CN111272764A - Large intelligent temporary platform non-contact image recognition mobile management and control system and method - Google Patents
Large intelligent temporary platform non-contact image recognition mobile management and control system and method Download PDFInfo
- Publication number
- CN111272764A CN111272764A CN202010074782.9A CN202010074782A CN111272764A CN 111272764 A CN111272764 A CN 111272764A CN 202010074782 A CN202010074782 A CN 202010074782A CN 111272764 A CN111272764 A CN 111272764A
- Authority
- CN
- China
- Prior art keywords
- node
- intelligent
- image
- nodes
- engineering
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8854—Grading and classifying of flaws
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8854—Grading and classifying of flaws
- G01N2021/8861—Determining coordinates of flaws
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a large intelligent temporary platform non-contact image recognition mobile management and control system and a method, which mainly comprise: 1) modeling, simulating, testing and contrastively analyzing a large temporary structure and engineering, determining the mapping relation between information to be acquired in the structure and engineering management and a model in a computer, and giving core parameters of an acquisition device; 2) intelligently collecting node images; 3) carrying out autonomous intelligent analysis on the acquired node images and giving an analysis result based on a model and a rule; 4) and intelligently generating and expressing results based on the Internet of things and cloud. The invention mainly adopts the non-contact mass data acquisition technology of the aircraft image, autonomously acquires the image pixels of the large temporary structure nodes and the connecting pieces, analyzes and processes the large data of the nodes and the engineering management image in real time, automatically generates a complete safety detection and control report of the temporary structure nodes, and provides the intelligent safety control and value creation of the large temporary structure.
Description
Technical Field
The invention relates to a large intelligent temporary platform non-contact image recognition mobile management and control system and a method.
Background
After the temporary structural rod piece is repeatedly transported, installed and used for a long time, certain initial defects are inevitably generated. In the field installation process, the installation quality is greatly influenced by the quality of workers and the technical level. In order to standardize the construction process of the temporary structure, ensure the construction quality of the temporary structure and discover the crack defects on the surface of the temporary structure node in time, efficient non-manual detection technology needs to be carried out on the surface cracks and the positions of main components of the temporary structure node.
The traditional upright post frame mainly utilizes a steel pipe as a main stress rod piece, and structural connection forms such as a straight buckle, a bowl buckle or a portal frame are formed through connecting nodes so as to support various loads in the construction process of a main body structure. The traditional inspection maintenance mode is that every node component is carried out the naked eye and is observed and combine handheld instrument to the manual work, discovers and the record problem, however to large-scale temporary structure, the node is numerous, and the higher, the darker check point in position is difficult to reach moreover, and the complete safety inspection efficiency of the traditional temporary structure of multinode is very low, and the accuracy degree is difficult to guarantee.
Disclosure of Invention
In order to solve the problems in the background art, the invention provides a large intelligent temporary platform non-contact image recognition mobile management and control method, which automatically processes large temporary structure safety check work which is difficult to realize manually by a mobile terminal, on one hand, comprehensive and reliable detection results can be provided, and the structure safety is powerfully maintained, and on the other hand, because the detection efficiency and the operation are simple and are beneficial to popularization and promotion, the large temporary multi-node structure detection system based on the non-contact image recognition mobile management and control technology has great significance and has important practical application value.
The technical scheme for solving the problems is as follows:
the non-contact image recognition, movement and control method for the large intelligent temporary platform is characterized in that a field non-reference non-contact image recognition method is taken as a main method, a deep learning model feature library formed by reference is combined, an anti-collision algorithm is used for controlling an image acquisition aircraft, high-speed image acquisition, intelligent analysis and cloud data processing are carried out on the large intelligent temporary structure in an all-node state, and a non-contact image movement and control system for the large intelligent temporary platform is obtained. It comprises the following steps:
1) modeling the structural engineering, determining the datamation of the structure and the information to be detected in engineering management, and finishing the mapping relation between the space coordinate and the space coordinate of the 3D model in the computer;
2) intelligently collecting node images;
3) analyzing the collected node images and giving detection and prediction results;
4) and intelligently generating and expressing the detection and prediction results.
Further, the step 1) is specifically:
1.1) customizing AUTOCAD plug-ins, importing design drawings of structure and management engineering, numbering style node components and management modes selected in the design drawings by using plug-in functions according to a certain numerical rule, and defining nodes as a class, wherein the class attribute comprises the number and relative spatial coordinates of the nodes;
1.2) then, obtaining the length and the position of the steel pipe by using a plug-in unit, and calculating the maximum space boundary which can fly by the intelligent collector theoretically;
1.3) finally saving all the information into a file with a specified format for the intelligent image acquisition system to use.
Further, the above step 2): the intelligent acquisition of the node image is realized by using an intelligent acquisition device to acquire the node image by pointing energy, and specifically comprises the following steps:
2.1) determining a flight control line of the intelligent collector in the 3D model, determining a mapping relation between a virtual route and an actual route, and controlling from 2 aspects of hardware speed and software functions;
2.2) guiding a third party designed by the intelligent collector chip to execute the wireless function and non-contact image collection characteristics required by the tasks of real-time collection and cloud analysis of a large number of node images of a large-scale structure by utilizing the terminal specific parameters;
and 2.3) transmitting the data acquired by the intelligent acquisition unit in real time to the cloud and IOTs-R terminals by selecting a specially designed wireless transmission enhancement program.
Further, the above step 3): the analyzing the collected node image specifically includes:
3.1) grading according to the node crack damage characteristics
Classifying each crack according to occurrence and development sources, wherein the classification principle is based on different crack-causing mechanisms, and grade definition is carried out according to the calculation of the stress level of fracture mechanics, unstable cracks and the propagation direction of sub-zero boundary cracks;
3.2) performing non-reference identification by using a crack deep learning model;
3.3) non-contact identification of node damage.
