad detection system provided by an embodiment of the present invention.
  • FIG. 1 A method for detecting leftover lock pads of an automated wharf bridge crane provided by the present invention will now be described. Please refer to FIG. 1 including:
  • Step 1 Obtain the image of the bottom corner of the box in advance, and establish an image library of the bottom corner of the box;
  • Step 1 specifically includes: placing a container on the platform of the bridge crane transfer platform; the container includes a 45-foot container, a 40-foot container, a double 20-foot container and a single 20-foot container;
  • the 4 bottom corners of the container are respectively hung with lock pads;
  • the lock pads include a large lock pad, a middle lock pad and a small lock pad;
  • a plurality of container bottom images are collected to establish the container bottom corner image library.
  • Step 2 Generate lock pad target training sample data and background training sample data according to the box bottom corner image library
  • Step 2 specifically includes:
  • lock pad target training sample data includes large lock pad samples, middle lock pad samples and small lock pad samples
  • the lock pad area is intercepted and marked in the image, and marked as a large lock pad, a middle lock pad, and a small lock pad, and the background is negative.
  • the samples are randomly intercepted from the non-lock pad target area in the image, and the lock pad target and background training sample data are obtained, and 8,000 positive samples of large pad targets, 7,000 positive samples of medium pad targets, and 4,000 positive samples of small pad targets can be obtained. , 50,000 background negative samples.
  • Step 3 Generate a legacy lock pad detection network model according to the lock pad target training sample data and the background training sample data;
  • Step 3 specifically includes:
  • L loc (x, l, g) is the position error
  • L conf (x, c) is the confidence error
  • N is the number of positive samples in the prior box
  • c is the category confidence prediction value
  • g is the position of the positive sample parameters
  • l is the bounding box prediction value
  • the category of the positive sample is p
  • ⁇ is the weight coefficient
  • the network model of the legacy lock pad detection is obtained.
  • the positive sample training data is first matched with the prior frame, and the positive sample training data The bounding box corresponding to the prior box matching the sample will be used to predict the positive sample; after the bounding box corresponding to the training sample is determined, the training loss function L(x,c,l,g) is defined as the position error Lloc ( Weighted sum of x,l,g) and confidence error L conf (x,c) where N is the number of positive samples of the prior box, c is the category confidence prediction value, g is the location parameter of the positive sample, l is the predicted value of the bounding box, and x represents is an indicator parameter, indicating that the i-th a priori frame matches the j-th pad target positive sample, and the category of the pad target positive sample is p; the training loss function L(x,c,l,g is optimized through the training samples ), the legacy lock pad detection network model can be pre-trained.
  • Step 4 Acquire the real-time video image of the bottom corner of the container during the operation; in the present invention, the real-time video image of the bottom corner of the container on the platform of the bridge crane transfer platform is acquired by using the real-time acquisition device of the video image of the bottom of the container.
  • Step 5 Input the real-time video image of the bottom corner of the box into the left lock pad detection network model for detection, and generate a lock pad detection result signal;
  • the lock pad detection result signal includes: a signal with a left lock pad and a signal without a left lock pad;
  • Step 5 specifically includes:
  • Step 501 Input the real-time video image at the bottom corner of the box into the legacy lock pad detection network model to obtain feature maps of different sizes;
  • the approximate pixel position of the marked bottom corner of the container in the video image of the bottom corner of the box is taken as the central part, and a 300 ⁇ 300 image is intercepted from the video image of the bottom corner of the box.
  • the 300 ⁇ 300 images are input into the pre-trained object detection classification network (legacy lock pad detection network model) to obtain feature maps of different sizes.
  • Step 502 Extract the feature maps of the Conv4_3, Conv7, Conv8_2, Conv9_2, Conv10_2, Conv11_2 layers, and then construct 6 bounding boxes of different scales at each point on the feature map respectively, classify the bounding boxes, and generate multiple Predicted boxes with class confidence.
  • Step 503 Generate a lock pad detection result signal according to the prediction frame and the confidence level.
  • Step 503 specifically includes:
  • the type of the lock pad includes a large lock pad, a middle lock pad and a small lock pad ;
  • the confidence threshold is set to 0.7.
  • the prediction frames with lower confidence are filtered out according to the confidence threshold of 0.7; the remaining prediction frames are decoded and arranged in descending order according to the confidence, and 200 are reserved.
  • the overlapping or incorrect prediction frames are removed by the non-maximum suppression method (NMS), and the remaining prediction frames are the final lock pad detection results, and then determine whether there is a lock pad target in the real-time video image of the bottom corner of the box.
  • NMS non-maximum suppression method
  • Step 6 When the lock pad detection result signal is a signal with a left lock pad, the suspension land side of the control bridge stops lifting from the trolley;
  • Step 7 When the lock pad detection result signal is that there is no remaining lock pad signal, the land side of the control bridge continues to lift from the trolley.
  • the invention discloses an automatic detection method for the left lock pad of a dock bridge crane, which is constructed based on the SSD target detection algorithm, has the advantages of fast recognition speed and high detection accuracy, and adopts the END-TO-END training method, even if The classification results are also very accurate when dealing with smaller resolution images.
  • the implementation case of the automatic dock bridge crane left lock pad detection system and method disclosed in the present invention can realize real-time processing for video images with a resolution of 1280 ⁇ 1080 and 25 frames per second, and the detection and identification of the left lock pad is correct. The rate is greater than 99%.
  • An automated dock bridge crane left lock pad detection system comprising:
  • the real-time acquisition device of the video image of the bottom corner of the box and the real-time processing system of the leftover information of the lock pad is used to acquire the video image of the bottom corner of the container on the pedestal of the bridge crane transfer platform in real time; All container bottom corners of 45-foot, 40-foot, double 20-foot, single 20-foot and other container types on the platform of the bridge crane transfer platform are observed and imaged; the real-time processing system for the information left by the lock pad is connected with the real-time acquisition device of the video image of the bottom corner of the box.
  • the lock pad detection result signal includes: pad signal and no legacy lock pad signal
  • the bridge suspension land side is controlled by the trolley electronic control system to stop and lift from the trolley;
  • the bridge suspension land side is controlled by the trolley electronic control system to continue to lift from the trolley.
  • the device for real-time acquisition of video images at the bottom corner of the box includes multiple sets of image acquisition sub-devices, and the preferred image acquisition sub-device in the present invention is a camera.
  • the camera of the present invention is preferably a Hikvision small hemisphere network camera.
