Classification of Bainitic Structures Using Textural Parameters and Machine Learning Techniques

Müller, Martín; Britz, Dominik; Ulrich, Laura; Staudt, Thorsten; Mücklich, Frank

doi:10.3390/met10050630

Cited by 33 publications

(27 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the present work, we refer to these microstructures as polygonal ferrite (PF, featureless gray regions), degenerated pearlite (DP, arrow highlighted bright region) and degenerated upper bainite (DUB, ellipse highlighted regions with fine substructure). The designations polygonal ferrite, degenerated pearlite and degenerated upper bainite and their abbreviations PF, DP and DUB are used as proposed by Zajac et al [19] and have been frequently adapted by many researchers such as [11,[31][32][33][34]. Figures 3a and b show that the PF and DP volume fractions decrease as b increases from 0.5 K/s to 1 K/s.…”

Section: Microstructural Features Observed By Semmentioning

confidence: 99%

Effect of cooling rate on the microstructure and mechanical properties of a low-carbon low-alloyed steel

Wang

Cao

et al. 2021

J Mater Sci

View full text Add to dashboard Cite

Heavy plate steels with bainitic microstructures are widely used in industry due to their good combination of strength and toughness. However, obtaining optimal mechanical properties is often challenging due to the complex bainitic microstructures and multiple phase constitutions caused by different cooling rates through the plate thickness. Here, both conventional and advanced microstructural characterization techniques which bridge the meso- and atomic-scales were applied to investigate how microstructure/mechanical property-relationships of a low-carbon low-alloyed steel are affected by phase transformations during continuous cooling. Mechanical tests show that the yield strength increases monotonically when cooling rates increase up to 90 K/s. The present study shows that this is associated with a decrease in the volume fraction of polygonal ferrite (PF) and a refinement of the substructure of degenerated upper bainite (DUB). The fine DUB substructures feature C-rich retained austenite/martensite-austenite (RA/M-A) constitutes which decorate the elongated micrograin boundaries in ferrite. A further increase in strength is observed when needle-shaped cementite precipitates form during water quenching within elongated micrograins. Pure martensite islands on the elongated micrograin boundaries lead to a decreased ductility. The implications for thick section plate processing are discussed based on the findings of the present work.

show abstract

Section: Microstructural Features Observed By Semmentioning

confidence: 99%

Effect of cooling rate on the microstructure and mechanical properties of a low-carbon low-alloyed steel

Wang

Cao

et al. 2021

J Mater Sci

View full text Add to dashboard Cite

show abstract

“…Furthermore, the suggested approach could also be attempted using other edge detectors or other operators like Fourier and Hough transform. By adding other features to the convolution operator, e.g., textural features as described in [36], the segmentation could be expanded to simultaneously perform a microstructure classification.…”

Section: Resultsmentioning

confidence: 99%

“…Regarding further analyses of the segmented image, using the binary image as a mask to the original image allows a separate analysis of the lath-like bainite, e.g., to further distinguish upper and degenerate upper bainite. This could be accomplished by calculating textural parameters from these regions, as suggested by Müller et al [36]. In addition, granular and lath-like regions can be analyzed independently, e.g., morphological parameters of the carbon-rich second phase such as size distributions, particle shapes and mean free distance between particles for microstructure-properties correlations can be calculated for each type of bainite instead of calculating it for the entire image.…”

mentioning

confidence: 99%

Segmentation of Lath-Like Structures via Localized Identification of Directionality in a Complex-Phase Steel

Müller

Stanke²,

Sonntag³

et al. 2020

Metallogr. Microstruct. Anal.

Self Cite

View full text Add to dashboard Cite

In this work, a segmentation approach based on analyzing local orientations and directions in an image, in order to distinguish lath-like from granular structures, is presented. It is based on common image processing operations. A window of appropriate size slides over the image, and the gradient direction and its magnitude inside this window are determined for each pixel. The histogram of all possible directions yields the main direction and its directionality. These two parameters enable the extraction of window positions which represent lath-like structures, and procedures to join these positions are developed. The usability of this approach is demonstrated by distinguishing lath-like bainite from granular bainite in so-called complex-phase steels, a segmentation task for which automated procedures are not yet reported. The segmentation results are in accordance with the regions recognized by human experts. The approach’s main advantages are its use on small sets of images, the easy access to the segmentation process and therefore a targeted adjustment of parameters to achieve the best possible segmentation result. Thus, it is distinct from segmentation using deep learning which is becoming more and more popular and is a promising solution for complex segmentation tasks, but requires large image sets for training and is difficult to interpret.

show abstract

“…Approaches to quantify the separate microstructure constituents using EBSD individually have been reported 11,12 . Müller et al 13 developed a procedure to segment lath-shaped bainite in CP steel micrographs consisting of lath-shaped and granular bainite by analyzing the microstructure constituents' directionality. Bulgarevich et al 14 used a trainable segmentation with a random forest classifier to segment ferrite, pearlite, and bainite in light optical micrographs of three-phase steels.…”

Section: Introductionmentioning

confidence: 99%

“…As opposed to these works, supporting correlative electron backscatter diffraction (EBSD) information is used in the LOM and SEM annotation procedure to lay an appropriate foundation for learning. Moreover, the aforementioned conventional CV or ML approaches require complex image processing pipelines and elaborate feature engineering to render predictions robust against variances 13 . In contrast, the applied DL methods are directly based on input and target output image pairs (representation learning).…”

Section: Introductionmentioning

confidence: 99%

A Deep Learning Approach for Complex Microstructure Inference

Durmaz

Müller

Lei

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Automated, reliable, and objective microstructure inference from micrographs is an essential milestone towards a comprehensive understanding of process-microstructure-property relations and tailored materials development. However, such inference, with the increasing complexity of microstructures, requires advanced segmentation methodologies. While deep learning (DL), in principle, offers new opportunities for this task, an intuition about the required data quality and quantity and an extensive methodological DL guideline for microstructure quantification and classification are still missing. This, along with a lack of open-access data sets and the seemingly intransparent decision-making process of DL models, hampers its breakthrough in this field. We address all aforementioned obstacles by a multidisciplinary DL approach, devoting equal attention to specimen preparation, contrasting, and imaging. To this end, we train distinct U-Net architectures with 30–50 micrographs of different imaging modalities and corresponding EBSD-informed annotations. On the challenging task of lath-bainite segmentation in complex-phase steel, we achieve accuracies of 90% rivaling expert segmentations. Further, we discuss the impact of image context, pre-training with domain-extrinsic data, and data augmentation. Network visualization techniques demonstrate plausible model decisions based on grain boundary morphology and triple points. As a result, we resolve preconceptions about required data amounts and interpretability to pave the way for DL's day-to-day application for microstructure quantification.

show abstract

Classification of Bainitic Structures Using Textural Parameters and Machine Learning Techniques

Cited by 33 publications

References 26 publications

Effect of cooling rate on the microstructure and mechanical properties of a low-carbon low-alloyed steel

Effect of cooling rate on the microstructure and mechanical properties of a low-carbon low-alloyed steel

Segmentation of Lath-Like Structures via Localized Identification of Directionality in a Complex-Phase Steel

A Deep Learning Approach for Complex Microstructure Inference

Contact Info

Product

Resources

About