Further, the above step 4): the intelligent generation and expression of the detection and prediction results are specifically as follows:
according to the calculation of fracture mechanics, after the nodes are analyzed and searched to be abnormal, the abnormal nodes are recorded into a problem list based on a model according to classification and grading, expression information comprising the self space position of an intelligent collector, an engineering modeling system, a microkernel controller, an identification process, environmental parameters, abnormal node pixel coordinates, intelligent positioning node codes, an analysis information chart and cartoon parameters of structure safety evolution is comprehensively generated, expression models and corresponding space coordinates of different identification structures are distributed according to different requirements of class members, finally problem nodes in a large temporary platform are marked through an image editing plug-in, a problem list table is generated, dynamic three-dimensional (namely, time-varying node structure x, y and z variation values, namely, temporary structure deformation and dangerous four-dimensional graph) information of the nodes and structure damage is given, and maintenance suggestions are provided, and the given expression result is stored, applied and developed in the Internet of things and cloud.
In addition, the invention also provides a large intelligent temporary platform non-contact image recognition mobile management and control system based on the large intelligent temporary platform non-contact image recognition mobile management and control method, which comprises an engineering modeling system, an image acquisition system, an image analysis system and a result output system;
the engineering modeling system is used for modeling a structure and engineering management, determining coordinates and numerical information of nodes to be detected in the structure and an engineering service stage, and mapping space coordinates of a computer 3D model;
the image acquisition system is used for intelligently acquiring and returning the node images;
the image analysis system is used for carrying out intelligent IOTs-R processing on the collected node images, determining world coordinates of each node, obtaining detailed recognition results of the node crack images, judging displacement and deformation of the whole structure and other service states, judging later-stage evolution expansion paths of the node surface cracks by utilizing a deep learning technology, and giving detection and prediction results;
and the result output system is used for intelligently generating and expressing detection and prediction results.
The invention has the advantages that:
the invention relates to a large intelligent temporary platform non-contact image recognition mobile management and control system and a method, and provides a mobile management and control technology based on non-contact image recognition.
Drawings
FIG. 1 is a flow chart of a large intelligent temporary platform non-contact image recognition mobile management method of the present invention;
FIG. 2 is a schematic structural engineering diagram;
FIG. 3 is a schematic track diagram of an intelligent collector;
FIG. 4 is one of the node and node area identification areas;
FIG. 5 is a second node and node area identification area;
FIG. 6 is a third node and node area identification area;
FIG. 7 is a schematic diagram of a node containing an artifact;
FIG. 8 illustrates the area under test;
fig. 9 is a schematic view of a crack.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings of the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention. Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention.
A large intelligent temporary platform non-contact image recognition mobile management and control system comprises an engineering modeling system, an image acquisition system, an image analysis system and a result output system;
the engineering modeling system is used for modeling a structural engineering and determining the mapping relation between the world space coordinate of a node to be detected in the structure and the space coordinate of a 3D model in the computer;
the image acquisition system is used for intelligently acquiring and returning the node images;
the image analysis system is used for intelligently processing the acquired node images, determining the world coordinates of each node, obtaining the detailed identification result of the node crack image, judging the displacement and deformation of the whole structure and other service states, judging the later-stage evolution expansion path of the node surface crack by using a deep learning technology, and giving a detection and prediction result;
and the result output system is used for intelligently generating and expressing detection and prediction results.
The non-contact image recognition, movement and control method for the large intelligent temporary platform of the invention is described in detail below with reference to the accompanying drawings.
1) Engineering modeling system
The system is implemented and finished by 4 core technologies, and comprises the following specific patent technical processes:
1.1) Intelligent input of engineering parameters
The mobile management and control technology related to the invention detects and identifies nodes generally more than 500 nodes or node domains, mostly more than 1000 scales, but is not limited to small nodes and is only applied to large nodes, and the application efficiency of the technology is more prominent. Aiming at large-scale node air mobile acquisition and identification, the information of the nodes needs to be input into the embedded storage controller in advance, and three methods are adopted for carrying out: the method comprises the steps of automatically identifying CAD information of a temporary structure design drawing, finishing space initial positioning and size pre-storage of nodes by a CAD drawing identifier, then carrying out numerical processing on an environment diagram around the nodes to serve as an environment parameter for node quantification, adopting an environment enhancement algorithm, superposing the environment parameter on a CAD value of the node by using a feature code set by budget, and finally carrying out field environment calibration by using language input according to special conditions of a field, such as welding sparks generated when workers carry out installation and maintenance during collection, so as to provide efficient intelligent parameter quick input work for strong engineering parameter calibration of image collection.
1.2) engineering node simulation and emulation technique
The invention discloses a non-contact mobile image collector embedded with rapid node type calculation, comparison, simulation and simulation algorithm, which firstly carries out primary shaping calculation on the collected node space information and size, then carries out comparison calculation of the two parties according to the recognition result of the first collected image of the node and the node domain, eliminates the error of CAD primary input information, rapidly establishes the type of the node and the node domain, determines the real space information and size of the actual node and the node domain, and uses the real space information and size as the calibrated first reference data, then calls a simulation calculation program stored in an embedded controller in advance, calculates the boundary characteristics in the node area recognition area, extracts the characteristic recognition key points, then establishes the simulation image of the node and the node domain to be collected with the data of CAD or voice recognition and environment recognition, combines the collected actual image, after the environmental error is eliminated, the initial reference image is used as an initial reference for image acquisition and is also used as a basic reference image for subsequent deep learning, and the identification, boundary determination and feature calibration of the initial information of the engineering node can be completed through 1-3 times of calculation of the 2 types of basic information.
1.3) track setting and control
The basic process of the track setting is shown in fig. 2 and 3. Wherein, FIG. 2 shows the node space information established by data input in engineering modeling, each unit comprises 9 nodes, as shown by solid circles marked with numbers a-i in the figure, a 3D version construction design drawing of a temporary structure is imported through the customized special AUTOCAD plug-in, node members of a selected pattern in the design drawing are numbered according to a certain rule by using the plug-in function according to the engineering parameter input information, the node number, relative space coordinates, a node domain (i.e. steel pipes or other material line segments connected on the nodes) and environment information are taken as a class, the length and the position of the steel pipes in the node domain are obtained by using the plug-in, the maximum space boundary of the non-contact image acquisition equipment (unmanned aerial vehicle and the like) in flight is calculated, the flight trace as shown in FIG. 3 is formed, all the information is stored as files with a specified format and is stored in an embedded communication controller, the system is used for an intelligent image acquisition system.