  • On the pedestal of the bridge crane transfer platform install the camera on the vertical support of the transfer platform that is closest to the bottom corner of each box-shaped box.
  • the installation height of the camera is roughly the same as the bottom of the container placed on the transfer platform pedestal.
  • Each camera is used to observe the operation process.
  • the camera is closest to the bottom corner of the container, and marks the approximate pixel position of the bottom corner of the container in the observed image; the size of the camera does not exceed the width of the vertical support, and does not exceed the outer contour of the vertical support, and the camera has a fill light function;
  • the fixed bracket of the camera on the vertical bracket of the platform is connected with the vertical bracket of the transfer platform by welding;
  • the real-time processing system for lock pad legacy information includes: a communication interface module and an information processing module; the communication interface module is connected to the electronic control system of the slave trolley on the land side of the bridge suspension, and is used for transmitting the land side of the bridge suspension
  • the operation information from the trolley includes: the box loading signal and the operation box type; the information processing module is connected with the communication interface module to receive the operation information and generate the lock pad detection result signal according to the video image of the bottom corner of the detection box according to the operation information. , and send the lock pad detection result signal to the electronic control system of the slave trolley on the bridge suspension land side through the communication interface module to control the stop and lift of the slave trolley on the bridge suspension land side.
  • the information processing module reads the operation information such as the landing signal and the operation box type of the trolley PLC on the land side of the bridge crane through the communication interface module. After receiving the box landing signal, the information processing system starts to process the real-time video images of the four box bottom corners corresponding to the bottom corners of the operating box, and detects whether the four box bottom corners are detected in the real-time video images of the four box bottom corners through the automatic detection method of the lock pad left by the dock bridge crane. There is a lock pad, and the lock pad detection result signal is sent to the electronic control system of the slave trolley on the bridge suspension land side through the communication interface module to control the stop or lift of the slave trolley on the bridge suspension land side.
  • the information processing module When the information processing module detects the lock pad target in any real-time video image of the 4 video images, it will output the lock pad left alarm signal to the electronic control system of the slave trolley on the bridge suspension land side through the communication interface module.
  • the side trolley electronic control system stops the lifting of the bridge crane and the land side trolley, warns the field operators, and manually removes the left lock pad; when the information processing module does not detect the lock pad target in the 4 video images, the bridge crane The land side rises straight from the trolley.
  • the communication interface module is also used to transmit the regular heartbeat signal of the normal working state of the information processing module to the electronic control system of the slave trolley on the land side of the bridge suspension. If the trolley electronic control system on the bridge suspension land side does not receive the normal working status timing heartbeat signal of the information processing module within 2 consecutive heartbeat signal cycles, the bridge suspension land side slave trolley electronic control system locks the bridge suspension land side slave trolley from the trolley. Lift until the land side of the bridge crane can receive the normal working state timing heartbeat signal of the information processing module from the trolley electronic control system.
  • the installation position of the device for real-time acquisition of video images at the bottom corner of the box proposed by the present invention is to install one camera on each of the 10 vertical brackets on both sides of each pedestal on the bridge crane transfer platform, and a total of 10 cameras (5 installed on each side).
  • the installation height of the camera is roughly the same as the bottom of the container placed on the pedestal of the transfer platform, the camera adopts a small hemisphere camera with the function of fill light, the size of the camera does not exceed the width of the vertical bracket, and does not exceed the vertical bracket
  • the fixed bracket of the camera on the vertical support of the transfer platform is connected with the vertical support of the transfer platform by welding; adjust the angle of view of the camera to observe the bottom corner of the container closest to the camera, and mark the approximate pixels of the bottom corner of the container in the observed image Location.
  • This installation method enables the real-time acquisition device of the video image of the bottom corner of the box to observe the bottom corner of the box in all cases such as 45-foot container, 40-foot container, double 20-foot container and single 20-foot container placed on the platform of the bridge crane transfer platform.
  • the work flow of the present invention is as follows: the information processing module of the lock pad legacy information real-time processing system reads the operation information such as the landing signal from the trolley PLC on the land side of the bridge crane, the operation box type of the container, etc. through the communication interface module. , After the information processing module receives the container landing signal from the trolley PLC on the land side of the bridge crane, the information processing system starts to process the video images of the four bottom corners of the container corresponding to the bottom corners of the container. Detect the real-time video images of the bottom corners of the 4 boxes. If a large lock pad target, a middle lock pad target or a small lock pad target is detected in any of the 4 video images, the lock pad is output through the communication interface module.
  • the pad left alarm signal is sent to the electric control system of the trolley from the land side of the bridge suspension, and the electric control system of the trolley from the land side of the bridge suspension stops the lifting of the trolley from the land side of the bridge suspension to warn the field operators to manually remove the left lock pad; After the big lock pad, the on-site operators resumed the lifting operation from the trolley on the land side of the bridge by operating the bypass switch.
  • the beneficial effect of the method and system for detecting leftover lock pads of an automated wharf bridge crane provided by the present invention is that compared with the prior art, the present invention obtains the bridge through the real-time acquisition device of the video image of the box bottom angle during the unloading operation of the automated wharf.
  • the real-time video image of the box bottom corner of the container on the pedestal of the crane transfer platform, the real-time processing system of the lock pad leftover information is used to generate the lock pad detection result signal according to the video image of the box bottom corner, and the lock pad detection result signal is sent to the bridge crane land side from the trolley Electronic control system; when there is a left lock pad, the electronic control system of the trolley from the land side of the bridge crane will control the land side of the bridge crane to stop lifting from the trolley, so as to remind the field operators to remove the lock pad in time.
  • the processing system detects the containers in real time, which can effectively prevent the lock pads that the field operators neglect to pick from entering the yard due to negligence, and avoid safety accidents in the yard.

