CN112767388B - Wood microspur characteristic image acquisition and AI identification system - Google Patents
Wood microspur characteristic image acquisition and AI identification system Download PDFInfo
- Publication number
- CN112767388B CN112767388B CN202110134276.9A CN202110134276A CN112767388B CN 112767388 B CN112767388 B CN 112767388B CN 202110134276 A CN202110134276 A CN 202110134276A CN 112767388 B CN112767388 B CN 112767388B
- Authority
- CN
- China
- Prior art keywords
- wood
- image
- image acquisition
- user
- micro
- 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.)
- Active
Links
- 239000002023 wood Substances 0.000 title claims abstract description 222
- 238000012545 processing Methods 0.000 claims abstract description 27
- 238000000034 method Methods 0.000 claims abstract description 25
- 238000003825 pressing Methods 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 10
- 238000007781 pre-processing Methods 0.000 claims description 8
- 238000012937 correction Methods 0.000 claims description 7
- 230000000903 blocking effect Effects 0.000 claims description 6
- 238000013527 convolutional neural network Methods 0.000 claims description 5
- 230000004044 response Effects 0.000 claims description 5
- 239000006002 Pepper Substances 0.000 claims description 3
- 238000012797 qualification Methods 0.000 claims 2
- 241000894007 species Species 0.000 description 29
- 238000007726 management method Methods 0.000 description 14
- 230000006870 function Effects 0.000 description 12
- 238000011835 investigation Methods 0.000 description 12
- 238000010586 diagram Methods 0.000 description 11
- 230000008569 process Effects 0.000 description 7
- 238000011161 development Methods 0.000 description 6
- 238000012986 modification Methods 0.000 description 6
- 230000004048 modification Effects 0.000 description 6
- 238000004891 communication Methods 0.000 description 5
- 239000013589 supplement Substances 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 238000007619 statistical method Methods 0.000 description 4
- 230000001502 supplementing effect Effects 0.000 description 4
- 238000012217 deletion Methods 0.000 description 3
- 230000037430 deletion Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 241000238631 Hexapoda Species 0.000 description 2
- WHXSMMKQMYFTQS-UHFFFAOYSA-N Lithium Chemical compound [Li] WHXSMMKQMYFTQS-UHFFFAOYSA-N 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 238000013079 data visualisation Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 229910052744 lithium Inorganic materials 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 102100040401 DNA topoisomerase 3-alpha Human genes 0.000 description 1
- 101000611068 Homo sapiens DNA topoisomerase 3-alpha Proteins 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000035764 nutrition Effects 0.000 description 1
- 235000016709 nutrition Nutrition 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 238000004171 remote diagnosis Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/30—Authentication, i.e. establishing the identity or authorisation of security principals
- G06F21/44—Program or device authentication
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4007—Interpolation-based scaling, e.g. bilinear interpolation
-
- G06T5/80—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30161—Wood; Lumber
Abstract
The invention discloses a wood micro-distance characteristic image acquisition and AI identification system, which comprises a client, a database, a server and a wood image AI identification module, wherein the client is used for receiving user operation, the database is used for storing user data, the server comprises a security authentication module and a service processing module, the wood image AI identification module is used for identifying a wood image to be identified to obtain wood species information of wood, the client receives a login request and sends the login request to the server, and after the server receives the login request, the security authentication module verifies whether the login information of a user is correct or not by comparing the user data in the user database, and the service processing module identifies the wood image through the wood image AI identification module and feeds the identification result back to the client. The invention provides a method for acquiring a clearer wood image through the micro-distance image acquisition terminal, which improves the identification accuracy, and simultaneously simplifies the system workload by adopting the load equalizer, and has simple operation and convenient use.
Description
Technical Field
The invention relates to the field of wood identification, in particular to a wood microspur characteristic image acquisition and AI identification system.
Background
At present, the wood species identification is mainly carried out by professionals under laboratory environment, and the macroscopic and microscopic characteristics of the wood are observed by means of instruments and equipment such as a magnifying glass, a microscope and the like, and then the wood species is determined after the comparison with a standard sample. Each wood forms a unique macroscopic and microscopic structure in the growth process, but the wood has a plurality of types, and even the same wood has a certain difference in macroscopic and microscopic structure due to the conditions of places, climate, nutrition and the like. The phenomena of similarity and intra-species difference increase the difficulty of wood species identification, and in addition, the expert with wood taxonomy knowledge in China's wood trade and supervision line is rare, wood or wood products are frequently transacted, and the artificial-based wood species identification method has the problems of strong specialization, heavy task, long period, high risk, non-real-time property and the like, and cannot meet the requirements of real-time property and high efficiency of wood supervision, so that a rapid and accurate wood species identification method is needed.
Disclosure of Invention
In view of the above, it is necessary to provide a wood micro-distance feature image acquisition and AI identification system, which can accurately and rapidly complete wood identification work, and the technical scheme provided by the invention is as follows:
The invention provides a wood micro-distance characteristic image acquisition and AI identification system, which comprises:
the client is used for receiving user operation;
a database comprising a user database for storing user data;
the server comprises a security authentication module and a service processing module;
the wood image AI identification module is used for identifying the wood image to be identified to obtain wood species information of the wood;
the client and the database are respectively connected with the server, the client receives the login request and sends the login request to the server, after the server receives the login request, the security authentication module verifies whether the login information of the user is correct by comparing the user data in the user database,
if not, the security authentication module refuses the login request and sends a re-login instruction to the client;
if the wood image to be identified is correct, the service processing module can receive an image identification request sent by the client, wherein the image identification request comprises the wood image to be identified;
and the business processing module recognizes the wood image through the wood image AI recognition module and feeds back a recognition result to the client.
Further, the system also comprises a wood database, and the service processing module retrieves the wood database according to the recognition result returned by the wood image AI recognition module and feeds back the retrieval result to the client.
Further, the system also comprises a macro image acquisition terminal which can be arranged on a camera lens of the mobile terminal, so that the camera lens of the mobile terminal shoots a detail image of the wood to be identified.
Preferably, the macro image acquisition terminal is an assembled macro image acquisition terminal, the assembled macro image acquisition terminal comprises a macro lens, an adapter arranged at the outer end part of the macro lens and an acquisition head arranged at the outer end part of the adapter, wherein,
the adapter is internally provided with a perspective cavity for the perspective of the macro lens, the acquisition head comprises an acquisition base provided with an annular cavity communicated with the perspective cavity and a lamp arranged in the acquisition base, wherein the acquisition base, the adapter and the surface of a sample to be detected form a closed image acquisition area, or the acquisition base, the adapter and the surface of a platform for placing the sample to be detected form a closed image acquisition area.