After the work is finished, the mapping relation between the flight route and the virtual route and the actual route needs to be determined, and the determining method comprises the following steps: firstly, simulating the maximum flight boundary of a collector according to information generated by an engineering modeling system, utilizing an AUTOCAD (auto computer aided design) customized plug-in to identify a boundary area by a special mark, then utilizing key points to draw a flight path in the flight area, storing the spatial positions of the key points in a class form according to a set sequence, starting a customized coordinate conversion function, converting the spatial coordinates of the key points in a 3D (three-dimensional) model into the real-flight relative coordinates of the collector, and storing the coordinate sequence of the key points for later use.
The method for drawing the flight path of the image acquisition terminal and the shooting trigger in the modeling system controls the shooting action which is not understood according to different identifications, the method for drawing the flight path can be realized by manually inputting key point coordinates or automatically learning the information acquired by the initial image, the terminal autonomously finishes the operation, and then, the acquisition route which is reasonably pushed and sent and the special shooting action and step are autonomously judged according to the quality of the recording operation.
1.4) node image collector
The node image collector utilizes a mature moving image to collect a front end, mainly embeds a state control instruction during collection into a plug-in system by using C and assembly language according to an API (application programming interface) provided by equipment and according to a track control instruction, and realizes special repeated collection, space pause and other exclusive collection tasks in image collection.
2) Image acquisition system
The system is realized as follows:
2.1) post-impact posture recovery structure and software and hardware design
When a non-contact collector collects images around a node or a node domain, if collision occurs, the posture of the collector needs to be recovered to a reasonable space position within the shortest time, a flexible spring anti-collision structure is additionally arranged on the collector to prevent the collector from being damaged by collision, the distance of a reference object is calculated by using a millimeter wave radar according to information input by the node and the space position of the collided collector, the flying speed and the front and back elevation angles are adjusted according to collision dynamics, the rotating speed of a propeller is realized by software, and finally the posture of the collector is recovered.
2.2) open embedded communication control and intelligent mobile Internet of things cloud platform
Because the front-end image collector of non-contact mobile management and control is an image sensor, after a preprocessing and control system is embedded, the front-end image collector has data distribution and kernel-level multi-user multi-task computing functions, each image collector for executing tasks on track is an exchangeable intelligent terminal of the Internet of things, and accordingly, in order to complete synchronous work of a plurality of image collectors, the invention mainly adopts an NB-IOT platform, simultaneously adopts a LoRa technology in a small amount, adopts a fusion and coordination mechanism of hybrid design image collection and information high-speed production places, implants a related protocol and a sharing mechanism of an Internet of things cloud into each collector, and enables the image collector in a working state to become the Internet of things through an execution layer, thereby completing autonomous data processing, and simultaneously also has the functions of interconnection, independent data processing, remote special cloud and the Internet of things, The intelligent mobile internet of things cloud always corresponds to a determined sensor, executes node and node domain information based on identification, and is based on bandwidth and capacity of a wireless carrier on the premise of ensuring completion of safety control. The method can expand third-party APP with very important value, and is an important tool for expanding intelligent characteristics of large temporary structures and corresponding commercial deployment by using ultra-wide-band high-capacity transmission rates (such as 5G and the like) in the patent of the invention.
Collector bottom control system's terminal software development contains cell-phone terminal APP and acquires the control right of unmanned aerial vehicle hardware, instruct unmanned aerial vehicle to carry out the flight, the task of shooing, high stability execution is to the node image acquisition task of whole engineering, provide space flight orbit orbital transfer control and shoot trigger strategy, acquire the space flight orbit of renewal, wherein shoot trigger strategy and decided the collector at the flight in-process intelligence execution shooting command, trigger strategy formulation process does: firstly, if no trigger triggers, the aircraft collects key points according to the set traversal sequence by using the initially set parameters and rules, if the image analysis system captures good shooting opportunity and feeds back feedback signals in real time, the collector hovers for a specified time, executes linear and rotary shooting tasks, and judges a reference template for continuing starting after the judgment according to the quality of the collected images. Due to the influence of the environment, the control system is provided with an image acquisition auxiliary enhancement function, when the image real-time analysis system feeds back environmental parameters to be unfavorable, for example, the light is dark or residues are on the surface of a node or other large interferences are generated, the unmanned aerial vehicle performs light supplement or blows out strong wind to clean the surface by hovering, image acquisition is performed when the shooting time meets the requirements, and when a trigger of the image acquisition auxiliary enhancement device is not triggered, the aircraft traverses according to the sequence of the original key points.
2.3) microkernel independent fast cloud storage
The method comprises the steps of adopting a Loose Coupling Asymmetric (LCA) architecture technology, independently compiling initial information of a large number of node images, reference mark characteristics, flight paths and other big data by utilizing a core algorithm in Cloud Computing, a cluster technology, a distributed file system and grid Computing, converting a compiled program into a binary system, embedding a micro-kernel independent function chip, and finishing a special function micro-kernel chip by a specified provider so As to finish quick real-time Computing, data migration, multi-use, multi-user, multi-scene and other Cloud Elastic Computing (ECC-Elastic Computing) of the large number of nodes, and simultaneously integrating application Integration, MAI (M2 mapping Integration) application Integration, and realizing non-contact image data and identification results of the large number of nodes with a large temporary structure and multiple applications based on a SOA technology and model for subsequent processing of identification results and distributed expression, The intelligent cloud and internet of things Service functions such As Multi-user Service MaaS (M2M As A Service) and MMO (Multi-Tenans) Multi-user leasing are integrated in the microkernel and become a function-independent unit for the image collector, even if a remote server is powered off or wireless connection fails, the image collection and the identification of core characteristics can still be continuously executed, which is particularly important under special environments such As severe and military requirements and the like, and is one of important and unique technologies of the non-contact rapid identification method used in the initial development and evolution stage of small damages such As microcracks and the like in the invention.