Landscapes

  • Engineering & Computer Science (AREA)
  • Mechanical Engineering (AREA)
  • Automation & Control Theory (AREA)
  • Ocean & Marine Engineering (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Control And Safety Of Cranes (AREA)

Abstract

一种自动化码头桥吊遗留锁垫检测方法及系统，属于目标检测技术领域，检测系统，包括箱底角视频图像实时获取装置和锁垫遗留信息实时处理系统。在自动化码头卸船作业过程中，通过箱底角视频图像实时获取装置获取桥吊中转平台台座上集装箱的箱底角实时视频图像，采用锁垫遗留信息实时处理系统根据箱底角实时视频图像生成锁垫检测结果信号，并将锁垫检测结果信号发送到桥吊陆侧从小车电控系统；在有遗留锁垫时，控制桥吊陆侧从小车停止起升，警告现场作业人员及时摘除锁垫，通过锁垫遗留信息实时处理系统实时对集装箱进行检测，可以有效防止现场作业人员因疏忽漏摘的锁垫进入堆场，避免锁垫进入堆场内发生安全事故。

Description

一种自动化码头桥吊遗留锁垫检测方法及系统 技术领域
本发明属于目标检测技术领域，更具体地说，是涉及一种自动化码头桥吊遗留锁垫检测方法及系统。
背景技术
目前，自动化集装箱码头的卸船作业流程是：桥吊海侧主小车从岸边的集装箱船上抓起集装箱，桥吊海侧主小车将集装箱移动和放置到桥吊中转平台的台座上；现场作业人员手动拆除集装箱底角的锁垫；然后，桥吊陆侧从小车将集装箱从中转平台的台座上抓起移动，放置到地面AGV自动导航卡车上，集装箱被AGV自动导航卡车运送到指定的堆场后，堆场海侧轨道吊抓起集装箱，将集装箱放到指定的堆场位置。
现场作业人员有时会漏摘桥吊转运平台上集装箱底角的锁垫，尤其是带有自锁的锁垫。若未摘除的锁垫没有被发现，集装箱放置到堆场时会损坏下层集装箱，更为严重的是带有自锁遗留锁垫的集装箱与下层集装箱会产生固联，后续当海侧轨道吊吊起该集装箱时，会将下层集装箱一同吊起，导致严重的摔箱事故。
海侧进堆场集装箱的锁垫控制是港口行业的重要关注点，目前锁垫拆除和确认是依靠现场作业人员，如果现场作业人员因疏忽漏摘掉桥吊转运平台上集装箱底角的锁垫，会给堆场内作业带来极大安全隐患。因此亟需一种自动化码头桥吊遗留锁垫检测系统对漏摘的锁垫进行检测预警，实现锁垫零进场目标，避免在堆场内发生事故。
发明内容
本发明的目的在于提供一种自动化码头桥吊遗留锁垫检测方法及系统，旨在解决遗漏锁垫进堆场的安全问题。
为实现上述目的，本发明采用的技术方案是：
一种自动化码头桥吊遗留锁垫检测方法，包括：
事先获取箱底角图像，建立箱底角图像库；
根据所述箱底角图像库生成锁垫目标训练样本数据和背景训练样本数据；
根据所述锁垫目标训练样本数据和所述背景训练样本数据生成遗留锁垫检测网络模型；
获取作业中的箱底角实时视频图像；
将所述作业中的箱底角实时视频图像输入到所述遗留锁垫检测网络模型进行检测，生成锁垫检测结果信号；所述锁垫检测结果信号包括：具有遗留锁垫信号和没有遗留锁垫信号；
当所述锁垫检测结果信号为具有遗留锁垫信号时，控制桥吊陆侧从小车停止起升；
当所述锁垫检测结果信号为没有遗留锁垫信号时，控制桥吊陆侧从小车继续起升。
优选的，所述建立箱底角图像库，包括：
在桥吊中转平台台座上放置集装箱；所述集装箱包括45尺集装箱、40尺集装箱、双20尺集装箱和单20尺集装箱；
在所述集装箱的4个底角分别挂上锁垫；所述锁垫包括大锁垫、中锁垫和小锁垫；
采集多个集装箱箱底图像建立所述箱底角图像库。
优选的，所述根据所述箱底角图像库生成锁垫目标训练样本数据和背景训练样本数据，包括：
对所述箱底角图像库中任意一个带锁垫的箱底角图像中的锁垫区域进行截 取和标注，生成所述锁垫目标训练样本数据；所述锁垫目标训练样本数据包括大锁垫样本、中锁垫样本和小锁垫样本；
对所述箱底角图像库中任意一个图像中的非锁垫区域进行随机截取，生成所述背景训练样本数据。
优选的，所述根据所述锁垫目标训练样本数据和所述背景训练样本数据生成遗留锁垫检测网络模型包括：
以所述锁垫目标训练样本数据为正样本，以所述背景训练样本数据为负样本，对SSD目标检测分类网络进行训练；
定义训练损失函数
Figure PCTCN2021123337-appb-000001
其中，L loc(x,l,g)为位置误差，L conf(x,c)为置信度误差，N是先验框的正样本数量，c是类别置信度预测值，g是正样本的位置参数，l是边界框预测值，
Figure PCTCN2021123337-appb-000002
是指示参数，表示第i个先验框与第j个正样本匹配，并且所述正样本的类别为p，α为权重系数；
通过所述锁垫目标训练样本数据和所述背景训练样本数据优化所述训练损失函数得到所述遗留锁垫检测网络模型。