Preferably, the macro image acquisition terminal is an embedded macro image acquisition terminal, the embedded macro image acquisition terminal comprises a macro lens, the macro lens comprises a lens holder and a lens, the lens holder comprises a holder body, a pressing ring detachably arranged at the front end part of the holder body and a lens external connection part arranged at the rear end part of the holder body, wherein the pressing ring is annular, and an inner cavity of the holder body, an inner cavity of the pressing ring and the surface of a sample to be detected form a closed image acquisition area, or the inner cavity of the holder body, the inner cavity of the pressing ring and a platform surface for placing the sample to be detected form a closed image acquisition area;
the embedded micro-distance image acquisition terminal further comprises a lamp which is arranged on the inner wall of the seat body or the inner wall of the pressing ring and used for compensating exposure, the lamp is annular, and the center line of the lamp and the center line of the lens are overlapped.
Further, the wood image AI-recognition module recognizing an image includes the steps of:
s1, inputting a wood image and preprocessing;
s2, processing the wood image in a blocking way;
s3, training the segmented sub-images through a convolutional neural network model;
S4, adopting different gradient values from the edge to the center of the wood image as weights of sub-image classification scores of different areas, increasing the proportion of the center area in the whole wood image classification score, converting the weighted score into a final probability value, obtaining the wood species information of the wood image, wherein,
the step of preprocessing in S1 includes:
s11, performing color correction on the wood image;
s12, carrying out data enhancement on the corrected wood image;
the step of processing the wood image in the S2 includes:
s21, dividing the preprocessed wood image to obtain a plurality of sub-images;
s22, unifying size pixels of each sub-image through a bilinear interpolation method.
Further, step S11 includes performing color correction on the wood cross-sectional image by a gray world method, calculating R, G, B a gain coefficient and a gray average value of three channels from the whole wood cross-sectional image and the expected values of pixels of the three channels R, G, B, and performing respective R, G, B three channel component adjustment on each pixel in the wood cross-sectional image according to the gain coefficient and the gray average value;
step S12 includes data enhancement of the corrected wood cross-sectional images using horizontal flipping, vertical flipping, and salt-and-pepper noise addition to bring each wood image training sample size into a specified number range.
Further, the system further comprises a load balancer, the load balancer is arranged between the client and the server, and data sent by the client are uniformly distributed on the server after being processed by the load balancer.
Further, the system also comprises an expert authentication support module for sending an expert authentication support request, wherein the expert authentication support request comprises the wood image to be authenticated and/or the recognition result returned by the wood image AI recognition module, and receiving expert authentication information returned in response to the expert authentication support request.
Further, the database comprises MySQL and/or Oracle and/or SQLite.
The invention has the following advantages:
a. the clearer wood image is acquired through the micro-distance image acquisition terminal, so that finer features can be conveniently extracted, wood recognition is facilitated, and the wood recognition accuracy is improved;
b. the load equalizer is adopted to enable the data to be distributed on the server side evenly, so that the workload of development and deployment is greatly simplified, the wood image recognition time length is reduced, and the long-term operation of the system is facilitated;
c. the system is simple in operation and convenient to use.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a block diagram of a system for acquiring and AI identification of macro characteristic images of wood according to an embodiment of the present invention;
FIG. 2 is a workflow diagram of a Web end of a wood macro characteristic image acquisition and AI identification system provided by an embodiment of the invention;
FIG. 3 is a flowchart of Web end image recognition of the wood micro-distance feature image acquisition and AI identification system provided by the embodiment of the invention;
FIG. 4 is a schematic diagram of internal storage of a user database of a wood macro characteristic image acquisition and AI identification system according to an embodiment of the invention;
fig. 5 is a schematic structural diagram of a wood image AI identification module of the wood macro feature image acquisition and AI identification system provided by the embodiment of the invention;
fig. 6 is a schematic structural diagram of an assembled macro image acquisition terminal of the wood macro characteristic image acquisition and AI identification system provided by the embodiment of the invention;
fig. 7 is an enlarged schematic diagram of an adapter of an assembled macro image acquisition terminal of the wood macro characteristic image acquisition and AI identification system provided by the embodiment of the invention;
fig. 8 is a schematic structural diagram of an embedded macro image acquisition terminal of the wood macro characteristic image acquisition and AI identification system provided by the embodiment of the invention;
fig. 9 is a schematic structural exploded view of an embedded macro image acquisition terminal of the wood macro characteristic image acquisition and AI identification system according to an embodiment of the present invention;
Fig. 10 is a schematic structural exploded view of an embedded macro image acquisition terminal of the wood macro characteristic image acquisition and AI identification system according to an embodiment of the present invention after replacing a press ring;
fig. 11 is a schematic top view of a luminaire of an embedded macro image acquisition terminal of the wood macro feature image acquisition and AI identification system according to an embodiment of the present invention.
Wherein, the reference numerals include: the lens comprises a 1-micro lens, a 10-lens, a 12-lens seat, a 120-seat body, a 120 a-first split, a 120 b-second split, a 121-compression ring, a 122-lens external connection part, a 2-adapter, a 21-first connection part, a 22-second connection part, a 3-acquisition head, a 30-acquisition base, a 31-flexible ring, a 32-lamp, a 320-annular lamp band, a 321-brightness body, a 322-external power supply connector, a 1-first annular cavity, a 2-second annular cavity, a 3-third annular cavity, b 1-annular clamping grooves, c-charging interfaces, q-perspective cavities, q 1-first cavities, q 2-second cavities and k-embedding gaps.
Detailed Description
For better understanding of the present invention, the objects, technical solutions and advantages thereof will be more clearly understood by those skilled in the art, and the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. It should be noted that the implementation manner not shown or described in the drawings is a manner known to those of ordinary skill in the art. Additionally, although examples of parameters including particular values may be provided herein, it should be appreciated that the parameters need not be exactly equal to the corresponding values, but may be approximated to the corresponding values within acceptable error margins or design constraints. It will be apparent that the described embodiments are merely some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, in the description and claims, are intended to cover a non-exclusive inclusion, such that a process, method, apparatus, article, or device that comprises a list of steps or elements is not necessarily limited to those steps or elements that are expressly listed or inherent to such process, method, article, or device.
In one embodiment of the invention, a wood micro-distance characteristic image acquisition and AI identification system is provided, as shown in FIG. 1, comprising a client, a database, a server, a wood image AI identification module and a micro-distance image acquisition terminal.
The client is used for receiving user operation; the database comprises a user database and a wood database, wherein the user database is used for storing user data, the wood database is used for storing wood information, the server side comprises a security authentication module and a service processing module, the security authentication module is used for verifying the user information, and the service processing module is used for processing related services; the wood image AI identification module is used for identifying the wood image to be identified to obtain wood species information of the wood; the macro image acquisition terminal is arranged on a camera lens of the mobile terminal and is used for shooting detailed images of wood to be identified.