3 image analysis system
The image analysis system is implemented as follows:
3.1) node crack Damage characterization and grading
The invention mainly takes nodes of a large temporary structure which are mainly manufactured by high-strength low-quality materials such as steel, magnesium-aluminum alloy and the like as main objects to carry out crack identification based on a non-contact mobile control image, wherein representative nodes are shown in figures 4-6, but the invention is not limited to the nodes. According to the statistics of a large number of tests and actual structure service conditions of the node, the identification of the crack damage of the node is mainly focused on two major types of plastic cracks and brittle cracks, wherein the brittle cracks are mainly convertible cracks, and for the cracks which are obvious in manufacturing, due to the safety inspection of products, therefore, cracks carried by unqualified products such as processing and initial welding cracks are not in the identification range, accordingly, four types of fine identification of intercrystalline cracks caused by stress concentration, eccentricity, explosion and the like and shearing bending, stretch bending, press bending and fatigue caused by partial crystal penetrating cracks after service are mainly carried out, in order to improve the identification speed and accuracy, each crack is graded according to the occurrence and development sources, the grading principle is based on the difference of the cracking mechanism, and carrying out grade definition according to the calculation of the stress level of fracture mechanics, unstable cracks and the propagation direction of the sub-zero boundary cracks.
3.2) crack deep learning model
In view of the complexity of the crack, the initial crack is a very dangerous crack due to the existence of a transition between brittle and ductile properties, and in view of the fact that not all nodes and node domains are simultaneously subjected to brittle crack failure, therefore, under the condition of supposing that all the recognizable deformation conditions exist, a neural network model for crack transformation, evolution, propagation and generation is established, in order to improve the mapping efficiency of the network, a deep learning model is adopted to classify cracks to finish the high efficiency and real-time feedback of crack feature extraction, the invention discloses an encoding method based on TensorFlow core codes, an SDNN (spiking Deepneural network) embedded type impulse neural network optimization calculation image recognition sample model is adopted, a novel momentum learning rule is adopted, and an STDP (Spike-timing Dependent reliability) training method is combined to greatly improve the learning speed of a random gradient descent algorithm.
Crack identification method based on no reference:
according to the identification technology, no special image characteristics are required to be arranged on a generation identification structure, filtering and denoising are firstly carried out according to feedback image pixel space information of a real-time return system, an RGB image is converted into a gray-scale image added with a learning model, then the position of a steel pipe in a visual field is determined by using a Hough straight line extraction function, and the intersection of the steel pipe (namely the area where nodes appear) is used as an ROI and is scratched for standby application. The patent algorithm uses a multi-redundancy fusion component boundary determination technology such as edge extraction, watershed and threshold segmentation (the color of a node component is effectively different from that of a steel pipe) to extract component core image pixels, then uses supervised deep learning algorithm, learns compared with normal node images on the basis of using a large number of damaged nodes to count image information in advance, determines whether the component is healthy and iterates state parameters through memory feedback, and synchronously learns the occurrence condition of obstacles in a time-sharing manner so as to cooperate with the follow-up calculation and analysis of an image acquisition auxiliary enhancement trigger and the real-time performance of process autonomous operation judgment. For the judgment function of the shooting time, a visual field image segmentation function is set in the algorithm, model items such as illumination, inclination angles, visual field ranges and the like are extracted and given a score value, and when the score value exceeds a certain threshold value, a shooting signal is fed back.
The prior identification characteristic information in crack deep learning is a high-speed non-contact image identification method which is formed by modeling according to a statistical characteristic library established based on an optimal error principle according to a component crack test and a structural actual crack evolution engineering experience, an acquired image comprises special information of a component to be detected and also comprises redundant background information, in order to accurately search a crack search range, a first stage firstly uses artificial marks for assistance in identification to distinguish a region to be detected from the redundant region, and the first stage comprises three parts: (1) priori knowledge; (2) solving the relation between the pixel coordinate and the world coordinate; (3) and (6) positioning. The priori knowledge here refers to the mapping relationship between the artificial auxiliary mark and the region to be searched (as shown in fig. 7, the engineering information input function unit automatically draws the core region of the whole member of the identification node from the graph paper according to the position of the specific mark point), a stable mapping relationship is established between the pixel coordinate in the image and the world coordinate corresponding to the point, and the region to be searched is accurately positioned according to the rule.
Writing an image recognition algorithm to recognize pixel coordinates of the artificial mark, as shown in fig. 8, a paper sheet with a white black bottom is a node coverage mark given by artificial simulation calculation, obtaining a world coordinate of the node coverage mark through mapping of the pixel coordinates and the world coordinate, and locking the world coordinate of the region to be detected through self error iterative calculation by depending on the established mapping relation between the mark and the region to be detected. Because the pixels have discreteness, the number of key points is required to be given according to specific conditions such as identification precision and service life environmental characteristics of the outline of the area to be detected, and the identification boundary is automatically determined.
After the above work is completed, the world coordinates of the key points of the region to be detected are converted into pixel coordinates through mapping, and the pixel coordinates of the key points are sequentially connected to obtain a connected domain, such as a crack boundary shown in fig. 9, wherein the recognized crack state is enlarged for convenience of expression.
The method comprises the steps of searching crack characteristics in a node to be detected and a node area, according to classification, classification and relevant definition of cracks and vector characteristics of the cracks in an image, defining the cracks with first-stage identification depth as 'the depth value of a node crack pixel is low and has strong directivity', if the two characteristics are provided at the same time, considering the first-stage depth as the cracks, accordingly introducing the cracks into a deep learning algorithm, compiling an identification crack depth function, and achieving the purpose of detecting other cracks in a target area through training and learning.