优选的，将所述作业中的箱底角实时视频图像输入到所述遗留锁垫检测网络模型进行检测，生成锁垫检测结果信号包括：
将所述作业中的箱底角实时视频图像输入到遗留锁垫检测网络模型中，获得不同大小的特征映射；
抽取Conv4_3、Conv7、Conv8_2、Conv9_2、Conv10_2、Conv11_2层的特征图，然后分别在所述特征图上面的每一个点构造6个不同尺度大小的边界框，对边界框分别进行分类，生成多个带有类别置信度的预测框；
根据所述预测框和置信度生成锁垫检测结果信号。
优选的，所述根据所述预测框和置信度生成锁垫检测结果信号包括：
设置置信度阈值；
根据所述类别置信度和置信度阈值过滤掉类别置信度低于置信度阈值的预 测框，生成过滤背景信息后的待处理预测框；
将所述待处理预测框进行解码，采用非极大值抑制方法去掉重叠或者不正确的预测框，确定锁垫的类别；所述锁垫的类别包括大锁垫、中锁垫和小锁垫；
生成锁垫检测结果信号。
一种自动化码头桥吊遗留锁垫检测系统，包括：
箱底角视频图像实时获取装置,用于获取桥吊中转平台台座上集装箱的箱底角实时视频图像；
锁垫遗留信息实时处理系统，与所述箱底角视频图像实时获取装置连接，用于根据所述箱底角实时视频图像生成锁垫检测结果信号，并将所述锁垫检测结果信号发送到桥吊陆侧从小车电控系统；所述锁垫检测结果信号包括：具有遗留锁垫信号和没有遗留锁垫信号；
当检测锁垫检测结果信号为具有遗留锁垫信号时，桥吊陆侧从小车电控系统控制桥吊陆侧从小车停止起升；
当检测锁垫检测结果信号为没有遗留锁垫信号时，桥吊陆侧从小车电控系统控制桥吊陆侧从小车继续起升。
优选的，所述锁垫遗留信息实时处理系统包括：
通迅接口模块，与桥吊陆侧从小车电控系统连接，用于传输桥吊陆侧从小车的作业信息；所述作业信息包括：着箱信号和作业箱型；
信息处理模块，与所述通迅接口模块连接，用于接受所述作业信息，并根据所述作业信息检测所述箱底角实时视频图像生成锁垫检测结果信号，并将所述锁垫检测结果信号发送到桥吊陆侧从小车电控系统。
本发明提供的一种自动化码头桥吊遗留锁垫检测方法及系统的有益效果在于：与现有技术相比，本发明在自动化码头卸船作业过程中，通过箱底角视频图像实时获取装置获取桥吊中转平台台座上集装箱的箱底角实时视频图像，采用锁垫遗留信息实时处理系统根据箱底角视频图像生成锁垫检测结果信号，并将所述锁垫检测结果信号发送到桥吊陆侧从小车电控系统；在有遗留锁垫时， 桥吊陆侧从小车电控系统会控制桥吊陆侧从小车停止起升，警告现场作业人员及时摘除锁垫，本发明通过锁垫遗留信息实时处理系统实时对集装箱进行检测，可以有效防止现场作业人员因疏忽漏摘的锁垫进入堆场，避免锁垫进入堆场内发生安全事故。
附图说明
为了更清楚地说明本发明实施例中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本发明实施例提供的一种自动化码头桥吊遗留锁垫检测方法的流程图；
图2为本发明实施例提供的提供SSD目标检测分类网络示意图；
图3为本发明实施例提供的相机安装位置示意图；
图4为本发明实施例提供的自动化码头桥吊遗留锁垫检测系统的工作流程图。
具体实施方式
为了使本发明所要解决的技术问题、技术方案及有益效果更加清楚明白，以下结合附图及实施例，对本发明进行进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。
现对本发明提供的一种自动化码头桥吊遗留锁垫检测方法进行说明，请参阅图1包括：
步骤1：事先获取箱底角图像，建立箱底角图像库；
步骤1具体包括：在桥吊中转平台台座上放置集装箱；所述集装箱包括45尺集装箱、40尺集装箱、双20尺集装箱和单20尺集装箱；
在所述集装箱的4个底角分别挂上锁垫；所述锁垫包括大锁垫、中锁垫和 小锁垫；
采集多个集装箱箱底图像建立所述箱底角图像库。
在实际应用中，在桥吊转运平台台座上分别放置45尺集装箱、40尺集装箱、双20尺集装箱和单20尺集装箱，对不同尺寸的集装箱，在箱底角都分别安装大锁垫、中锁垫和小锁垫，在白天和晚上采集带锁垫的箱底角图像，构建数量不少于1万张的箱底角图像库。
步骤2：根据箱底角图像库生成锁垫目标训练样本数据和背景训练样本数据；
步骤2具体包括：
对箱底角图像库中任意一个图像中的锁垫区域进行截取和标注，生成锁垫目标训练样本数据；锁垫目标训练样本数据包括大锁垫样本、中锁垫样本和小锁垫样本；
对箱底角图像库中任意一个图像中的非锁垫区域进行随机截取，生成背景训练样本数据；
在实际应用中，对箱底角图像库中任一个带锁垫的箱底角图像，在图像中对锁垫区域进行截取和标注，并标记为大锁垫、中锁垫、小锁垫，背景负样本从图像中非锁垫目标区域随机截取，获取锁垫目标和背景训练样本数据，可以得到8000个大锁垫目标正样本、7000个中锁垫目标正样本、4000个小锁垫目标正样本，5万个背景负样本。
步骤3：根据锁垫目标训练样本数据和背景训练样本数据生成遗留锁垫检测网络模型；
步骤3具体包括：
以锁垫目标训练样本数据为正样本，以背景训练样本数据为负样本，对SSD目标检测分类网络进行训练；
定义训练损失函数