The client and the database are respectively connected with the server, the client receives the login request and sends the login request to the server, after the server receives the login request, the security authentication module verifies whether the login information of the user is correct by comparing the user data in the user database,
If not, the security authentication module refuses the login request and sends a re-login instruction to the client;
if the identification result is correct, the service processing module can receive an image identification request sent by the client, the image identification request comprises a wood image to be identified, the service processing module identifies the wood image through the wood image AI identification module, the wood image AI identification module receives the wood image to be identified, obtains the identification result of the wood image to be identified through a convolutional neural network and feeds the identification result back to the service processing module, and the service processing module searches the wood database according to the identification result returned by the wood image AI identification module and feeds the search result back to the client.
The wood micro-pitch feature image acquisition and AI identification system also comprises an expert identification support module for sending an expert identification support request, wherein the expert identification support request comprises the wood image to be identified and/or the identification result returned by the wood image AI identification module, and receives expert identification information returned in response to the expert identification support request.
For the client, it is divided into a Web terminal, a mobile Web terminal and an App mobile terminal, in one embodiment of the present invention, as shown in fig. 2, the Web terminal is taken as an example for a specific description, and the development and deployment of the Web terminal of the wood macro feature image acquisition and AI authentication system are as follows:
the system is developed by utilizing HTML, CSS and JavaScript technology, and a system frame is built by utilizing Vue-CLI; designing a system interface by using an Element UI, modularizing each function of the front end, using an ES2015+ standard, downwards compatible with ES5, and adapting to all main stream browsers; using the Vue Router as a route manager, supporting functions of nesting of a route/view chart, modularized component configuration, navigation fine granularity control and the like; the network request uses an Axios network request based on an HTTP protocol to support functions of XMLHttpRequest, converting request data and response data, automatically converting JSON data, preventing XSRF attack and the like; the identification information is stored through the Token, so that unnecessary interaction information is reduced, and the robustness of the system is improved; and when the front end is developed, the mock. Js analog data response is used, so that the front end development and the back end development are completely separated, and the whole system is highly decoupled.
As shown in fig. 2, after the user logs in successfully, the server returns a corresponding function list and Token according to the user rights, where it is to be noted that the users are classified into a management user, an expert user, and a general user, where the user rights are classified into a management user right, an expert user right, and a general user right, where the general user right is relatively minimum, the expert user right is relatively more, and the management user right includes all rights, and the management user can operate information, rights, and the like of all users, and other users cannot operate, which is described in detail below.
When the Web terminal initiates a user request, the Token is encapsulated in an HTTP request header, and the server terminal carries out security and authority verification on the user information through Token information. And when the image is identified, the user encapsulates the image into an HTTP request body in a Form, the server receives the picture, returns the wood related information, the related investigation case and the customs information after the identification is completed, and renders the information on a webpage for the user to refer to. If the user needs to browse the wood information, the Web end sends a related request to the server end, and because the wood information is more in quantity, the information needs to be paged for the user to inquire and browse, the request parameters need to carry pageNum and pageSize fields which respectively represent the current page number and the number of information items contained in one page, and the server end returns corresponding data to the Web end for the user to browse and reference. If the user has a question about the recognition result, the user can ask questions to the expert through the expert online support function, wherein the question asking mode is in an image-text mode and an audio-video mode, in the image-text mode, the user can check the expert opinion in own question asking records by filling in question description and uploading related images, the Web terminal packages information into an HTTP request body and uploads the information to the server terminal, the server terminal stores data into the database, and the expert gives guidance opinion after checking the questions. If the user needs to obtain expert support in a short time, the audio and video mode can be selected, and the encountered problems are solved by face-to-face communication with the expert.
In one embodiment of the present invention, when the client performs data transmission with the server, there is an overload fault occurring in one or more servers, so the wood macro feature image acquisition and AI authentication system further includes a load balancer, where the load balancer is disposed between the client and the server, and data sent by the client is uniformly distributed on the servers after being processed by the load balancer.
For the server, a Spring frame may be selected, or other frames may be selected, such as a structures 2 frame, a socket frame, a label frame, and a strips frame, which do not limit the protection scope of the present invention. In one embodiment of the present invention, as shown in fig. 3, a Spring Boot frame is taken as an example to specifically describe:
the Spring Boot framework realizes the automatic configuration of a plurality of frameworks on the basis of the Spring framework, and meanwhile, the Tomcat server is built in, so that the workload of development and deployment is greatly simplified. And the Web terminal interacts with the server terminal to use a RESTFul format request. The user request is sent to the server through the HTTP protocol, and the request is firstly subjected to load balancing through Nginx and then forwarded to the background of the server through a polling mode. After the background receives the user request, the user authentication and authorization are performed by combining the JWT by using the Spring Security. The Spring Security is a Security management framework provided by the Spring framework, and can be used for filtering illegal requests and verifying whether user information is correct, and after the user is successfully verified, relevant service processing is performed and returned to the Web terminal. When the user performs wood image recognition, the background of the server side firstly verifies the legality of the image after receiving the wood image, and then forwards the result to the Django server side after confirming that the problem exists. The Django server is a Web framework developed based on Python, and the wood macro characteristic image acquisition and AI authentication system uses the Django framework to avoid performance loss caused by direct language interconversion. The method comprises the steps that after a wood image to be identified is received by the Django server, a wood image identification model is called to obtain an identification result, the identification result is returned to the Django server, relevant information is extracted after the identification result is received by the Django server, the relevant information of the wood is queried from the wood database, the relevant information of the wood comprises information corresponding to Latin names, wood places, classification positions and the like, after the information is packaged, the information is returned to a Web end in a JSON format, and a user can see the identification result at the Web end, and the whole process is about 1-2 seconds.
One or more of MySQL, oracle, SQLite or other databases may be used for the database, where the database is selected according to actual requirements, and the protection scope of the present invention is not limited. In one embodiment of the present invention, as shown in fig. 4, mySQL is taken as an example for specific explanation:
the data of the wood micro-distance characteristic image acquisition and AI identification system is divided into two parts, namely user information and wood information.
The user information data storage adopts RBAC (Role-BasedAccess Control ) model, and user related information is managed by means of Role-associated users and Role-associated authorities. Fig. 4 shows a structure of a user related information table in the wood micro-distance feature image acquisition and AI identification system, user information such as user id, job number, user name, password, etc. is stored in a sys_user table, role authority information such as authority name, authority key, etc. is stored in a sys_role table, and all function interfaces in the system are stored in a sys_menu table, which is associated with the sys_role table to determine authority information to which the user belongs.
The wood information is stored with reference to an RBAC model, the wood information table comprises wood name, classification status, production place and other information, the investigation and transmission case information and the customs internal information are respectively stored in other two tables, and then the two association tables are used for associating the wood with the investigation and transmission case information and the customs internal information. Meanwhile, by combining with Redis as a cache database, storing hot spot information in the Redis can enhance the capability of the wood macro characteristic image acquisition and AI identification system for processing high concurrence user requests, and improve the system stability.
For the wood image AI-recognition module, as shown in fig. 5, the specific steps of its operation are as follows:
s1, inputting a wood image and preprocessing;
s2, processing the wood image in a blocking way;
s3, training the segmented sub-images through a convolutional neural network model;
s4, adopting different gradient values from the edge to the center of the wood image as weights of sub-image classification scores of different areas, increasing the proportion of the center area in the whole wood image classification score, converting the weighted score into a final probability value, obtaining the wood species information of the wood image, wherein,
the step of preprocessing in S1 includes:
s11, performing color correction on the wood image;
s12, carrying out data enhancement on the corrected wood image;
the step of processing the wood image in the S2 includes:
s21, dividing the preprocessed wood image to obtain a plurality of sub-images;
s22, unifying size pixels of each sub-image through a bilinear interpolation method.
Step S11 includes performing color correction on the wood cross-sectional image by a gray world method, calculating R, G, B a gain coefficient and a gray average value of three channels from the whole wood cross-sectional image and the expected pixel values of the three channels R, G, B, and adjusting R, G, B three channel components of each pixel in the wood cross-sectional image according to the gain coefficient and the gray average value.
Step S12 includes data enhancement of the corrected wood cross-sectional images using horizontal flipping, vertical flipping, and salt-and-pepper noise addition to bring each wood image training sample size into a specified number range.
It should be noted that the above preprocessing method is only an example, and the preprocessing may also use methods such as contrast broadening, logarithmic transformation, density layering, histogram equalization, gaussian filtering, and laplace filtering, which do not limit the protection scope of the present invention.
In one embodiment of the present invention, as shown in fig. 5, a wood image is first preprocessed; secondly, processing the wood image in a 7 multiplied by 7 blocking mode, and thirdly, training the image through a ResNet101 convolutional neural network model; and fourthly, adopting different gradient values from the edge to the center of the wood image as weights of different sub-region image classification scores, increasing the proportion of the center region in the whole wood image classification score, converting the weighted scores of various types into final probability values by using a Softmax method, obtaining the probability values of various types of wood by using the Softmax method, automatically identifying the wood image, and inquiring the wood type information of the wood image by combining the wood database.
It should be noted that the above-mentioned blocking method is only illustrative, other blocking methods, such as 5×5 and 10×10, and similarly, the above-mentioned training model is also only illustrative, and other training models, such as VggNet16, googleNet, denseNet, mobileNetv3, resNet50 and ResNet152, may be used, which do not limit the protection scope of the present invention.
In one embodiment of the invention, a HTML, CSS, javaScript technology and Vue-CTI are utilized to build a Web end; establishing a timber database by using MySQL for storing timber information, investigation and distribution cases, recognition history records and other information; establishing a wood recognition model based on ResNet101, and deploying the model on a server developed based on a Spring Boot frame; and calling a deep learning model by using a Django framework, calling the wood recognition model and returning a recognition result when a wood image recognition request is sent by a Spring Boot server, searching corresponding wood information, associated searching cases and other information in the wood database by the Spring Boot server through the recognition result, feeding back the information to a user and displaying the information on a Web end, and storing the recognition result in the database. When a user encounters unknown wood, the wood image is acquired through a microscope or a mobile phone and the like and is imported into a computer, the wood image is uploaded through a Web end, feedback information is fed back after the recognition is completed by the service end and is displayed on the Web end, and the recognition process is about 1-2s. If the user has a question about the recognition result, an expert remote diagnosis may be requested. The wood micro-distance characteristic image acquisition and AI identification system provides an auxiliary tool for rapidly identifying the wood types, and economic cost and time cost are greatly saved.
For the macro image acquisition terminal, an assembled macro image acquisition terminal and an embedded macro image acquisition terminal can be adopted, and other macro image acquisition terminals can be adopted, wherein the macro image acquisition terminal is selected according to actual requirements, and the protection scope of the invention is not limited. The assembled macro image acquisition terminal and the embedded macro image acquisition terminal are respectively described:
(1) Assembled micro-distance image acquisition terminal
As shown in fig. 6, the assembled macro image acquisition terminal includes a macro lens 1, an adapter 2 connected to the outer end of the macro lens 1, and an acquisition head 3 disposed at the outer end of the adapter 2 and capable of compensating exposure.
The collecting head 3 comprises a collecting base 30 with an annular cavity, a soft light ring 31 and a lamp 32 which are arranged on the collecting base 30, wherein the adapter 2, the soft light ring 31 and the surface of a sample to be detected or the adapter 2, the soft light ring 31 and the surface of a platform for placing the sample to be detected form a closed image collecting area.
The end surface of the collecting base 30 far away from the adapter 2 is arranged in parallel with the lens 10 of the macro lens 1. After the lens is connected to the shooting equipment (mobile phone), the lens of the shooting equipment and the end face of the acquisition base, which is far away from the adapter, can be ensured to be parallel, so that a structure enlarged picture of the surface of the sample to be detected can be obtained under the same focal length.
The soft light ring 31 is arranged at the end part of the collection base 30 far away from the adapter 2, and the end surface of the soft light ring 31 far away from the collection base 30 is arranged in parallel with the lens 10 of the macro lens 1. Therefore, when the fixed focus photographing is performed, the lens is ensured to be parallel to the surface of the sample to be detected, and further necessary conditions are provided for obtaining satisfactory photos.
In one embodiment of the present invention, the flexible ring 31 is formed with a first annular cavity a1, a second annular cavity a2, and a third annular cavity a3 formed by extending the second annular cavity a2 from the outer end surface to the outside in order from the inside to the outside.
The cross section of the first annular cavity a1 is rectangular, the cross sections of the second annular cavity a2 and the third annular cavity a3 are isosceles trapezoids, and the soft light ring 31 is erected on the surface of a sample to be detected or the surface of a platform for placing the sample to be detected from the end surface of the third annular cavity a3. By the arrangement, better illumination conditions are created, and focusing is more accurate when the macro lens is used for photographing.
In one embodiment of the present invention, the centerlines of the first, second and third annular cavities a1, a2 and a3 coincide with the centerline of the lens 10.
Meanwhile, in one embodiment of the present invention, the inner diameter of the first annular cavity a1 is larger than the outer diameter of the lens 10, and the inner diameter of the inner end of the second annular cavity a2 is larger than the inner diameter of the first annular cavity a 1. The exposure effect is optimal at this time.
The light fixture 32 is an annular light strip circumferentially disposed along the collection base 30, a plurality of LED lights disposed on the annular light strip, and a power source and switch, wherein the power source is a rechargeable lithium battery. By the arrangement of the annular light band, the light can be compensated from all directions to the acquisition area, so that photographing is facilitated; meanwhile, the lamp 32 can be directly powered by the lithium battery, so that the lamp is more convenient to use.