3.3) node Damage non-contact identification
Node damage identification is a determination of the node status of a large temporary structure based on crack identification information. Considering the effectiveness of damage identification, and the number of nodes of a large temporary structure is large, according to a structure collapse mechanism, 2 methods are adopted for carrying out damage quantitative evaluation: the first method is that the extracted node crack state is identified, the node fracture life is calculated, the stress resistance level of the node with the expansion crack is obtained, and the quantitative conclusion of whether the node has continuous bearing or can bear the load for a long time is judged according to the load distributed by the node structure, so that the related follow-up measures such as the damage state and the maintenance strategy of the node are given; the second is that according to the deformation condition of the node domain, because the node domain has or has such a trend, the node can not continuously complete the established function due to the possible excessive deformation, so as to provide the damage state caused by the large deformation of the substructure or the local structure formed by the node domain, and provide the corresponding judgment information and subsequent processing measures.
3.4) node crack and safety control expression technology
The safety control and expression of the node cracks are important expression forms of final achievements and application of the technology, and are also important interfaces and tools for recalculating, analyzing and information transferring of the identification structure on the Internet of things and the cloud, the technology of the Internet of things and the cloud is specially developed on the management and control of the node crack information and safety in order to realize better application of the non-contact mobile control technology of the large intelligent temporary platform nodes, the microkernel independent control technology, the embedded control communication and the intelligent cloud platform of the Internet of things in the technology of the invention independently develop and apply the acquired and identified nodes and the image information in the node domain, and the key point of the implementation of the working technology at this stage is that the model-based Internet of things storage and the special cloud multifunctional computing development and model opening system are carried out on the basis of the data of the acquired and identified results, all temporary structure platforms can be connected into the expression system without being limited to nodes of a rigid structure, so that a large data platform is established for the huge amount of node information of a large temporary structure, a basis is further provided for establishing data with excellent quality for a wider deep learning model, a network distributed database technology is adopted for the crack and safety expression, programming is carried out by Paython, the crack is used as a basic class and is used as an attribute and a function of a safety class, standardized design is carried out, the crack and safety are established on an Ubuntu OS system, a crack and safety welcome relationship is established according to a multi-sensor fusion mode of class aircraft information acquisition, a graph relation expression node, a node domain, a substructure and a structure 4-level safety comprehensive transfer mode is established, and safety control information which can be understood by experts, common personnel and even disabled persons is rapidly, accurately expressed, the intelligent management and control technology really achieves information distribution according to needs.
Referring to fig. 1, a non-contact image recognition, movement and control method for a large intelligent temporary platform comprises the following steps:
the method comprises the following steps:
1) modeling the structural engineering, determining the datamation of the structure and the information to be detected in engineering management, and finishing the mapping relation between the space coordinate and the space coordinate of the 3D model in the computer;
2) intelligently collecting node images;
3) analyzing the collected node images and giving detection and prediction results;
4) and intelligently generating and expressing the detection and prediction results.
Further, the step 1) is specifically:
1.1) customizing AUTOCAD plug-ins, importing design drawings of structure and management engineering, numbering style node components and management modes selected in the design drawings by using plug-in functions according to a certain numerical rule, and defining nodes as a class, wherein the class attribute comprises the number and relative spatial coordinates of the nodes;
1.2) then, obtaining the length and the position of the steel pipe by using a plug-in unit, and calculating the maximum space boundary which can fly by the intelligent collector theoretically;
1.3) finally saving all the information into a file with a specified format for the intelligent image acquisition system to use.
Further, the above step 2): the intelligent acquisition of the node image is realized by using an intelligent acquisition device to acquire the node image by pointing energy, and specifically comprises the following steps:
2.1) determining a flight control line of the intelligent collector in the 3D model, determining a mapping relation between a virtual route and an actual route, and controlling from 2 aspects of hardware speed and software functions;
2.2) guiding a third party designed by the intelligent collector chip to execute the wireless function and non-contact image collection characteristics required by the tasks of real-time collection and cloud analysis of a large number of node images of a large-scale structure by utilizing the terminal specific parameters;
and 2.3) transmitting the data acquired by the intelligent acquisition unit in real time to the cloud and IOTs-R terminals by selecting a specially designed wireless transmission enhancement program.
Further, the above step 3): the analyzing the collected node image specifically includes:
3.1) grading according to the node crack damage characteristics
Classifying each crack according to occurrence and development sources, wherein the classification principle is based on different crack-causing mechanisms, and grade definition is carried out according to the calculation of the stress level of fracture mechanics, unstable cracks and the propagation direction of sub-zero boundary cracks;
3.2) performing non-reference identification by using a crack deep learning model;
3.3) non-contact identification of node damage.
Further, the above step 4): the intelligent generation and expression of the detection and prediction results are specifically as follows:
according to the calculation of fracture mechanics, after the nodes are analyzed and searched to be abnormal, the abnormal nodes are recorded into a problem list based on a model according to classification and grading, expression information comprising the self space position of an intelligent collector, an engineering modeling system, a microkernel controller, an identification process, environmental parameters, abnormal node pixel coordinates, intelligent positioning node codes, an analysis information chart and cartoon parameters of structure safety evolution is comprehensively generated, expression models and corresponding space coordinates of different identification structures are distributed according to different requirements of class members, finally problem nodes in a large temporary platform are marked through an image editing plug-in, a problem list table is generated, dynamic three-dimensional (namely, time-varying node structure x, y and z variation values, namely, temporary structure deformation and dangerous four-dimensional graph) information of the nodes and structure damage is given, and maintenance suggestions are provided, and the given expression result is stored, applied and developed in the Internet of things and cloud.
In addition, the invention also provides a large intelligent temporary platform non-contact image recognition mobile management and control system based on the large intelligent temporary platform non-contact image recognition mobile management and control method, which comprises an engineering modeling system, an image acquisition system, an image analysis system and a result output system;
the engineering modeling system is used for modeling a structure and engineering management, determining coordinates and numerical information of nodes to be detected in the structure and an engineering service stage, and mapping space coordinates of a computer 3D model;
the image acquisition system is used for intelligently acquiring and returning the node images;
the image analysis system is used for carrying out intelligent IOTs-R processing on the collected node images, determining world coordinates of each node, obtaining detailed recognition results of the node crack images, judging displacement and deformation of the whole structure and other service states, judging later-stage evolution expansion paths of the node surface cracks by utilizing a deep learning technology, and giving detection and prediction results;
and the result output system is used for intelligently generating and expressing detection and prediction results.