Figure PCTCN2021123337-appb-000003
其中，L loc(x,l,g)为位置误差，L conf(x,c)为置信度误差，N是先验框的正样本数量，c是类别置信度 预测值，g是正样本的位置参数，l是边界框预测值，
Figure PCTCN2021123337-appb-000004
是指示参数，表示第i个先验框与第j个正样本匹配，并且正样本的类别为p，α为权重系数；
通过锁垫目标训练样本数据和背景训练样本数据优化训练损失函数得到遗留锁垫检测网络模型。
在实际应用中，请参考图2，其中，Image表示图像，conv表示卷积层，FC表示全连接层，classifler表示分类器，Classes为类别，Extra Feature Layers表示额外的功能层，detections表示分类结果。根据锁垫目标训练样本数据和背景训练样本数据，训练一个SSD目标检测分类网络；SSD的基础网络部分采用了VGG16，在训练过程中，先将正样本训练数据与先验框进行匹配，与正样本相匹配的先验框所对应的边界框将用于预测该正样本；在训练样本对应的边界框确定后，定义训练损失函数L(x,c,l,g)为位置误差L loc(x,l,g)与置信度误差L conf(x,c)的加权和
Figure PCTCN2021123337-appb-000005
其中N是先验框的正样本数量，c是类别置信度预测值，g是正样本的位置参数，l是边界框预测值，x表示
Figure PCTCN2021123337-appb-000006
是指示参数，表示第i个先验框与第j个锁垫目标正样本匹配，并且该锁垫目标正样本的类别为p；通过训练样本优化训练损失函数L(x,c,l,g)，可以预训练完成遗留锁垫检测网络模型。
步骤4：获取作业中的箱底角实时视频图像；本发明中通过采用箱底角视频图像实时获取装置获取桥吊中转平台台座上集装箱的箱底角实时视频图像。
步骤5：将箱底角实时视频图像输入到遗留锁垫检测网络模型进行检测，生成锁垫检测结果信号；锁垫检测结果信号包括：具有遗留锁垫信号和没有遗留锁垫信号；
步骤5具体包括：
步骤501：在箱底角实时视频图像输入到遗留锁垫检测网络模型中，获得不同大小的特征映射；
在实际应用中，对实时获取的箱底角实时视频图像，以标记的集装箱底角 在箱底角视频图像中的大致像素位置为中央部位，在箱底角视频图像中截取一个300×300图像，将该300×300图像输入到预训练好的目标检测分类网络中(遗留锁垫检测网络模型)获得不同大小的特征映射。
步骤502：抽取Conv4_3、Conv7、Conv8_2、Conv9_2、Conv10_2、Conv11_2层的特征图，然后分别在特征图上面的每一个点构造6个不同尺度大小的边界框，对边界框分别进行分类，生成多个带有类别置信度的预测框。
步骤503：根据预测框和置信度生成锁垫检测结果信号。
步骤503具体包括：
设置置信度阈值；
根据所述类别置信度和置信度阈值过滤掉类别置信度低于置信度阈值的预测框，生成过滤背景信息后的待处理预测框；
将所述待处理预测框进行解码，采用非极大值抑制方法去掉重叠或者不正确的预测框，确定锁垫的类别；所述锁垫的类别包括大锁垫、中锁垫和小锁垫；
生成锁垫检测结果信号。
在实际应用中，设置置信度阈值0.7，对于每个预测框，根据置信度阈值0.7，过滤掉置信度较低的预测框；对于余下的预测框进行解码，按照置信度进行降序排列，保留200个预测框；再经过非极大值抑制方法(NMS)去掉重叠或者不正确的预测框，剩余的预测框就是最终锁垫检测结果，进而判断箱底角实时视频图像中是否存在锁垫目标。
步骤6：当锁垫检测结果信号为具有遗留锁垫信号时，控制桥吊陆侧从小车停止起升；
步骤7：当锁垫检测结果信号为没有遗留锁垫信号时，控制桥吊陆侧从小车继续起升。
本发明公开的一种自动化码头桥吊遗留锁垫检测方法，是基于SSD目标检测算法所构建的，具有识别速度快和检测精度高的优点，并且采用了END-TO-END的训练方式，即使在处理分辨率较小的图片，分类结果也很准确。 根据试验数据统计，本发明公开的自动化码头桥吊遗留锁垫检测系统及方法的实施案例，对于1280×1080分辨率、25帧/秒的视频图像可实现实时处理，遗留锁垫的检测识别正确率大于99％。
现对本发明提供的一种自动化码头桥吊遗留锁垫检测系统进行说明。一种自动化码头桥吊遗留锁垫检测系统，包括：
箱底角视频图像实时获取装置和锁垫遗留信息实时处理系统；箱底角视频图像实时获取装置用于实时获取桥吊中转平台台座上集装箱的箱底角视频图像；箱底角实时视频图像是可对放置在桥吊中转平台台座上的45尺、40尺、双20尺、单20尺等集装箱型的所有箱底角进行观测成像的；锁垫遗留信息实时处理系统，与箱底角视频图像实时获取装置连接，用于接受箱底角实时视频图像并根据箱底角视频图像生成锁垫检测结果信号，并将锁垫检测结果信号发送到桥吊陆侧从小车电控系统；锁垫检测结果信号包括：具有遗留锁垫信号和没有遗留锁垫信号；
当检测锁垫检测结果信号为具有遗留锁垫信号时，桥吊陆侧从小车电控系统控制桥吊陆侧从小车停止起升；
