A method of image feature description based on database

Abstract

The present invention belongs to the field of electron microscope analysis technology, and specifically relates to a method for describing image features based on a database, which is characterized by using a description method of pixel-line segment-feature-particle-frame image, and storing these structured data in a relational database. Compared with the prior art, the present invention has the following beneficial effects: the present invention describes the particles identified after removing the background with this feature data structure, and then stores it in the database, which has faster data retrieval speed and storage speed than directly storing pictures or other custom binary data structures, and relying on the powerful query capability of the database SQL language, it can realize various data joint queries very conveniently, which has no negative impact on the overall measurement accuracy, but has a great effect on improving the measurement speed and report retrieval speed, and does not require a large amount of memory, and can run very quickly.

Description

Technical Field

[0001] The invention belongs to the technical field of electron microscope analysis, and in particular relates to an image feature description method based on a database. Background Art

[0002] The mineral automatic identification system continuously and automatically obtains multiple backscatter image (BSE) information and X-ray (X[1]Ray) information of mineral samples, and obtains the type of mineral after analysis. The main purpose of obtaining the BSE image is to identify the mineral particles in the sample; after identifying the mineral particles, the X-Ray information of the mineral particles is obtained, and the type of mineral is identified based on the spectral peak characteristics of each element in the mineral X-Ray information.

[0003] The automatic material recognition and analysis system of the electron microscope generally collects the BSE image frame of the sample through the electron microscope API, then manually sets the gray value to distinguish the image foreground and background, removes the background pixels through pixel-by-pixel processing, and finally realizes the identification of particles through the connectivity algorithm. After identifying the position of the particle, the elemental composition of the particle is collected through the energy spectrum X-ray. Finally, the identification of the particle material is realized.

[0004] The existing mineral particle identification algorithm based on BSE image pre-sets the background gray value of the ore sample, and uses a binarization method according to the background gray range, that is, the pixels belonging to the background gray are set as the background, and the pixels not in the background gray range are identified as mineral particles. The above method generally uses binary files or binary image files to describe and store the morphological characteristics of the particles after removing the background, such as using contours or ROI methods to describe in OpenCV. Since the working conditions of the electron microscope are easily affected by the environment, when multiple BSE images are taken continuously, the gray value, brightness value, and contrast information of each BSE image will change. In the "OTS Inclusion Analysis System", these feature information is originally stored in binary files. If data retrieval and analysis are to be performed, all data must be loaded into the memory at one time, and then all data must be analyzed in the memory using traversal technology to obtain the results, which makes the program have a large demand for memory, and sometimes only a small analysis query requires all data to be loaded, which is very slow. Summary of the invention

[0005] The present invention aims to provide a method for describing image features based on a database, overcome the shortcomings of the prior art, use a description method of pixel-line segment-feature-particle-frame image, store these structured data in a relational database, and use the efficient data management capabilities of the database to achieve rapid retrieval and analysis of image features, thereby meeting the accuracy and speed requirements of electron microscope analysis.

[0006] In order to solve the above technical problems, the technical solution of the present invention is:

[0007] A method for describing image features based on a database is characterized by using a description method of pixel-line segment-feature-particle-frame image and storing these structured data in a relational database, and the steps are as follows:

[0008] 1) Obtain the positions of all frames to be measured and control the electron microscope to move the sample stage to the specified position;

[0009] 2) Take a BSE image of each frame: Set the working position of the SEM and the working parameters for taking the BSE image and then take the BSE image;

[0010] 3) Set or calculate the background grayscale range of the image frame, and remove the frame background according to the background grayscale range;

[0011] 4) Binarize the remaining pixels, with the removed background part as 0 and the remaining foreground part as 255;

[0012] 5) According to the connectivity principle, all pixels with a value of 255 and connected left and right are organized into segment objects, i.e. line segments;

[0013] 6) According to the connectivity principle, all segment object sets connected up and down are merged into feature objects to obtain a feature object set;

[0014] 7) Calculate any graphic attribute of the feature object, including ID, diameter, perimeter, Feret diameter or minimum enclosing rectangle, and assign these attributes together with its segment set to a particle object, i.e., a particle object, to obtain a particle object set list, which is the particle feature set of this frame object;

[0015] 8) Store all segment objects in the segment data table, with the foreign key being the particle ID; the particle ID can generate images of foreign key relationships in the database. The system reads the associated data between all tables in the relational database management system through the reading unit;

[0016] 9) Store all particle objects in the particle data table, with the foreign key being the Field ID; the Field ID can generate images of foreign key relationships in the database. The system reads the associated data between all tables in the relational database management system through a reading unit;

[0017] 10) Store the frame graph object in the Field data table, and generate a global association network diagram between tables based on the associated data through the frame graph object, which can intuitively display the primary and foreign key relationships in the relational database management system;

[0018] 11) Control the electron microscope to take the next frame image, and repeat steps 3) to 10) to complete the collection and analysis of all frame image information features and store them in the database.

[0019] Compared with the prior art, the present invention has the following beneficial effects:

[0020] The present invention describes the particles identified after removing the background with this characteristic data structure and then stores it in a database. It has faster data retrieval speed and storage speed than directly storing pictures or other custom binary data structures. Moreover, relying on the powerful query capability of the database SQL language, it can very conveniently implement various data joint queries, which has no negative impact on the overall measurement accuracy, but has a great effect on improving the measurement speed and report retrieval speed, and does not require a large amount of memory and can run very quickly. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a control flow chart of an embodiment of the present invention.

[0022] Figure 2 This is the particle surface structure of the embodiment of the present invention.