The above description is only an embodiment of the present invention, and not intended to limit the scope of the present invention, and all equivalent structures or equivalent flow transformations made by using the contents of the specification and the drawings, or applied directly or indirectly to other related systems, are included in the scope of the present invention.
Claims (6)
1. A non-contact image recognition, movement and management method for a large intelligent temporary platform is characterized by comprising the following steps:
1) modeling the structural engineering, determining the datamation of the structure and the information to be detected in engineering management, and finishing the mapping relation between the space coordinate and the space coordinate of the 3D model in the computer;
2) intelligently collecting node images;
3) analyzing the collected node images and giving detection and prediction results;
4) and intelligently generating and expressing the detection and prediction results.
2. The large intelligent temporary platform non-contact image recognition, movement and management method as claimed in claim 1, characterized in that:
the step 1) is specifically as follows:
1.1) customizing AUTOCAD plug-ins, importing design drawings of structure and management engineering, numbering style node components and management modes selected in the design drawings by using plug-in functions according to a certain numerical rule, and defining nodes as a class, wherein the class attribute comprises the number and relative spatial coordinates of the nodes;
1.2) then, obtaining the length and the position of the steel pipe by using a plug-in unit, and calculating the maximum space boundary which can fly by the intelligent collector theoretically;
1.3) finally saving all the information into a file with a specified format for the intelligent image acquisition system to use.
3. The large intelligent temporary platform non-contact image recognition, movement and management method as claimed in claim 1, characterized in that:
step 2): the intelligent acquisition of the node image is realized by using an intelligent acquisition device to acquire the node image by pointing energy, and specifically comprises the following steps:
2.1) determining a flight control line of the intelligent collector in the 3D model, determining a mapping relation between a virtual route and an actual route, and controlling from 2 aspects of hardware speed and software functions;
2.2) guiding a third party designed by the intelligent collector chip to execute the wireless function and non-contact image collection characteristics required by the tasks of real-time collection and cloud analysis of a large number of node images of a large-scale structure by utilizing the terminal specific parameters;
and 2.3) transmitting the data acquired by the intelligent acquisition unit in real time to the cloud and IOTs-R terminals by selecting a specially designed wireless transmission enhancement program.
4. The large intelligent temporary platform non-contact image recognition, movement and management method as claimed in claim 1, characterized in that:
step 3): the analyzing the collected node image specifically includes:
3.1) grading according to the node crack damage characteristics
Classifying each crack according to occurrence and development sources, wherein the classification principle is based on different crack-causing mechanisms, and grade definition is carried out according to the calculation of the stress level of fracture mechanics, unstable cracks and the propagation direction of sub-zero boundary cracks;
3.2) performing non-reference identification by using a crack deep learning model;
3.3) non-contact identification of node damage.
5. The large intelligent temporary platform non-contact image recognition, movement and management method as claimed in claim 1, characterized in that:
step 4): the intelligent generation and expression of the detection and prediction results are specifically as follows:
according to the calculation of fracture mechanics, after the nodes are analyzed and searched to be abnormal, the abnormal nodes are recorded into a problem list based on a model according to classification and grading, expression information comprising the self space position of an intelligent collector, an engineering modeling system, a microkernel controller, an identification process, environmental parameters, abnormal node pixel coordinates, intelligent positioning node codes, an analysis information chart and cartoon parameters of structure safety evolution is comprehensively generated, expression models and corresponding space coordinates of different identification structures are distributed according to different requirements of class members, finally problem nodes in a large temporary platform are marked through an image editing plug-in, a problem list table is generated, dynamic three-dimensional (namely, time-varying node structure x, y and z variation values, namely, temporary structure deformation and dangerous four-dimensional graph) information of the nodes and structure damage is given, and maintenance suggestions are provided, and the given expression result is stored, applied and developed in the Internet of things and cloud.
6. The utility model provides a management and control system is removed in non-contact image recognition of large-scale interim platform which characterized in that:
the system comprises an engineering modeling system, an image acquisition system, an image analysis system and a result output system;
the engineering modeling system is used for modeling a structure and engineering management, determining coordinates and numerical information of nodes to be detected in the structure and an engineering service stage, and mapping space coordinates of a computer 3D model;
the image acquisition system is used for intelligently acquiring and returning the node images;
the image analysis system is used for carrying out intelligent IOTs-R processing on the collected node images, determining world coordinates of each node, obtaining detailed recognition results of the node crack images, judging displacement and deformation of the whole structure and other service states, judging later-stage evolution expansion paths of the node surface cracks by utilizing a deep learning technology, and giving detection and prediction results;