当检测锁垫检测结果信号为没有遗留锁垫信号时，桥吊陆侧从小车电控系统控制桥吊陆侧从小车继续起升。
在实际应用中，箱底角视频图像实时获取装置，包括多组图像获取子装置，本发明优选图像获取子装置为相机。本发明相机优选为采用海康威视小半球网络摄像机。在桥吊转运平台台座上，将相机安装在与各箱型箱底角距离最近的转运平台竖支架上，相机安装高度与转运平台台座上放置的集装箱底大体持平，每个相机用于观测作业过程中桥吊转运平台台座上放置的45尺集装箱、40尺集装箱、双20尺集装箱和单20尺集装箱等箱型的箱底角；转运平台竖支架上的相机，可以通过调整相机的视角去观测距离本相机最近的集装箱底角，并标记集装箱底角在观测的图像中的大致像素位置；相机的尺寸不超过竖支架的宽度，不越过竖支架的外轮廓，且相机具有补光灯功能；转运平台竖支架上相机 的固定支架与转运平台竖支架采用焊接方式相连；转运平台竖支架上相机的上方安装遮雨罩，且不越过竖支架的外轮廓。
作为本申请另一实施例，锁垫遗留信息实时处理系统包括：通迅接口模块和信息处理模块；通迅接口模块，与桥吊陆侧从小车电控系统连接，用于传输桥吊陆侧从小车的作业信息；作业信息包括：着箱信号和作业箱型；信息处理模块，与通迅接口模块连接，用于接受作业信息，并根据作业信息检测箱底角视频图像生成锁垫检测结果信号，并将锁垫检测结果信号通过通迅接口模块发送到桥吊陆侧从小车电控系统，控制桥吊陆侧从小车的停止和起升。
作为本申请另一实施例，信息处理模块通过通迅接口模块读取桥吊陆侧从小车PLC的着箱信号、作业箱型等作业信息，信息处理模块在接收到桥吊陆侧从小车PLC的着箱信号后，信息处理系统开始处理与作业箱型底角对应的4个箱底角实时视频图像，通过自动化码头桥吊遗留锁垫检测方法，分别在4个箱底角实时视频图像中检测是否有锁垫，并将锁垫检测结果信号通过通迅接口模块发给桥吊陆侧从小车电控系统，控制桥吊陆侧从小车的停止或起升。
当信息处理模块在4个视频图像中的任一个实时视频图像中检测到锁垫目标时，则通过通迅接口模块输出锁垫遗留报警信号给桥吊陆侧从小车电控系统，桥吊陆侧从小车电控系统停止桥吊陆侧从小车的起升，警告现场作业人员，手动拆卸遗留的锁垫；当信息处理模块在4个视频图像中都没有检测到锁垫目标时，桥吊陆侧从小车一直起升。在中转平台处还设置有旁路开关，现场作业人员在拆卸遗留的锁垫后，通过操作旁路开关恢复桥吊陆侧从小车起升作业。
通迅接口模块还用于将信息处理模块的正常工作状态定时心跳信号传输给桥吊陆侧从小车电控系统。如果桥吊陆侧从小车电控系统连续2个心跳信号周期内都没有收到信息处理模块的正常工作状态定时心跳信号，则桥吊陆侧从小车电控系统锁定桥吊陆侧从小车的起升，直到桥吊陆侧从小车电控系统能收到信息处理模块的正常工作状态定时心跳信号。
在实际应用中，本发明提出的箱底角视频图像实时获取装置安装位置在桥 吊转运平台上每个台座两侧的10个竖支架上各安装1个相机，总10个相机(每侧安装5个相机)，请参阅图3，相机安装高度与转运平台台座上放置的集装箱底大体持平，相机采用具有补光灯功能的小半球相机，相机的尺寸不超过竖支架的宽度，不越过竖支架的外轮廓；转运平台竖支架上相机的固定支架与转运平台竖支架采用焊接方式相连；调整相机的视角去观测距离本相机最近的集装箱底角，并标记集装箱底角在观测图像中的大致像素位置。这种安装方式使得箱底角视频图像实时获取装置可观测到桥吊转运平台台座上放置的45尺集装箱、40尺集装箱、双20尺集装箱和单20尺集装箱等箱型所有情况下的箱底角。
请参考图4，本发明的工作流程为：锁垫遗留信息实时处理系统的信息处理模块通过通迅接口模块读取桥吊陆侧从小车PLC的着箱信号、集装箱的作业箱型等作业信息，信息处理模块在接受到桥吊陆侧从小车PLC的着箱信号后，信息处理系统开始处理与集装箱底角对应的4个箱底角视频图像，通过自动化码头桥吊遗留锁垫检测方法，分别对4个箱底角实时视频图像中进行检测，在4个视频图像中的任一个视频图像中检测到有大锁垫目标或中锁垫目标或小锁垫目标，则通过通迅接口模块输出锁垫遗留报警信号给桥吊陆侧从小车电控系统，桥吊陆侧从小车电控系统停止桥吊陆侧从小车的起升，警告现场作业人员，手动拆卸遗留的锁垫；在拆卸遗留大锁垫后，现场作业人员通过操作旁路开关恢复桥吊陆侧从小车起升作业。
本发明提供的一种自动化码头桥吊遗留锁垫检测方法及系统的有益效果在于：与现有技术相比，本发明在自动化码头卸船作业过程中，通过箱底角视频图像实时获取装置获取桥吊中转平台台座上集装箱的箱底角实时视频图像，采用锁垫遗留信息实时处理系统根据箱底角视频图像生成锁垫检测结果信号，并将所述锁垫检测结果信号发送到桥吊陆侧从小车电控系统；在有遗留锁垫时，桥吊陆侧从小车电控系统会控制桥吊陆侧从小车停止起升，以提醒现场作业人员及时摘除锁垫，本发明通过锁垫遗留信息实时处理系统实时对集装箱进行检 测，可以有效防止现场作业人员因疏忽漏摘的锁垫进入堆场，避免在堆场内发生安全事故。
以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等，均应包含在本发明的保护范围之内。

