Search

KR-20260064792-A - Apparatus and method for phase segmentation of steel microstructure

KR20260064792AKR 20260064792 AKR20260064792 AKR 20260064792AKR-20260064792-A

Abstract

The present invention relates to a device and method for separating phases of a metal's microstructure. A device for distinguishing microstructure phases of a metal according to the present invention comprises: an image extraction unit that extracts one or more filtered images using one or more filters on an image of a microstructure phase of a metal; and a phase distinguishing unit that distinguishes the microstructure phases of the metal based on the one or more filtered images.

Inventors

  • 김영재
  • 김현기
  • 강민우
  • 이충안

Assignees

  • 현대자동차주식회사
  • 기아 주식회사

Dates

Publication Date
20260508
Application Date
20241029

Claims (18)

  1. As a device for distinguishing phases of the microstructure of a metal, An image extraction unit that extracts one or more filtered images using one or more filters on an image of a microstructure phase of a metal; and A phase separator comprising a phase separator that distinguishes the microstructure phases of the metal based on one or more of the above-mentioned filtering images, Separation device for the microstructure phases of a metal.
  2. In paragraph 1, A noise removal unit further comprising removing noise included in one or more of the filtered images, Separation device for the microstructure phases of a metal.
  3. In paragraph 2, The above one or more filtered images include at least one of an image quality (IQ) image, a phase image extracted from a diffraction pattern, a Kernel Average Misorientation (KAM) image, a deviation image in the Kurdjumov-Sachs (KS) orientation relationship, and a masking image. Separation device for the microstructure phases of a metal.
  4. In paragraph 3, The image extraction unit calculates the smallest angular deviation among the angular deviations between the microstructures of the metal, and generates a deviation image in the KS orientation relationship corresponding to the angular deviation. Separation device for the microstructure phases of a metal.
  5. In paragraph 4, The above noise removal unit classifies noise included in the one or more filtered images using a noise classification model that classifies the types of noise in the one or more filtered images. Separation device for the microstructure phases of a metal.
  6. In paragraph 5, The above noise removal unit removes the noise included in the one or more filtered images using noise removal logic corresponding to the type of noise classified above. Separation device for the microstructure phases of a metal.
  7. In paragraph 6, The types of noise included in the filtered image above include at least one of grain boundaries, crystal structure distortion due to dislocations, crystal structure distortion due to precipitation, Hough transform recognition errors, scratches, and contamination. Separation device for the microstructure phases of a metal.
  8. In Paragraph 7, The microstructural phase of the above metal is at least one of ferrite, bainite, and martensite, Separation device for the microstructure phase inside.
  9. In paragraph 8, The above upper section is, A first CNN (Convolutional Neural Network) for extracting a first feature value for at least one of the morphological texture and defect density of the metal microstructure determined at each pixel of the above clarity image; A second CNN that extracts a second feature value for a crystal structure detected at each pixel of the above-mentioned image; A third CNN for extracting a third feature value for the potential density detected at each pixel of the above KAM image; and It includes a fourth CNN that extracts a fourth feature value for the angular deviation between the microstructures of the metal in the above deviation image, and The above-mentioned phase classification unit calculates a probability value for each microstructure phase of the metal based on at least one of the first to fourth feature values, and classifies the microstructure phase with the highest probability value as the microstructure phase corresponding to the Region of Interest (ROI) of the masking image. Separation device for the microstructure phases of a metal.
  10. As a method for distinguishing microstructural phases of a metal, A step of extracting one or more filtered images using one or more filters on an image of the microstructure of a metal; and A step comprising distinguishing the microstructure phase of the metal based on one or more of the filtered images, Method for classifying the microstructural phases of a metal.
  11. In Paragraph 10, Prior to the step of distinguishing the microstructural phases of the above metal, A method further comprising the step of removing noise included in one or more filtered images, Method for classifying the microstructural phases of a metal.
  12. In Paragraph 11, The above one or more filtered images include at least one of an image quality (IQ) image, a phase image extracted from a diffraction pattern, a Kernel Average Misorientation (KAM) image, a deviation image in the Kurdjumov-Sachs (KS) orientation relationship, and a masking image. Method for classifying the microstructural phases of a metal.
  13. In Paragraph 12, The step of extracting the filtered image above is, A process for calculating the smallest angular deviation among the angular deviations between the microstructures of the metal; and A process including generating a deviation image in the KS orientation relationship corresponding to the above angle deviation, Method for classifying the microstructural phases of a metal.
  14. In Paragraph 13, The step of removing the above noise is, A process comprising classifying noise contained in one or more filtered images using a noise classification model that classifies the types of noise in one or more filtered images, Method for classifying the microstructural phases of a metal.
  15. In Paragraph 14, The step of removing the above noise is, A process further comprising removing the noise included in the one or more filtered images using noise removal logic corresponding to the types of noise classified above, Method for classifying the microstructural phases of a metal.
  16. In paragraph 15, The types of noise included in the filtered image above include at least one of grain boundaries, crystal structure distortion due to dislocations, crystal structure distortion due to precipitation, Hough transform recognition errors, scratches, and contamination. Method for classifying the microstructural phases of a metal.
  17. In Paragraph 16, The microstructural phase of the above metal is at least one of ferrite, bainite, and martensite, Method for classifying the microstructural phases of a metal.
  18. In Paragraph 17, The step of distinguishing the microstructural phases of the above metal is, A process of extracting a first feature value for at least one of the morphological texture and defect density of the metal microstructure identified at each pixel of the above clarity image; A process of extracting a second feature value for a crystal structure detected at each pixel of the above-mentioned image; A process of extracting a third feature value for the potential density detected at each pixel of the above KAM image; and It includes at least one of the processes for extracting a fourth feature value for the angular deviation between the microstructures of the metal of the above deviation image, and The step of distinguishing the microstructural phases of the above metal is, A process of calculating a probability value for each microstructure phase of a metal based on at least one of the first to fourth feature values; and A process further comprising classifying the microstructure image with the highest probability value as the microstructure image corresponding to the Region of Interest (ROI) of the masked image. Method for classifying the microstructural phases of a metal.