[0023] Figure 3 This is the Segment table structure of an embodiment of the present invention.

[0024] Figure 4 This is the field table structure of the embodiment of the present invention. DETAILED DESCRIPTION

[0025] The technical solution of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] Terminology in this article: Pixcel: Pixel, the basic unit of grayscale images, with a value range of 0-255.

[0027] Segment: Line segment. All pixels connected to the left and right constitute a line segment.

[0028] Feature: All the line segments connected up and down constitute a feature.

[0029] Particle: Particle, generally one feature constitutes a particle.

[0030] Field: Field of view. One frame can be captured in one field of view.

[0031] See Figure 1 - Figure 4 , is a schematic diagram of the effect of a database-based image feature description method of the present invention, using a pixel-line segment-feature-particle-frame image description method, and storing these structured data in a relational database, the steps are as follows:

[0032] 1) Obtain the positions of all frames to be measured and control the electron microscope to move the sample stage to the specified position;

[0033] 2) Take a BSE image of each frame: Set the working position of the SEM and the working parameters for taking the BSE image and then take the BSE image;

[0034] 3) Set or calculate the background grayscale range of the image frame, and remove the frame background according to the background grayscale range;

[0035] 4) Binarize the remaining pixels, with the removed background part being 0 and the remaining foreground part being 255; use the Canny edge detection operator based on the binary image to obtain its edge image and find out the pixel grayscale value of 255 in the edge image.

[0036] 5) According to the connectivity principle, all pixels with a value of 255 and connected left and right are organized into segment objects, i.e. line segments;

[0037] 6) According to the connectivity principle, all segment object sets connected up and down are merged into feature objects to obtain a feature object set;

[0038] 7) Calculate any graphic attribute of the feature object, including ID, diameter, perimeter, Feret diameter or minimum enclosing rectangle, and assign these attributes together with its segment set to a particle object, i.e., a particle object, to obtain a particle object set list, which is the particle feature set of this frame object;

[0039] 8) Store all segment objects in the segment data table, with the foreign key being the particle ID; the particle ID can generate the image system from the foreign key relationship in the database through the reading unit to read the associated data of all tables in the relational database management system; the reading unit is used to generate a header base address (HeaderBaseAddress), a vector base address (VectorBaseAddress), multiple headers (Header) and multiple vectors (Vector) according to these image feature descriptions. The header includes a feature number and a feature coordinate (FeatureCoordinates).

[0040] 9) Store all particle objects in the particle data table, with the foreign key being the Field ID; the Field ID can generate images of foreign key relationships in the database. The system reads the associated data between all tables in the relational database management system through a reading unit;

[0041] 10) Store the frame graph object in the Field data table, and generate a global association network diagram between tables based on the associated data through the frame graph object, which can intuitively display the primary and foreign key relationships in the relational database management system;

[0042] 11) Control the electron microscope to take the next frame image, repeat steps 3) to 11), and the acquisition and analysis of all frame image information features are completed and stored in the database. The database is used to continuously write these headers to the header storage unit according to the header base address, and continuously write these vectors to the vector storage unit according to the vector base address.

[0043] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the invention.

Claims

1. A method for describing image features based on a database, characterized in that: Using a description method of pixel Pixcel - line segment Segment - feature Feature - particle Particle - frame field, and storing these structured data in a relational database, the steps are as follows: 1) Obtain the positions of all frames to be measured and control the electron microscope to move the sample stage to the specified position; 2) Take a backscattered electron (BSE) image of each frame: Set the working position of the SEM and the working parameters for taking the backscattered electron (BSE) image, and then take the backscattered electron (BSE) image; 3) Set or calculate the background grayscale range of the image frame, and remove the frame background according to the background grayscale range; 4) Binarize the remaining pixels, set the removed background to 0, and the remaining foreground to 255; 5) According to the connectivity principle, all pixels with a value of 255 and connected to the left and right are organized into segment objects, i.e. line segments; 6) According to the connectivity principle, all segment object sets connected up and down are merged into feature objects to obtain a feature object set; 7) Calculate any graphic attribute of the feature object, including ID, diameter, perimeter, Feret diameter or minimum enclosing rectangle, and assign these attributes together with its segment set to a particle object, i.e., a particle object, to obtain a particle object set list, which is the particle feature set of this frame object; 8) Store all segment objects in the segment data table, whose foreign key is the particle ID; The ID of the particle particle can generate an image of the foreign key relationship in the database. The system reads the associated data of all tables in the relational database management system through a reading unit; the reading unit is used to generate a header base address HeaderBaseAddress, a vector base address VectorBaseAddress, multiple headers Header and multiple vectors Vector according to these image feature descriptions; the header includes a feature number and a feature coordinate FeatureCoordinates; 9) Store all particle objects in the particle data table, whose foreign key is the ID of the frame image Field; The ID of the frame image Field can generate an image of the foreign key relationship in the database. The system reads the associated data between all tables in the relational database management system through the reading unit; 10) The frame graph object is stored in the frame graph Field data table, and a global association network diagram between tables is generated according to the associated data through the frame graph object, which can intuitively display the primary and foreign key relationships in the relational database management system; the database is used to continuously write these headers to the header storage unit according to the header base address, and continuously write these vectors to the vector storage unit according to the vector base address; 11) Control the electron microscope to take the next frame image, and repeat steps 3) to 11) to complete the collection and analysis of all frame image information features and store them in the database.

2. The method for describing image features based on a database according to claim 1, characterized in that: In the step 4), the edge image is obtained by using the Canny edge detection operator according to the binary image, and all pixels with a grayscale value of 255 in the edge image are found.

Images