Claims (8)

  1. 一种自动化码头桥吊遗留锁垫检测方法，其特征在于，包括：
    事先获取箱底角图像，建立箱底角图像库；
    根据所述箱底角图像库生成锁垫目标训练样本数据和背景训练样本数据；
    根据所述锁垫目标训练样本数据和所述背景训练样本数据生成遗留锁垫检测网络模型；
    获取作业中的箱底角实时视频图像；
    将所述作业中的箱底角实时视频图像输入到所述遗留锁垫检测网络模型进行检测，生成锁垫检测结果信号；所述锁垫检测结果信号包括：具有遗留锁垫信号和没有遗留锁垫信号；
    当所述锁垫检测结果信号为具有遗留锁垫信号时，控制桥吊陆侧从小车停止起升；
    当所述锁垫检测结果信号为没有遗留锁垫信号时，控制桥吊陆侧从小车继续起升。
  2. 如权利要求1所述的一种自动化码头桥吊遗留锁垫检测方法，其特征在于，所述建立箱底角图像库，包括：
    在桥吊中转平台台座上放置集装箱；所述集装箱包括45尺集装箱、40尺集装箱、双20尺集装箱和单20尺集装箱；
    在所述集装箱的4个底角分别挂上锁垫；所述锁垫包括大锁垫、中锁垫和小锁垫；
    采集多个集装箱箱底图像建立所述箱底角图像库。
  3. 如权利要求2所述的一种自动化码头桥吊遗留锁垫检测方法，其特征在于，所述根据所述箱底角图像库生成锁垫目标训练样本数据和背景训练样本数据，包括：
    对所述箱底角图像库中任意一个带锁垫的箱底角图像中的锁垫区域进行截取和标注，生成所述锁垫目标训练样本数据；所述锁垫目标训练样本数据包括 大锁垫样本、中锁垫样本和小锁垫样本；
    对所述箱底角图像库中任意一个图像中的非锁垫区域进行随机截取，生成所述背景训练样本数据。
  4. 如权利要求3所述的一种自动化码头桥吊遗留锁垫检测方法，其特征在于，所述根据所述锁垫目标训练样本数据和所述背景训练样本数据生成遗留锁垫检测网络模型包括：
    以所述锁垫目标训练样本数据为正样本，以所述背景训练样本数据为负样本，对SSD目标检测分类网络进行训练；
    定义训练损失函数
    Figure PCTCN2021123337-appb-100001
    其中，L loc(x,l,g)为位置误差，L conf(x,c)为置信度误差，N是先验框的正样本数量，c是类别置信度预测值，g是正样本的位置参数，l是边界框预测值，
    Figure PCTCN2021123337-appb-100002
    是指示参数，表示第i个先验框与第j个正样本匹配，并且所述正样本的类别为p，α为权重系数；
    通过所述锁垫目标训练样本数据和所述背景训练样本数据优化所述训练损失函数得到所述遗留锁垫检测网络模型。
  5. 如权利要求4所述的一种自动化码头桥吊遗留锁垫检测方法，其特征在于，将所述作业中的箱底角实时视频图像输入到所述遗留锁垫检测网络模型进行检测，生成锁垫检测结果信号包括：
    将所述作业中的箱底角实时视频图像输入到遗留锁垫检测网络模型中，获得不同大小的特征映射；
    抽取Conv4_3、Conv7、Conv8_2、Conv9_2、Conv10_2、Conv11_2层的特征图，然后分别在所述特征图上面的每一个点构造6个不同尺度大小的边界框，对边界框分别进行分类，生成多个带有类别置信度的预测框；
    根据所述预测框和置信度生成锁垫检测结果信号。
  6. 如权利要求5所述的一种自动化码头桥吊遗留锁垫检测方法，其特征在于，所述根据所述预测框和置信度生成锁垫检测结果信号包括：
    设置置信度阈值；
    根据所述类别置信度和置信度阈值过滤掉类别置信度低于置信度阈值的预测框，生成过滤背景信息后的待处理预测框；
    将所述待处理预测框进行解码，采用非极大值抑制方法去掉重叠或者不正确的预测框，确定锁垫的类别；所述锁垫的类别包括大锁垫、中锁垫和小锁垫；
    生成锁垫检测结果信号。
  7. 一种自动化码头桥吊遗留锁垫检测系统，其特征在于，包括：
    箱底角视频图像实时获取装置,用于获取桥吊中转平台台座上集装箱的箱底角实时视频图像；
    锁垫遗留信息实时处理系统，与所述箱底角视频图像实时获取装置连接，用于根据所述箱底角实时视频图像生成锁垫检测结果信号，并将所述锁垫检测结果信号发送到桥吊陆侧从小车电控系统；所述锁垫检测结果信号包括：具有遗留锁垫信号和没有遗留锁垫信号；
    当检测锁垫检测结果信号为具有遗留锁垫信号时，桥吊陆侧从小车电控系统控制桥吊陆侧从小车停止起升；
    当检测锁垫检测结果信号为没有遗留锁垫信号时，桥吊陆侧从小车电控系统控制桥吊陆侧从小车继续起升。
  8. 如权利要求7所述的一种自动化码头桥吊遗留锁垫检测系统，其特征在于，所述锁垫遗留信息实时处理系统包括：
    通迅接口模块，与桥吊陆侧从小车电控系统连接，用于传输桥吊陆侧从小车的作业信息；所述作业信息包括：着箱信号和作业箱型；
    信息处理模块，与所述通迅接口模块连接，用于接受所述作业信息，并根据所述作业信息检测所述箱底角实时视频图像生成锁垫检测结果信号，并将所述锁垫检测结果信号发送到桥吊陆侧从小车电控系统。
PCT/CN2021/123337 2020-12-17 2021-10-12 一种自动化码头桥吊遗留锁垫检测方法及系统 WO2022127311A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011502610.3A CN112661013B (zh) 2020-12-17 2020-12-17 一种自动化码头桥吊遗留锁垫检测方法及系统
CN202011502610.3 2020-12-17

Publications (1)

Publication Number Publication Date
WO2022127311A1 true WO2022127311A1 (zh) 2022-06-23

Family

ID=75406325

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/123337 WO2022127311A1 (zh) 2020-12-17 2021-10-12 一种自动化码头桥吊遗留锁垫检测方法及系统

Country Status (2)

Country Link
CN (1) CN112661013B (zh)
WO (1) WO2022127311A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115293443A (zh) * 2022-08-17 2022-11-04 上海国际港务（集团）股份有限公司 桥吊和集装箱船舶装卸作业时间的预测方法、系统及介质
CN116074616A (zh) * 2023-04-06 2023-05-05 上海知率智能科技有限公司 集装箱图像采集系统