and the result output system is used for intelligently generating and expressing detection and prediction results.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010074782.9A CN111272764B (en) | 2020-01-22 | 2020-01-22 | Non-contact image identification mobile management and control system and method for large intelligent temporary platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010074782.9A CN111272764B (en) | 2020-01-22 | 2020-01-22 | Non-contact image identification mobile management and control system and method for large intelligent temporary platform |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111272764A true CN111272764A (en) | 2020-06-12 |
| CN111272764B CN111272764B (en) | 2023-04-28 |
Family
ID=70995375
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010074782.9A Active CN111272764B (en) | 2020-01-22 | 2020-01-22 | Non-contact image identification mobile management and control system and method for large intelligent temporary platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111272764B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112130118A (en) * | 2020-08-19 | 2020-12-25 | 复旦大学无锡研究院 | SNN-based ultra-wideband radar signal processing system and processing method |
| CN112304852A (en) * | 2020-10-20 | 2021-02-02 | 中国特种设备检测研究院 | Portable pressure-bearing equipment damage mode intelligent identification method and system |
Citations (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6075880A (en) * | 1994-03-16 | 2000-06-13 | Jenoptik Technologie Gmbh | Method for detection of defects in the inspection of structured surfaces |
| US6584420B1 (en) * | 1999-11-29 | 2003-06-24 | Olympus Optical Co., Ltd. | Defect examination apparatus |
| CN1566907A (en) * | 2003-07-09 | 2005-01-19 | 何林 | Time base varying monitoring method for large-scale construction damage status real time identification |
| JP2007198762A (en) * | 2006-01-24 | 2007-08-09 | Canon Chemicals Inc | Flaw detection method and detector |
| US20070265790A1 (en) * | 2006-05-09 | 2007-11-15 | Lockheed Martin Corporation | System to monitor the health of a structure, program product and related methods |
| CN201251554Y (en) * | 2008-07-08 | 2009-06-03 | 武汉理工大学 | Deepwater pier detection device |
| CN101532926A (en) * | 2008-12-12 | 2009-09-16 | 齐齐哈尔华工机床制造有限公司 | On-line vision detecting system for automatic impact specimen processing device and image processing method thereof |
| JP2011095178A (en) * | 2009-10-30 | 2011-05-12 | Itochu Techno-Solutions Corp | Method and program for analyzing crack development |
| CN104678954A (en) * | 2015-01-23 | 2015-06-03 | 中国长江三峡集团公司 | Dam safety intelligent monitoring and pre-warning system based on full life circle and method thereof |
| US20170372470A1 (en) * | 2016-06-27 | 2017-12-28 | Sun Yat-Sen University | Method of separating, identifying and characterizing cracks in 3d space |
| US20180106609A1 (en) * | 2015-03-20 | 2018-04-19 | Nec Corporation | Structure status determination device, status determination system, and status determination method |
| CN108593656A (en) * | 2018-04-17 | 2018-09-28 | 中国公路工程咨询集团有限公司 | A kind of structure detection method, device and the UAV system for structure detection |
| US20180292328A1 (en) * | 2015-12-25 | 2018-10-11 | Fujifilm Corporation | Information processing device and information processing method |
| CN108650245A (en) * | 2018-04-24 | 2018-10-12 | 上海奥孛睿斯科技有限公司 | Internet of things system based on augmented reality and operation method |
| CN109060281A (en) * | 2018-09-18 | 2018-12-21 | 山东理工大学 | Integrated Detection System for Bridge based on unmanned plane |
| CN109300126A (en) * | 2018-09-21 | 2019-02-01 | 重庆建工集团股份有限公司 | A kind of bridge defect high-precision intelligent detection method based on spatial position |
| CN109580649A (en) * | 2018-12-18 | 2019-04-05 | 清华大学 | A kind of identification of engineering structure surface crack and projection modification method and system |
| CN109580657A (en) * | 2019-01-23 | 2019-04-05 | 郑州工程技术学院 | A kind of crack detection method in bridge quality testing |
| CN109816626A (en) * | 2018-12-13 | 2019-05-28 | 深圳高速工程检测有限公司 | Road surface crack detection method, device, computer equipment and storage medium |
| CN109901625A (en) * | 2019-04-11 | 2019-06-18 | 株洲时代电子技术有限公司 | A kind of bridge cruising inspection system |
| CN110031477A (en) * | 2019-04-04 | 2019-07-19 | 中设设计集团股份有限公司 | Bridge key component disease early warning system and method based on image monitoring data |
| CN110059631A (en) * | 2019-04-19 | 2019-07-26 | 中铁第一勘察设计院集团有限公司 | The contactless monitoring defect identification method of contact net |
| WO2019163329A1 (en) * | 2018-02-21 | 2019-08-29 | 富士フイルム株式会社 | Image processing device and image processing method |
| JP2019200512A (en) * | 2018-05-15 | 2019-11-21 | 株式会社日立システムズ | Structure deterioration detection system |
| CN110532591A (en) * | 2019-07-12 | 2019-12-03 | 中南大学 | Method based on DIC-EFG associative simulation analysis crack tip strain field |
| KR20190142626A (en) * | 2018-06-18 | 2019-12-27 | 세종대학교산학협력단 | System and method for autonomous crack evaluation of structure using hybrid image scanning |
-
2020
- 2020-01-22 CN CN202010074782.9A patent/CN111272764B/en active Active
Patent Citations (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6075880A (en) * | 1994-03-16 | 2000-06-13 | Jenoptik Technologie Gmbh | Method for detection of defects in the inspection of structured surfaces |
| US6584420B1 (en) * | 1999-11-29 | 2003-06-24 | Olympus Optical Co., Ltd. | Defect examination apparatus |
| CN1566907A (en) * | 2003-07-09 | 2005-01-19 | 何林 | Time base varying monitoring method for large-scale construction damage status real time identification |
| JP2007198762A (en) * | 2006-01-24 | 2007-08-09 | Canon Chemicals Inc | Flaw detection method and detector |
| US20070265790A1 (en) * | 2006-05-09 | 2007-11-15 | Lockheed Martin Corporation | System to monitor the health of a structure, program product and related methods |
| CN201251554Y (en) * | 2008-07-08 | 2009-06-03 | 武汉理工大学 | Deepwater pier detection device |
| CN101532926A (en) * | 2008-12-12 | 2009-09-16 | 齐齐哈尔华工机床制造有限公司 | On-line vision detecting system for automatic impact specimen processing device and image processing method thereof |
| JP2011095178A (en) * | 2009-10-30 | 2011-05-12 | Itochu Techno-Solutions Corp | Method and program for analyzing crack development |