The center line of the annular lamp band is arranged to coincide with the center line of the lens 10. Thus, the annular light supplementing effect is optimal.
In addition, a charging interface c which is communicated with a power supply is further arranged on the collecting base 30. Therefore, the mobile phone can directly supply power to the bright body, and the practicability of the micro-distance image acquisition terminal is further improved.
Referring to fig. 7, the adapter 2 includes a first connecting portion 21 connected to the macro lens 1, and a second connecting portion 22 detachably connected to the collecting base 30, wherein the first connecting portion 21 and the second connecting portion 22 are internally penetrated to form a perspective cavity q, an annular clamping groove b1 is formed at an end portion of the second connecting portion 22 away from the macro lens 1, and an annular protrusion b2 matched with the annular clamping groove b1 is provided on an outer periphery of the collecting base 30.
Therefore, the assembled macro image acquisition terminal has the following advantages:
a. the annular light surrounding the camera can provide accurate annular light supplement taking the shot object as the center (no annular light supplement lamp taking the mobile phone camera as the center at present) for the shot object, and the shadow of the shot object can be eliminated greatly relative to other lateral light;
b. the light supplementing camera can be flexibly replaced aiming at the multi-camera mobile phone by matching with a plurality of fixing clamps;
c. the size is very small, the requirement of high flexibility of micro-distance photography can be met, and the portability is high;
d. the power supply is convenient to acquire, and the mobile phone can directly supply power;
e. the micro-lens and the acquisition head are quickly assembled through the adapter so as to be convenient to carry, and meanwhile, the micro-lens and the acquisition head can be attached to the surface of an object to shoot a microstructure of a plane, and can shoot tiny objects (such as insects, jewelry, flowers and the like);
f. the assembled macro image acquisition terminal and the macro shooting equipment are matched for use, and the lamp supplements light in the image acquisition area, so that the focusing distance is reduced under the macro lens, and the lens is stable and shadowless in the shooting process, so that a photo with good effect (such as high definition) can be quickly and accurately obtained, and meanwhile, the camera can be operated by a single person and is quite convenient.
(2) Embedded micro-distance image acquisition terminal
As shown in fig. 8, the in-line macro image acquisition terminal includes a macro lens 1 and a lamp 32 for compensating exposure.
As shown in fig. 9, the macro lens 1 includes a lens holder 12 and a lens 10, wherein the lens holder 12 includes a holder body 120, a pressing ring 121 detachably provided at a front end portion of the holder body 120, and a lens external connection portion 122 provided at a rear end portion of the holder body 120.
In one embodiment of the present invention, the pressure ring 121 is annular, and the inner cavity of the seat body 120, the inner cavity of the pressure ring 121, and the surface of the sample to be detected form a closed image acquisition area.
The seat body 120 includes a first split 120a and a second split 120b in which a first cavity q1 and a second cavity q2 are formed to communicate with each other, respectively.
Specifically, the first cavity q1 is cylindrical, the second cavity q2 is conical with gradually larger diameter from the first cavity forward, and the lens 10 is installed at the communication position of the first cavity q1 and the second cavity q2 in an interception mode.
The lens 10 is arranged in parallel with the front end face of the press ring 121, so that a structure enlarged picture of the surface of the sample to be detected can be obtained under the same focal length.
Referring to fig. 10, the pressing ring 121 is detachably connected to the second sub 120b, and the inner cavity of the pressing ring 121 is in a straight cylindrical shape or in a conical shape with a freely forward inner diameter gradually increasing. Selection is made according to the subject to be photographed, for example: for objects, the small objects are positioned in the straight cylinder area and are close to the lenses, so that the enlarged pictures of the clear object tissue structures can be obtained more easily; once the volume of the shot object is large, as in a conventional wood section texture structure, the frustum shape is adopted, so that an image acquisition area can be enlarged, and focusing can be facilitated.
The lens external connection portion 122 is an external thread provided on the outer periphery of the first split body 120 a.
The lamp 32 is annular, the lamp 32 is embedded in the second cavity q2, and the pressing ring 121 abuts against the lamp 32.
The diameter of the annular inner cavity of the lamp 32 is larger than the outer diameter of the lens 10, and the center line of the lamp 32, the center line of the lens 10 and the center line of the compression ring 121 are overlapped.
As shown in fig. 11, the lamp 32 includes an annular lamp band 20, a plurality of luminous bodies 21 provided on the annular lamp band 20, and a power supply and a control switch.
Specifically, in one embodiment of the present invention, the illuminating body 21 is a conventional bulb such as: an LED lamp. The lamp 32 further comprises an external power connector 22 extending outwards from one side of the annular lamp strip 20, wherein the external power connector 21 is embedded from the embedded notch k, and the annular lamp strip 20 is abutted against the wall surface of the second cavity q2 and is arranged at a distance from the lens. By arranging the external power connector, on one hand, the positioning of the annular lamp belt can be realized, and the displacement is prevented; on the other hand, the power supply of the mobile phone can be communicated with the power supply of the lamp through the transmission line, so that the mobile phone directly supplies power.
Therefore, the embedded macro image acquisition terminal has the following advantages:
a. The annular light surrounding the camera can provide accurate annular light supplement taking the shot object as the center (no annular light supplement lamp taking the mobile phone camera as the center at present) for the shot object, and the shadow of the shot object can be eliminated greatly relative to other lateral light;
b. the light supplementing camera can be flexibly replaced aiming at the multi-camera mobile phone by matching with a plurality of fixing clamps;
c. the size is very small, the requirement of high flexibility of micro-distance photography can be met, and the portability is high;
d. the power supply is convenient to acquire, and the mobile phone can directly supply power;
e. after the micro-lens and the shooting equipment are combined, the micro-lens and the shooting equipment can be attached to the surface of an object to shoot a microstructure of a plane, and can shoot tiny objects (such as insects, jewelry and flowers), meanwhile, the light is supplemented in an image acquisition area by a lamp, the focusing distance is reduced under the micro-lens, and the lens is stable and shadowless in light supplementing condition in shooting, so that a photo with good effect (such as high definition) can be quickly and accurately obtained, and meanwhile, the micro-lens can be operated by a single person and is quite convenient.
The development and deployment of the Web end, the user and the user authority are thoroughly described above, and the following specific description is made on the page layout of the Web end:
the Web page comprises eight major modules of a home page, intelligent identification, a database, online support, statistical analysis, learning management, authority setting and mobile terminal management. Wherein, for the management user, the authority thereof relates to all the above modules; for expert users, the authority of the expert users relates to six modules, namely a home page, intelligent identification, a database, online support, statistical analysis and learning management; for the common user, the authority of the common user relates to three modules of a home page, intelligent identification and a database. In relatively terms, the ordinary users have the least authority, and the expert users have relatively more authority.
The above eight modules will now be specifically described:
(1) "front page"
The first page is divided into three columns, wherein the first column shows the condition that the user uses App to identify wood yesterday, and the number of active users yesterday, the number of times of yesterday identification and the type of high-frequency identification wood yesterday are respectively shown; the second column shows the tree species identification accuracy ranking, three modes of a data report, a line graph and a histogram are provided to realize data visualization, and a user can download the visualized image; the third column shows the wood identification number ranking and the daily user identification number.
(2) Intelligent identification "
The intelligent identification is divided into two steps, firstly, pictures are added, the pictures can be selected from a picture library to upload, 20 jpg or png format pictures can be uploaded at most at a time, the uploaded pictures can be checked online again, and invalid images can be selectively or completely cleared; and secondly, starting identification, clicking a 'starting identification' after checking, so as to perform wood identification, and immediately displaying an identification result below, wherein the identification result respectively displays the first three tree species with larger confidence, clicking a 'wood species image-text' can check the detailed information of the tree species, and clicking a 'checking case' can check the detailed information of the checking case of the tree species if the checking case exists. In addition, after the recognition is finished, the 'clear recognition result' can be clicked to clear the recognition information, so that the page is kept clean.
(3) "database"
The "database" includes the following four parts:
the first part is a wood pattern library, at present, the wood pattern library contains 122 trees in total, the interface supports a wood searching function, a user can input Chinese names, latin's names, market names, family information, distribution places and endangered grades of the wood for searching, if the wood exists in the pattern library, the detailed information of the wood can be automatically popped up from the lower part, otherwise, temporary data is output. In addition, the interface also supports four functions of 'new addition', 'modification', 'deletion' and 'export' of the wood, and a user can click a corresponding module according to the requirement to correct the wood image-text library.
The second part is the management of the family and genus, the order of the family and genus and the related information of whether the family and genus exist or not can be obtained through the search of the name of the input family and genus, and if the family and genus does not exist, the user can add the family and genus on line. In addition, the user can jump to the corresponding position according to the genus of the family to manage the new addition, modification and deletion of the corresponding genus of the family.
The third part is case finding, the user can search cases according to the case title, the finding time, the related tree species, the finding customs, the origin and the endangered level, and if the finding case does not exist in the database, the user can carry out online new addition. The new case of the user needs to be added with information such as case title, investigation time, related tree species, investigation customs, origin, detailed description, text and the like. The interface also supports "add", "modify", "delete" and "export" of the case under investigation.
The fourth part is front lookout, and the user can input the front lookout title, whether the front lookout title is rotated or not and the related tree species to perform information retrieval. If the library is not present, online addition is supported. The user can click a "new" button, fill in the front lookout header, select the relevant tree species, the pictures of the front lookout, the front lookout content, and determine whether to add as a rotation map. The supporting rotation blog graph can realize that a common user can know the latest information when the related information is displayed on an App mobile terminal. The user can also modify the added front lookout header, select related tree species, pictures of the front lookout, front lookout content and whether the added front lookout is related to the alternate broadcast pictures.
(4) "Online support"
"online support" includes the following three parts:
the first part is image-text online support, a common user can initiate a post online, input a theme title and a theme content, and add related timber pictures to wait for an expert or other users to give a comment; the part also supports the retrieval of historical posts, and a user can input the user name of an application post or the title of the post for retrieval; the part also supports the application of voice-video call to carry out voice or video communication with an online expert, and can also check the historical call record.
The second part is the audio and video online support, which supports online communication, and can select users needing communication and also supports adding new users.
The third part is expert authentication support, the part is a re-verification link of machine recognition, the expert can see the image of the recognized timber uploaded by the common user and the recognition result, and the expert can diagnose whether the recognition is accurate or not on line and give comments. The part also supports the inquiry of the historical identification information, and a user can provide a user name, identify the type name of the wood species, identify the rate and the like to search the historical information.
(5) Statistical analysis "
"statistical analysis" is in two parts:
the first part is user information statistics, mainly displaying the total user number, the identification times in the last 30 days and the historical identification total times, and selecting the last five months, the last five weeks and the last five days to check the user identification times. The part provides three data visualization modes of a data report, a line graph and a bar graph for users to select and supports downloading of images. In addition, the section provides a top ten number of thirty days or historical tree species identification times to be presented, visualizes the data with data reports, line graphs and bar graphs, and supports downloading of images.
The second part is identification information statistics, which shows the identification times ranking (top ten tree species), the tree species identification accuracy ranking, and the tree species identification accuracy ranking (reverse order), respectively. The user may select recognition accuracy ranks at different time periods.
(6) "study management"
"learning management" is divided into three parts:
the first part is auditing and warehousing, an administrator can select 'warehousing' or 'deletion' according to data uploaded by a user, a retrieval interface is also provided, and the administrator can input a user name, identify the wood species type, identify the rate, whether auditing and creating time are performed for retrieval.
The second part is data warehouse entry, the part supports searching and checking warehouse entry conditions, and a user can input information such as wood species type, creation time and the like to inquire warehouse entry conditions.
The third part is model training, an administrator can select a model for identifying wood and add a new model, and when the administrator needs to add the new model, the new model can be added by clicking 'new addition'.
(7) Authority setting "
The authority setting is respectively divided into three parts:
the first part is authorized by the user, and the part gathers all customs departments in China, can narrow the search range by inputting the names of the customs departments, and can also search by inputting information such as user numbers, mobile phone numbers, user work numbers and the like. The user can select to delete or enter a stop-use state. Click modification may modify relevant information for the user, such as name, home department, cell phone number, user job number, etc. The administrator can click on 'new creation' to add new users, the users are required to provide real names, home departments, mobile phone numbers, user work numbers and the like, when the users forget passwords, the administrator can help the users to reset the passwords, and the administrator can click on 'reset' to input the new passwords.
The second part is user authority modification, the part provides online retrieval, an administrator can input user names, authority characters, states and creation time, and the user authority can be modified after the user is retrieved. The project engineering is currently divided into four user roles, namely an administrator, a customs administrator, an expert and a common user. The administrator may click "modify" to pick a function in front of the corresponding function that the user may use.
The third part provides an on-line searching function for the departments, the parts can search through inputting the names or the states of the departments, the information correction of the China customs departments is convenient, when the information of the customs departments is changed, an administrator can click on 'modification', the basic information of the departments is maintained, and when a new customs department is added, the administrator can click on 'addition', and the new customs department is added.
(8) Mobile terminal management "
"Mobile end management" is two parts:
the first part is version updating service, an administrator can click on 'add new version', an App installation package of the latest version is uploaded, and a user can download and install App software of the latest version in time at a webpage end.
The second part is a mobile terminal home page display diagram, the part provides a mobile terminal first display diagram, an administrator can set an App terminal home page display diagram, and a new display diagram can be added.