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112661013B (zh) * 2020-12-17 2023-06-30 北京航天自动控制研究所 一种自动化码头桥吊遗留锁垫检测方法及系统

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2014174021A (ja) * 2013-03-08 2014-09-22 Mitsui Eng & Shipbuild Co Ltd トラック搭載コンテナの重心計測システム及び重心計測方法
CN106203239A (zh) * 2015-05-04 2016-12-07 杭州海康威视数字技术股份有限公司 用于集装箱理货的信息处理方法、装置和系统
CN107680092A (zh) * 2017-10-12 2018-02-09 中科视拓（北京）科技有限公司 一种基于深度学习的集装箱锁扣检测及预警方法
CN108491851A (zh) * 2018-01-29 2018-09-04 江苏大学 一种基于机器视觉的集装箱锁孔快速识别与吊具纠偏方法
CN109335964A (zh) * 2018-09-21 2019-02-15 北京航天自动控制研究所 一种集装箱旋锁检测系统及检测方法
CN111275682A (zh) * 2020-01-19 2020-06-12 上海箱云物流科技有限公司 集装箱检测方法、装置及计算机可读存储介质
CN112661013A (zh) * 2020-12-17 2021-04-16 北京航天自动控制研究所 一种自动化码头桥吊遗留锁垫检测方法及系统

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103207879B (zh) * 2012-01-17 2016-03-30 阿里巴巴集团控股有限公司 图像索引的生成方法及设备
CN106600581B (zh) * 2016-12-02 2019-09-24 北京航空航天大学 一种基于双目立体视觉的列车运行故障自动检测系统及方法
CN108846440B (zh) * 2018-06-20 2023-06-02 腾讯科技（深圳）有限公司 图像处理方法及装置、计算机可读介质和电子设备
CN110033026B (zh) * 2019-03-15 2021-04-02 深圳先进技术研究院 一种连续小样本图像的目标检测方法、装置及设备
CN211644372U (zh) * 2019-12-20 2020-10-09 青岛港国际股份有限公司 一种用于岸桥的集装箱锁垫框吊装装置

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2014174021A (ja) * 2013-03-08 2014-09-22 Mitsui Eng & Shipbuild Co Ltd トラック搭載コンテナの重心計測システム及び重心計測方法
CN106203239A (zh) * 2015-05-04 2016-12-07 杭州海康威视数字技术股份有限公司 用于集装箱理货的信息处理方法、装置和系统
CN107680092A (zh) * 2017-10-12 2018-02-09 中科视拓（北京）科技有限公司 一种基于深度学习的集装箱锁扣检测及预警方法
CN108491851A (zh) * 2018-01-29 2018-09-04 江苏大学 一种基于机器视觉的集装箱锁孔快速识别与吊具纠偏方法
CN109335964A (zh) * 2018-09-21 2019-02-15 北京航天自动控制研究所 一种集装箱旋锁检测系统及检测方法
CN111275682A (zh) * 2020-01-19 2020-06-12 上海箱云物流科技有限公司 集装箱检测方法、装置及计算机可读存储介质
CN112661013A (zh) * 2020-12-17 2021-04-16 北京航天自动控制研究所 一种自动化码头桥吊遗留锁垫检测方法及系统

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115293443A (zh) * 2022-08-17 2022-11-04 上海国际港务（集团）股份有限公司 桥吊和集装箱船舶装卸作业时间的预测方法、系统及介质
CN115293443B (zh) * 2022-08-17 2024-03-22 上海国际港务（集团）股份有限公司 桥吊和集装箱船舶装卸作业时间的预测方法、系统及介质
CN116074616A (zh) * 2023-04-06 2023-05-05 上海知率智能科技有限公司 集装箱图像采集系统

Also Published As

Publication number Publication date
CN112661013A (zh) 2021-04-16
CN112661013B (zh) 2023-06-30

Similar Documents

Publication Publication Date Title
WO2022127311A1 (zh) 一种自动化码头桥吊遗留锁垫检测方法及系统
US7508956B2 (en) Systems and methods for monitoring and tracking movement and location of shipping containers and vehicles using a vision based system
EP3033293B1 (en) Method and system for automatically landing containers on a landing target using a container crane
CN104085794B (zh) 一种用于集装箱分布轮廓和位置智能化检测系统及方法
CN109335964B (zh) 一种集装箱旋锁检测系统及检测方法
CN105893940B (zh) 基于边缘检测的集装箱起吊防撞定准系统的实现方法
CN103879741A (zh) 一种用于散货料场的激光防碰撞预警系统
CN206751220U (zh) 一种自动集装箱堆垛机控制系统
CN109534177A (zh) 一种基于机器视觉的集卡防吊起装置及集卡防吊起方法
CN104683772A (zh) 港口集装箱全方位图形识别理货信息监测系统及其监测方法
CN114998824A (zh) 一种车辆装卸任务监控方法、装置及系统
CN105836673A (zh) 一种堆料机堆垛高度实时数据采集系统
CN112462373A (zh) 基于多传感器融合的船舱位检测方法
CN108792951A (zh) 图像识别技术的集卡防吊起方法、系统、设备及存储介质
CN107200274A (zh) 一种基于机器视觉的防集装箱卡车被吊起方法
CN110015617A (zh) 一种自动化集装箱门式起重机控制系统
CN111539344A (zh) 基于视频流及人工智能的集卡防吊起控制系统及方法
CN208802612U (zh) 装船机作业系统和装船机
WO2024104192A1 (zh) 基于点云的吊具二次锚定方法、系统、设备及存储介质
EP1594791B1 (en) A container crane and a method for container identification
CN204481965U (zh) 港口集装箱全方位图形识别理货信息监测系统
CN113415727A (zh) 一种集装箱装卸识别安全引导装置和方法
CN207957554U (zh) 自动化码头集卡防吊起系统
CN206389462U (zh) 集装箱卡车防吊起系统
CN204433483U (zh) 集装箱验残识别装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21905240

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM1205N DATED 08-08-2023)