Description

Apparatus and method for phase segmentation of steel microstructure The present invention relates to an apparatus and method for distinguishing microstructure phases of a metal, and more specifically, to an apparatus and method for distinguishing microstructure phases of a metal, such as ferrite, bainite, and martensite, of a multiphase composite structure steel. The application of the hot stamping method in the manufacture of automobile frames is on the rise. Here, the steel sheets used for hot stamping require high formability, such as Advanced Multi-Phase (AMP) steel, which is a third-generation steel sheet. Accordingly, third-generation steel sheets utilize the TRIP (Transformation Induced Plasticity) phenomenon to overcome the low formability, a disadvantage of conventional steel. TRIP-based steel sheets consist of a multi-phase microstructure composed of ferrite, bainite, martensite, and austenite. Since the microstructure of steel sheets is closely related to formability and impact performance, accurate differentiation and quantitative analysis of the microstructural phases are essential. Accordingly, conventionally, an EBSD phase separation technique was used to separate sections by phase by converting specific data in the EBSD (Electron Back Scatter Diffraction) measurement area into histograms and spectroscopic data. However, conventional technology had the problem that users had to manually determine and distinguish the microstructure phases from images and distribution maps. Additionally, there was a problem where the phase distinction results were inaccurate due to unclear regions in the images. FIG. 1 is a diagram showing the configuration of a device for distinguishing the microstructure phases of a metal according to one embodiment of the present invention. FIG. 2 is a flowchart of a method for distinguishing microstructure phases of a metal according to one embodiment of the present invention. Figure 3 is a detailed flowchart of step S210 of Figure 2. Figure 4 is a detailed flowchart of step S220 of Figure 2. Figure 5 is a detailed flowchart of step S230 of Figure 2. FIG. 6 is a graph illustrating an IQ (Image Quality) map image and an IQ profile according to one embodiment of the present invention. FIG. 7 is an image showing a region of interest according to one embodiment of the present invention and an image showing a phase of the adjacent region of the grain of interest. FIG. 8 is an image showing a region of interest and a KAM image according to one embodiment of the present invention. FIG. 9 is an image showing a region of interest according to one embodiment of the present invention and an image showing the degree of deviation from the KS orientation relationship. FIG. 10 is an image showing a region of interest and an image with masking processing applied outside the region of interest according to one embodiment of the present invention. FIG. 11 is an EBSD image according to one embodiment of the present invention, illustrating images before and after removing noise caused by grain boundaries in the image. Figure 12 compares the accuracy of phase separation prediction values on the microstructure by manual, first-generation model, and second-generation model. Figure 13 compares the accuracy of phase separation prediction values on the microstructure by manual, first-generation model, and second-generation model. Figure 14 is a graph showing the IQ overlap of the ferrite and bainite metals used as comparison subjects in Figure 13. Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the attached drawings. Identical or similar components regardless of drawing symbols are assigned the same reference number, and redundant descriptions thereof will be omitted. The suffixes "module" and "part" used for components in the following description are assigned or used interchangeably solely for the ease of drafting the specification and do not inherently possess distinct meanings or roles. Furthermore, in describing the embodiments disclosed in this specification, if it is determined that a detailed description of related prior art could obscure the essence of the embodiments disclosed in this specification, such detailed description will be omitted. Additionally, the attached drawings are intended only to facilitate understanding of the embodiments disclosed in this specification; the technical concept disclosed in this specification is not limited by the attached drawings, and it should be understood that they include all modifications, equivalents, and substitutions that fall within the spirit and technical scope of the invention. Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but said components are not limited by said terms. These terms are used solely for the purpose of distinguishing one component from another. When it is stated that one component is "connected" or "connected" to