The above is the page layout of the Web terminal, and the following is the page layout of the App mobile terminal:
the App mobile terminal page comprises five modules of 'front lookout', 'knowledge dictionary', 'intelligent identification', 'online support', 'My', and the five modules are specifically described below:
(1) Front lookout "
The top carousel graph of the front lookout can realize carousel content playing in turn. The carousel content needs to be set at a Web end; the corresponding front lookout information can be clicked to browse details; content viewing can be performed by sliding down; in addition, when browsing the article, the user can click a collection button at the upper right to collect the content, and can click a picture in the content of the article to browse the picture details.
(2) "knowledge dictionary"
By default, the material type image-text page is entered, and the upper 'case finding' button can be clicked to enter the case finding page. In the upper search bar, text may be entered for a fuzzy search, such as entering "wood," i.e., wood with "wood" in all names may be searched. The timber macro structure, the timber name, the timber Latin school name and the timber description are displayed in the timber species column from left to right in sequence, and corresponding buttons can be clicked to enter a timber species detail page. The wood pattern and text detail page is sequentially a wood name, a wood detail and a wood picture from top to bottom. In addition, the user can click the collection button at the upper right part to collect wood during browsing. The timber pictures can be slid left and right to check a plurality of pictures, and the timber pictures are clicked to enter picture details. The case search and issue page comprises a search field and a case search and issue field, and the case search and issue field can be clicked arbitrarily to enter case search and issue details. The detailed information of the investigation case is investigation case title, collection button, investigation case information, investigation case text and investigation case picture from top to bottom.
(3) Intelligent identification "
Clicking the middle picture button to take a picture or upload an album. In the uploading process, a progress bar prompt exists. The result page after successful uploading comprises TOP3 tree pictures and confidence. If the result is expert user, clicking the button of "enter result" on the upper right after clicking "confirm the timber for this", save the result to the operation of the database; if the request is a common user, clicking a 'request expert authentication button', storing the request to the background, and uniformly processing the authentication request by the expert at the Web end.
(4) "Online support"
The same as the "online support" module in the Web terminal is not described here again.
(5) "My"
"My" refers to user information and rights settings.
The wood micro-distance characteristic image acquisition and AI identification system comprises a client, a database, a server, a wood image AI identification module and a micro-distance image acquisition terminal, and has the functions of intelligent wood image identification, wood information browsing and inquiring, expert remote online support, user identification history browsing and the like. When workers encounter the wood which is difficult to distinguish, the acquired wood image is uploaded to a server, the server feeds the identification result back to the user within 1-2s, and an expert can be applied to a questionable place for remote online identification, so that wood type information reference is provided for the user rapidly. And, the wood species with the questionable recognition result and not in the database can be further identified by an expert, so that the accuracy of wood identification is ensured.
The foregoing description of the preferred embodiments of the invention is not intended to limit the invention to the precise form disclosed, and any such modifications, equivalents, and alternatives falling within the spirit and scope of the invention are intended to be included within the scope of the invention.
Claims (9)
1. The utility model provides a timber microspur characteristic image gathers and AI appraisal system which characterized in that includes:
the client is used for receiving user operation;
a database comprising a user database for storing user data;
the server comprises a security authentication module and a service processing module;
the wood image AI identification module is used for identifying the wood image to be identified to obtain wood species information of the wood;
the client and the database are respectively connected with the server, the client receives the login request and sends the login request to the server, after the server receives the login request, the security authentication module verifies whether the login information of the user is correct by comparing the user data in the user database,
if not, the security authentication module refuses the login request and sends a re-login instruction to the client;
if the wood image to be identified is correct, the service processing module can receive an image identification request sent by the client, wherein the image identification request comprises the wood image to be identified;
The business processing module recognizes the wood image through the wood image AI recognition module, and feeds back a recognition result to the client;
the wood image AI recognition module recognizes an image including the steps of:
s1, inputting a wood image and preprocessing;
s2, processing the wood image in a blocking way;
s3, training the segmented sub-images through a convolutional neural network model;
s4, adopting different gradient values from the edge to the center of the wood image as weights of sub-image classification scores of different areas, increasing the proportion of the center area in the whole wood image classification score, converting the weighted score into a final probability value, obtaining the wood species information of the wood image, wherein,
the step of preprocessing in S1 includes:
s11, performing color correction on the wood image;
s12, carrying out data enhancement on the corrected wood image;
the step of processing the wood image in the S2 includes:
s21, dividing the preprocessed wood image to obtain a plurality of sub-images;
s22, unifying size pixels of each sub-image through a bilinear interpolation method.
2. The wood micro-scale feature image acquisition and AI identification system of claim 1, further comprising a wood database, wherein the business processing module retrieves the wood database according to the recognition result returned by the wood image AI recognition module, and feeds back the retrieval result to the client.
3. The wood-micro-feature image acquisition and AI-qualification system of claim 1, further comprising a macro-image acquisition terminal mountable on a camera lens of a mobile terminal such that the camera lens of the mobile terminal captures a detail image of wood to be qualified.
4. The wood-micro-scale feature image acquisition and AI identification system of claim 3, wherein the macro-scale image acquisition terminal is an assembled macro-scale image acquisition terminal comprising a macro lens, an adapter disposed at an outer end of the macro lens, and an acquisition head disposed at an outer end of the adapter, wherein,
the adapter is internally provided with a perspective cavity for the perspective of the macro lens, the acquisition head comprises an acquisition base provided with an annular cavity communicated with the perspective cavity and a lamp arranged in the acquisition base, wherein the acquisition base, the adapter and the surface of a sample to be detected form a closed image acquisition area, or the acquisition base, the adapter and the surface of a platform for placing the sample to be detected form a closed image acquisition area.
5. The wood micro-distance characteristic image acquisition and AI identification system according to claim 3, wherein the micro-distance image acquisition terminal is an embedded micro-distance image acquisition terminal, the embedded micro-distance image acquisition terminal comprises a micro-distance lens, the micro-distance lens comprises a lens base and a lens, the lens base comprises a base body, a pressing ring detachably arranged at the front end part of the base body and a lens external connection part arranged at the rear end part of the base body, the pressing ring is annular, and a closed image acquisition area is formed by an inner cavity of the base body, an inner cavity of the pressing ring and the surface of a sample to be detected, or a closed image acquisition area is formed by an inner cavity of the base body, an inner cavity of the pressing ring and a platform surface on which the sample to be detected is placed;
the embedded micro-distance image acquisition terminal further comprises a lamp which is arranged on the inner wall of the seat body or the inner wall of the pressing ring and used for compensating exposure, the lamp is annular, and the center line of the lamp and the center line of the lens are overlapped.