| CN104678954A (en) * | 2015-01-23 | 2015-06-03 | 中国长江三峡集团公司 | Dam safety intelligent monitoring and pre-warning system based on full life circle and method thereof |
| US20180106609A1 (en) * | 2015-03-20 | 2018-04-19 | Nec Corporation | Structure status determination device, status determination system, and status determination method |
| US20180292328A1 (en) * | 2015-12-25 | 2018-10-11 | Fujifilm Corporation | Information processing device and information processing method |
| US20170372470A1 (en) * | 2016-06-27 | 2017-12-28 | Sun Yat-Sen University | Method of separating, identifying and characterizing cracks in 3d space |
| WO2019163329A1 (en) * | 2018-02-21 | 2019-08-29 | 富士フイルム株式会社 | Image processing device and image processing method |
| CN108593656A (en) * | 2018-04-17 | 2018-09-28 | 中国公路工程咨询集团有限公司 | A kind of structure detection method, device and the UAV system for structure detection |
| CN108650245A (en) * | 2018-04-24 | 2018-10-12 | 上海奥孛睿斯科技有限公司 | Internet of things system based on augmented reality and operation method |
| JP2019200512A (en) * | 2018-05-15 | 2019-11-21 | 株式会社日立システムズ | Structure deterioration detection system |
| KR20190142626A (en) * | 2018-06-18 | 2019-12-27 | 세종대학교산학협력단 | System and method for autonomous crack evaluation of structure using hybrid image scanning |
| CN109060281A (en) * | 2018-09-18 | 2018-12-21 | 山东理工大学 | Integrated Detection System for Bridge based on unmanned plane |
| CN109300126A (en) * | 2018-09-21 | 2019-02-01 | 重庆建工集团股份有限公司 | A kind of bridge defect high-precision intelligent detection method based on spatial position |
| CN109816626A (en) * | 2018-12-13 | 2019-05-28 | 深圳高速工程检测有限公司 | Road surface crack detection method, device, computer equipment and storage medium |
| CN109580649A (en) * | 2018-12-18 | 2019-04-05 | 清华大学 | A kind of identification of engineering structure surface crack and projection modification method and system |
| CN109580657A (en) * | 2019-01-23 | 2019-04-05 | 郑州工程技术学院 | A kind of crack detection method in bridge quality testing |
| CN110031477A (en) * | 2019-04-04 | 2019-07-19 | 中设设计集团股份有限公司 | Bridge key component disease early warning system and method based on image monitoring data |
| CN109901625A (en) * | 2019-04-11 | 2019-06-18 | 株洲时代电子技术有限公司 | A kind of bridge cruising inspection system |
| CN110059631A (en) * | 2019-04-19 | 2019-07-26 | 中铁第一勘察设计院集团有限公司 | The contactless monitoring defect identification method of contact net |
| CN110532591A (en) * | 2019-07-12 | 2019-12-03 | 中南大学 | Method based on DIC-EFG associative simulation analysis crack tip strain field |
Non-Patent Citations (3)
| Title |
|---|
| JIN HUI ET,AL: "Permissible Casting Defects Calculation of Cast Steel Nodes Based On Fracture Mechanics" * |
| SASABE,K: "Effect of Joint Clearance on Fatigue-Strength of Brazed Joint" * |
| 张维峰;尹冠生;刘萌;贺正权;: "数字图像处理技术在桥梁裂纹测量中的应用" * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112130118A (en) * | 2020-08-19 | 2020-12-25 | 复旦大学无锡研究院 | SNN-based ultra-wideband radar signal processing system and processing method |
| CN112130118B (en) * | 2020-08-19 | 2023-11-17 | 复旦大学无锡研究院 | Ultra-wideband radar signal processing system and method based on SNN |
| CN112304852A (en) * | 2020-10-20 | 2021-02-02 | 中国特种设备检测研究院 | Portable pressure-bearing equipment damage mode intelligent identification method and system |
| CN112304852B (en) * | 2020-10-20 | 2021-10-29 | 中国特种设备检测研究院 | Portable pressure-bearing equipment damage mode intelligent identification method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111272764B (en) | 2023-04-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110796186A (en) | Dry and wet garbage identification and classification method based on improved YOLOv3 network | |
| Khan et al. | Comparative study of hand gesture recognition system | |
| CN107316058A (en) | Improve the method for target detection performance by improving target classification and positional accuracy | |
| CN110222767B (en) | Three-dimensional point cloud classification method based on nested neural network and grid map | |
| CN109159113B (en) | Robot operation method based on visual reasoning | |
| CN114782626B (en) | Transformer substation scene map building and positioning optimization method based on laser and vision fusion | |
| CN108846359A (en) | It is a kind of to divide the gesture identification method blended with machine learning algorithm and its application based on skin-coloured regions | |
| CN111272764B (en) | Non-contact image identification mobile management and control system and method for large intelligent temporary platform | |
| CN103106265A (en) | Method and system of classifying similar images | |
| CN104036550A (en) | Laser radar point-cloud interpretation and reconstruction method for building elevations on basis of shape semantics | |
| CN114387499A (en) | Island coastal wetland waterfowl identification method, distribution query system and medium | |
| CN110135237B (en) | Gesture recognition method | |
| Zelener et al. | Cnn-based object segmentation in urban lidar with missing points | |
| CN111539422A (en) | Flight target cooperative identification method based on fast RCNN | |
| CN112052736A (en) | Cloud computing platform-based field tea tender shoot detection method | |
| Alidoost et al. | Y-shaped convolutional neural network for 3d roof elements extraction to reconstruct building models from a single aerial image | |
| CN113705445A (en) | Human body posture recognition method and device based on event camera | |
| Chen et al. | Research and implementation of robot path planning based on computer image recognition technology | |
| CN113971775B (en) | Method and system for identifying illegal behaviors based on optimization yolov algorithm | |
| CN111008607A (en) | Automatic laser scanning method and system for cabin door gap of visual servo aircraft | |
| Wang et al. | Building3d: A urban-scale dataset and benchmarks for learning roof structures from point clouds | |
| Xu et al. | Solar panel recognition of non-cooperative spacecraft based on deep learnin | |
| Mo et al. | Identification and detection of automotive door panel solder joints based on YOLO | |
| CN116690988A (en) | 3D printing system and method for large building model | |
| Zhongyang et al. | Classification of LiDAR point cloud based on multiscale features and pointnet |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |