Ferrite structure edge extraction method based on microstructural phase diagram of steel

Abstract

A ferrite structure edge extraction method based on a microstructural phase diagram of steel, comprising: performing edge detection on a metallographic image of a low-carbon steel cold rolled sheet; closing grain boundaries; removing redundant boundaries; detecting grain parts of unconnected parts and grain unclosed parts in a labeled image on the basis of a network topology structural diagram, performing grain boundary edge determination, and when the vertex of each unclosed part grain remains unchanged, continuously traversing different grain boundary arcs connected to the unclosed part grain so as to determine whether the unclosed part grain satisfies the characteristics of a completely closed grain, thereby achieving grain reconstruction; and then performing reconstruction detection to finally complete extraction and reconstruction of ferrite structure edges. Thus, ferrite grains of a microstructural phase diagram of steel are accurately reconstructed, two grains connected by a narrow area are effectively segmented, accurate ferrite grains of a low-carbon steel cold rolled sheet can be extracted, and ferrite grain size grading of the low-carbon steel cold rolled sheet is assisted on the basis of the extracted accurate ferrite grains of the low-carbon steel cold rolled sheet.

Description

A method for ferrite microstructure edge extraction based on steel microphase diagram Technical Field

[0001] This invention relates to the field of metallographic image analysis and processing technology for steel, and in particular to a method for ferrite structure edge extraction based on the microscopic phase diagram of steel. Background Technology

[0002] Ferrite is an interstitial solid solution of carbon dissolved in α-Fe, commonly represented by the symbol F. It has a body-centered cubic lattice and a very low carbon solubility, capable of dissolving only 0.0008% of carbon at room temperature, with a maximum carbon solubility of 0.02% at 727℃. Hypoeutectoid austenite precipitates through proeutectoid processes to form ferrite. Ferrite has smooth grain boundaries, few twins or slip lines within the grains, and a light green, shiny color that darkens after deep corrosion. Ferrite in steel exists in platy, massive, acicular, and network forms; this portion of ferrite is called proeutectoid ferrite or structurally free ferrite. Depending on the formation conditions, proeutectoid ferrite has different morphologies, such as equiaxed, intergranular, spindle-shaped, serrated, and acicular. Ferrite is also the matrix of pearlite. In the hot-rolled (normalized) and annealed structures of carbon steel and low-alloy steel, ferrite is the main constituent phase. The composition and structure of ferrite have a significant impact on the processing properties of steel and, in some cases, also affect the service performance of steel.

[0003] The method of this invention mainly targets the edge extraction method of ferrite structure in the microphase diagram of steel proposed in the current GB / T 4335-2013 standard for the determination of ferrite grain size in cold-rolled sheet of low carbon steel. Because the morphology of ferrite is complex, while the ferrite grain boundaries in cold-rolled sheet of low carbon steel are clear and often exist in the form of network, block, and plate, the method is an edge extraction technique for areas with rough grain boundaries.

[0004] Most software currently on the market uses image grayscale intensity to distinguish between different types of steel, but the results are not very good. In addition to considering grayscale intensity, information such as texture pattern, edge direction and morphological pattern must also be considered. Furthermore, some methods are limited to a specific type of steel and have weak transfer learning capabilities.

[0005] Some methods rely entirely on machine learning and deep learning, but their segmentation and training processes are limited by the training set and model, making it impossible to completely solve the problem of ferrite edge extraction.

[0006] Summary of the Invention

[0007] This invention provides a method for ferrite grain edge extraction based on the microscopic phase diagram of steel, which accurately reconstructs the ferrite grains in the microscopic phase diagram of steel, effectively segments two grains connected by narrow regions, and helps to extract accurate ferrite grains of cold-rolled low-carbon steel sheets, thereby assisting in the ferrite grain size rating of cold-rolled low-carbon steel sheets.

[0008] To achieve the above objectives, the present invention employs the following technical solution:

[0009] A method for ferrite microstructure edge extraction based on steel microphase diagram includes the following steps:

[0010] S1. Edge detection: Perform edge detection on the metallographic image of low carbon steel cold-rolled sheet, detect the grain edge information of the ferrite structure image, and generate a grain edge map;

[0011] S2, Closed Grain Boundaries: The grain edge map is sequentially processed by connecting fractured edges, filling voids, and refining grain boundaries to generate a grain processing map;

[0012] S3. Remove redundant boundaries: Use the region growing method to re-label the grains in the image. The outer contour of the labeled region is used as the boundary of the grain. Select unlabeled pixels one by one and label them until all are labeled.

[0013] S4. Grain Reconstruction: Based on the network topology diagram, the image detects unconnected parts and unclosed grain parts of the labeled image, performs grain boundary edge judgment and re-aggregation, and uses a region adjacency graph to connect the nodes to the corresponding regions with adjacent regions using arcs. While keeping each vertex of the grain unchanged, it continuously traverses different connected grain boundary arcs to determine whether it meets the characteristics of a fully closed grain, thus completing the grain reconstruction.

[0014] S5. Perform reconstruction detection to complete the extraction and reconstruction of the ferrite structure edge;

[0015] S6. Generate a ferrite phase boundary map and save the resulting image data.

[0016] Furthermore, the edge detection in step S1 employs a multi-level edge detection Canny algorithm to detect the edge information of the grains.

[0017] Furthermore, step S2, closing the boundary, specifically includes:

[0018] S2.1. Use morphological closure methods to process the edge image, then refine the edges and reconnect the broken edges.

[0019] S2.2. Using a hole-filling method, the area of ​​each hole is calculated, and a threshold is used to fill holes with an area smaller than the threshold.

[0020] S2.3. Refine grain boundaries using a skeletalization method.

[0021] Furthermore, step S4 involves continuously traversing different connected grain boundary arcs, and its traversal model is represented as follows: G = (V, E) (1)

[0022] Where V represents the vertex of the unclosed grain; E represents the grain boundary arc connected to it.

[0023] Furthermore, step S5, which involves reconstruction detection to extract and reconstruct the ferrite structure edge, specifically includes the following steps:

[0024] S5.1. Count the adjacent adjacent grain blocks of each grain to form a network topology diagram;

[0025] S5.2 Calculate the average gray level of each grain;

[0026] S5.3 Detect whether each grain contains a grain boundary;

[0027] S5.4 The inspection conditions are: 1) the average gray level is less than the threshold; 2) the grain does not contain the boundary of the image. Adjacent grains that meet both conditions are judged to belong to the same grain.

[0028] S5.5 Grain Reconstruction: If a case is found that meets the above inspection conditions but is not marked as correctly reconstructed, the grain boundaries of the adjacent grains will be changed to the same one.

[0029] S5.6 Output and save the image results after the above processing.

[0030] Compared with the prior art, the beneficial effects of the present invention are:

[0031] 1) By reconstructing the crystal, using the topological structure and pixel changes of the graph, and through traversal reconstruction and analytical reconstruction, the ferrite grains of the microphase diagram of steel can be accurately reconstructed.

[0032] 2) Effectively separating two grains connected by a narrow region helps to extract accurate ferrite grains from cold-rolled low-carbon steel sheets;

[0033] 3) Accurate ferrite grains of low-carbon steel cold-rolled sheet were extracted to assist in the ferrite grain size rating of low-carbon steel cold-rolled sheet. Attached Figure Description

[0034] Figure 1 is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0035] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings:

[0036] The following embodiments are implemented based on the technical solution of the present invention, providing detailed implementation methods and specific operation processes. However, the scope of protection of the present invention is not limited to the following embodiments. Unless otherwise specified, the methods used in the following embodiments are conventional methods.

[0037]

Example 1

[0038] Figure 1 shows the flowchart of the method of this invention. This invention provides a method for extracting ferrite grain edges based on the microscopic phase diagram of steel. It is used to determine the ferrite grain size of cold-rolled thin sheets of low-carbon steel with a carbon content of less than 0.2%. This method can accurately reconstruct the ferrite grains in the microscopic phase diagram of steel. By using a model to determine grain edges, whether they are closed, whether they are polygonal, and other grain features, the method ultimately completes the extraction and construction of the ferrite grain edges.

[0039] The method of this invention uses metallographic images of cold-rolled low-carbon steel sheets as data for this embodiment. The method of this invention includes the following steps:

[0040] S1. Edge Detection: The multi-level edge detection Canny algorithm is used to detect grain edge information of ferrite microstructure in the metallographic image of low carbon steel cold-rolled sheet. Gaussian filter is used to remove noise in the grain image. Finite difference of first-order partial derivative is used to calculate the gradient magnitude and direction of each pixel in the image. First-order difference convolution template is used to convolve the image. Non-maximum suppression is applied to the gradient magnitude to obtain refined edges. Double threshold algorithm is used to detect and connect edges to generate grain edge map. The preprocessed data is saved locally.

[0041] S2, Closed Grain Boundaries: The grain edge map is sequentially processed by connecting fractured edges, filling voids, and refining grain boundaries to generate a grain processing map;

[0042] S3. Remove redundant boundaries: General edge detection algorithms may detect some false edges. This method uses the region growing method to re-label the grains in the image. By using the points inside the grain as seed points, the region growing method is followed to recursively label the pixels inside the grain. The outer contour of the labeled area is the boundary of the grain. Then, the unlabeled pixels are selected and labeled one by one until all are labeled.

[0043] S4. Grain Reconstruction: Based on the network topology diagram, the image detects unconnected parts and unclosed grain parts of the labeled image, performs grain boundary edge judgment and re-aggregation, and uses a region adjacency graph to connect the nodes to the corresponding regions with adjacent regions using arcs. While keeping each vertex of the grain unchanged, it continuously traverses different connected grain boundary arcs to determine whether it meets the characteristics of a fully closed grain, thus completing the grain reconstruction.

[0044] S5. Perform reconstruction detection to complete the extraction and reconstruction of the ferrite structure edge;

[0045] S6. Generate a ferrite phase boundary map and save the resulting image data.

[0046]

Example 2

[0047] This embodiment is basically the same as Embodiment 1, except that:

[0048] In this embodiment, the specific steps for closed grain boundaries include: determining grain closure based on the grain edge image; for unclosed grain boundaries, processing the edge image using a morphological closure method, employing a 2x2 convolution structure to enlarge the edges in the image, achieving an edge expansion effect; then refining the edges so that broken edges can be reconnected; for discontinuous edges, a hole-filling method is used, calculating the area of ​​each hole and filling holes with areas smaller than the threshold using a threshold; finally, refining the grain boundaries using a skeletalization method to address the problem of excessively thick grain edges caused by the above processing, generating a grain processing image, and saving it locally.

[0049] For grain reconstruction, the specific steps include: based on the network topology diagram, detecting unconnected parts of the labeled image and grain parts of unclosed grains, performing grain boundary edge judgment and re-aggregation, using a region adjacency graph to connect the nodes to the corresponding regions with adjacent regions using arcs, and continuously traversing different connected grain boundary arcs while keeping each vertex of the grain unchanged to determine whether it satisfies the characteristics of a completely closed grain, thus completing the grain reconstruction. The traversal model is represented as follows: G=(V,E) (2)

[0050] Where V represents the vertex of the unclosed grain; E represents the grain boundary arc connected to it.

[0051]

Example 3

[0052] This embodiment is basically the same as embodiment two, except that:

[0053] In this embodiment, the extraction and reconstruction of the ferrite structure edge for reconstruction detection includes the following steps:

[0054] S5.1. Count the adjacent adjacent grain blocks of each grain to form a network topology diagram;

[0055] S5.2 Calculate the average gray level of each grain;

[0056] S5.3 Detect whether each grain contains a grain boundary;

[0057] S5.4 The inspection conditions are: 1) the average gray level is less than the threshold; 2) the grain does not contain the boundary of the image. Adjacent grains that meet both conditions are judged to belong to the same grain.

[0058] S5.5 Grain Reconstruction: If a case is found that meets the above inspection conditions but is not marked as correctly reconstructed, the grain boundaries of the adjacent grains will be changed to the same one.

[0059] S5.6 Output and save the image results after the above processing.

Claims

1. A method for extracting the edge of ferrite structure based on the microphase diagram of iron and steel, characterized in that, Includes the following steps: S1. Edge detection: Perform edge detection on the metallographic image of low carbon steel cold-rolled sheet, detect the grain edge information of the ferrite structure image, and generate a grain edge map; S2, Closed Grain Boundaries: The grain edge map is sequentially processed by connecting fractured edges, filling voids, and refining grain boundaries to generate a grain processing map; S3. Remove redundant boundaries: Use the region growing method to re-label the grains in the image. The outer contour of the labeled region is used as the boundary of the grain. Select unlabeled pixels one by one and label them until all are labeled. S4. Grain Reconstruction: Based on the network topology diagram, the image detects unconnected parts and unclosed grain parts of the labeled image, performs grain boundary edge judgment and re-aggregation, and uses a region adjacency graph to connect the nodes to the corresponding regions with adjacent regions using arcs. While keeping each vertex of the grain unchanged, it continuously traverses different connected grain boundary arcs to determine whether it meets the characteristics of a fully closed grain, thus completing the grain reconstruction. S5. Perform reconstruction detection to complete the extraction and reconstruction of the ferrite structure edge; S6. Generate a ferrite phase boundary map and save the resulting image data.

2. The method for extracting ferrite structure edges based on the microstructure phase diagram of steel according to claim 1, characterized in that, The edge detection in step S1 uses the multi-level edge detection Canny algorithm to detect the edge information of the grains.

3. The method for extracting ferrite structure edges based on the microphase diagram of steel as described in claim 1, characterized in that, The specific steps of closing the boundary in step S2 include: S2.

1. Use morphological closure methods to process the edge image, then refine the edges and reconnect the broken edges. S2.

2. Using a hole-filling method, the area of ​​each hole is calculated, and a threshold is used to fill holes with an area smaller than the threshold. S2.

3. Refine grain boundaries using a skeletalization method.

4. The method for extracting ferrite structure edges based on the microstructure phase diagram of iron and steel according to claim 1, characterized in that, The step S4 involves traversing different connected grain boundary arcs, and its traversal model is represented as follows: G = (V, E) (1) Where V represents the vertex of the unclosed grain; E represents the grain boundary arc connected to it.

5. The method for extracting ferrite structure edges based on the microstructure phase diagram of iron and steel according to claim 1, characterized in that, Step S5, which involves reconstruction detection to extract and reconstruct the ferrite tissue edge, specifically includes the following steps: S5.

1. Count the adjacent adjacent grain blocks of each grain to form a network topology diagram; S5.2 Calculate the average gray level of each grain; S5.3 Detect whether each grain contains a grain boundary; S5.4 The inspection conditions are: 1) the average gray level is less than the threshold; 2) the grain does not contain the boundary of the image. Adjacent grains that meet both conditions are judged to belong to the same grain. S5.5 Grain Reconstruction: If a case is found that meets the above inspection conditions but is not marked as correctly reconstructed, the grain boundaries of the adjacent grains will be changed to the same one. S5.6 Output and save the image results after the above processing.

Images