6. The wood-micro-scale feature image acquisition and AI-authentication system of claim 1, wherein step S11 includes performing color correction on the wood cross-sectional image by a gray world method, calculating R, G, B a gain coefficient and a gray average value of three channels from the whole wood cross-sectional image and expected values of pixels of three channels R, G, B, and performing respective R, G, B three-channel component adjustment on each pixel in the wood cross-sectional image according to the gain coefficient and the gray average value;
Step S12 includes data enhancement of the corrected wood cross-sectional images using horizontal flipping, vertical flipping, and salt-and-pepper noise addition to bring each wood image training sample size into a specified number range.
7. The wood micro-distance characteristic image acquisition and AI identification system according to claim 1, further comprising a load balancer, wherein the load balancer is arranged between the client and the server, and data sent by the client are uniformly distributed on the server after being processed by the load balancer.
8. The wood-micro-feature image acquisition and AI-authentication system of claim 1, further comprising an expert authentication support module for sending an expert authentication support request, the expert authentication support request including the wood image to be authenticated and/or the recognition result returned by the wood image AI-recognition module, and receiving expert authentication information returned in response to the expert authentication support request.
9. The wood-micro-feature image acquisition and AI-qualification system of claim 1, wherein the database comprises MySQL and/or Oracle and/or SQLite.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110134276.9A CN112767388B (en) | 2021-01-29 | 2021-01-29 | Wood microspur characteristic image acquisition and AI identification system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110134276.9A CN112767388B (en) | 2021-01-29 | 2021-01-29 | Wood microspur characteristic image acquisition and AI identification system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112767388A CN112767388A (en) | 2021-05-07 |
| CN112767388B true CN112767388B (en) | 2023-12-26 |
Family
ID=75704399
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110134276.9A Active CN112767388B (en) | 2021-01-29 | 2021-01-29 | Wood microspur characteristic image acquisition and AI identification system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112767388B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| BRPI0704329A2 (en) * | 2007-10-23 | 2009-06-23 | De Oliveira Luiz Eduardo Soares | computational process of wood image analysis |
| CN108648191A (en) * | 2018-05-17 | 2018-10-12 | 吉林大学 | Pest image-recognizing method based on Bayes's width residual error neural network |
| CN110503051A (en) * | 2019-08-27 | 2019-11-26 | 西南林业大学 | A kind of precious timber identifying system and method based on image recognition technology |
| CN211352280U (en) * | 2019-12-23 | 2020-08-25 | 联想(北京)有限公司 | Light-emitting device and electronic equipment |
| CN111652307A (en) * | 2020-05-29 | 2020-09-11 | 广西大学 | Intelligent nondestructive identification method and device for redwood furniture based on convolutional neural network |
| CN111696033A (en) * | 2020-05-07 | 2020-09-22 | 中山大学 | Real image super-resolution model and method for learning cascaded hourglass network structure based on angular point guide |
| CN111898649A (en) * | 2020-07-07 | 2020-11-06 | 西南林业大学 | Wood tree species classification and identification method based on visible light or near infrared spectrum analysis |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA3054959C (en) * | 2017-03-13 | 2023-07-25 | Lucidyne Technologies, Inc. | Method of board lumber grading using deep learning techniques |
-
2021
- 2021-01-29 CN CN202110134276.9A patent/CN112767388B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| BRPI0704329A2 (en) * | 2007-10-23 | 2009-06-23 | De Oliveira Luiz Eduardo Soares | computational process of wood image analysis |
| CN108648191A (en) * | 2018-05-17 | 2018-10-12 | 吉林大学 | Pest image-recognizing method based on Bayes's width residual error neural network |
| CN110503051A (en) * | 2019-08-27 | 2019-11-26 | 西南林业大学 | A kind of precious timber identifying system and method based on image recognition technology |
| CN211352280U (en) * | 2019-12-23 | 2020-08-25 | 联想(北京)有限公司 | Light-emitting device and electronic equipment |
| CN111696033A (en) * | 2020-05-07 | 2020-09-22 | 中山大学 | Real image super-resolution model and method for learning cascaded hourglass network structure based on angular point guide |
| CN111652307A (en) * | 2020-05-29 | 2020-09-11 | 广西大学 | Intelligent nondestructive identification method and device for redwood furniture based on convolutional neural network |
| CN111898649A (en) * | 2020-07-07 | 2020-11-06 | 西南林业大学 | Wood tree species classification and identification method based on visible light or near infrared spectrum analysis |
Non-Patent Citations (3)
| Title |
|---|
| Wood species identification using convolutional neural network(CNN) architectures on macroscopic images;Anindita Safna et al.;《JITeCS》;第4卷(第3期);第274-283页 * |
| 基于轻量化卷积神经网络的木材种类识别;曾铮等;《现代计算机》;第59-63、75页 * |
| 曾铮等.基于轻量化卷积神经网络的木材种类识别.《现代计算机》.2020,第59-63、75页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112767388A (en) | 2021-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20180322353A1 (en) | Systems and methods for electronically identifying plant species | |
| JP6785305B2 (en) | Equipment and methods for video analysis techniques to identify individuals with face recognition and contextual video streams | |
| US11899681B2 (en) | Knowledge graph building method, electronic apparatus and non-transitory computer readable storage medium | |
| US8194985B2 (en) | Product identification using image analysis and user interaction | |
| CA2890158C (en) | Automated color processing and selection platform | |
| CN104239408B (en) | The data access of content based on the image recorded by mobile device | |
| US20150149454A1 (en) | Preferred image retrieval | |
| US20130329060A1 (en) | Overlay image providing system, server and method | |
| KR102089100B1 (en) | System for providing image database based interior design service minimizing difference between client and designer | |
| CN105450987A (en) | Intelligent recognition platform video monitoring early warning system | |
| US20120089676A1 (en) | Live data form viewing and updating system | |
| US9411839B2 (en) | Index configuration for searchable data in network | |
| US20070070217A1 (en) | Image analysis apparatus and image analysis program storage medium | |
| US20200082199A1 (en) | Method and apparatus for inspecting burrs of electrode slice | |
| CN107392238A (en) | Outdoor knowledge of plants based on moving-vision search expands learning system | |
| JP2019520662A (en) | Content-based search and retrieval of trademark images | |
| CN110166525A (en) | Resource information processing management system applied to garden | |
| CN109711443A (en) | Floor plan recognition methods, device, equipment and storage medium neural network based | |
| WO2019171803A1 (en) | Image search device, image search method, electronic equipment, and control method | |
| CN104850636A (en) | Animal/plant identification method and system for mobile terminal | |
| CN108229780B (en) | Workshop management system and method based on tablet computer | |
| CN108230220B (en) | Watermark adding method and device | |
| CN112767388B (en) | Wood microspur characteristic image acquisition and AI identification system | |
| Luciani et al. | Large-scale overlays and trends: Visually mining, panning and zoomingthe observable universe | |
| EP3850538A1 (en) | Image management with region-based metadata indexing |
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 |