Segmenting images with complex textures by using hybrid algorithm

Han, Yuexing; Lai, Chuanbin; Wang, Bing; Gu, Hui

doi:10.1117/1.jei.28.1.013030

Cited by 10 publications

(8 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The need for such automation is well motivated from the perspective of material science [30][31][32][33][34][35] , as well as from the perspective of microscopic imaging in general 36,37 . In relatively-simple cases, classic algorithms for image pre-processing, pixel classification, region extraction, edge detection and so forth, occasionally combined with data mining methods, proved to yield moderate to good results [38][39][40][41][42][43] . However, as the image complexity and resolution increases, and the image domain varies, the error rate in the needed tasks using those methods increases as well, thus demanding human expertise in tuning said algorithms 38 .…”

Section: Previous Workmentioning

confidence: 99%

“…In relatively-simple cases, classic algorithms for image pre-processing, pixel classification, region extraction, edge detection and so forth, occasionally combined with data mining methods, proved to yield moderate to good results [38][39][40][41][42][43] . However, as the image complexity and resolution increases, and the image domain varies, the error rate in the needed tasks using those methods increases as well, thus demanding human expertise in tuning said algorithms 38 . For example, in cases where segmentation depends on the color histogram or some hyperparameters of other edge detection algorithms, the expert is required to tune the algorithm until reaching the desired result 38 .…”

Section: Previous Workmentioning

confidence: 99%

“…However, as the image complexity and resolution increases, and the image domain varies, the error rate in the needed tasks using those methods increases as well, thus demanding human expertise in tuning said algorithms 38 . For example, in cases where segmentation depends on the color histogram or some hyperparameters of other edge detection algorithms, the expert is required to tune the algorithm until reaching the desired result 38 . Nevertheless, as many alloys -even of the same materials and under similar conditions -exhibit extreme variance in colors, boundaries, structures and in their corresponding distributions.…”

Section: Previous Workmentioning

confidence: 99%

See 2 more Smart Citations

An End-to-End Computer Vision Methodology for Quantitative Metallography

Rusanovsky¹,

Beeri²,

Oren³

2021

Preprint

View full text Add to dashboard Cite

Metallography is crucial for a proper assessment of material's properties. It involves mainly the investigation of spatial distribution of grains and the occurrence and characteristics of inclusions or precipitates. This work presents an holistic artificial intelligence model for Anomaly Detection that automatically quantifies the degree of anomaly of impurities in alloys. We suggest the following examination process: (1) Deep semantic segmentation is performed on the inclusions (based on a suitable metallographic database of alloys and corresponding tags of inclusions), producing inclusions masks that are saved into a separated database. (2) Deep image inpainting is performed to fill the removed inclusions parts, resulting in 'clean' metallographic images, which contain the background of grains. (3) Grains' boundaries are marked using deep semantic segmentation (based on another metallographic database of alloys), producing boundaries that are ready for further inspection on the distribution of grains' size. (4) Deep anomaly detection and pattern recognition is performed on the inclusions masks to determine spatial, shape and area anomaly detection of the inclusions. Finally, the system recommends to an expert on areas of interests for further examination. The performance of the model is presented and analyzed based on few representative cases. Although the models presented here were developed for metallography analysis, most of them can be generalized to a wider set of problems in which anomaly detection of geometrical objects is desired. All models as well as the data-sets that were created for this work, are publicly available at 1 .

show abstract

Section: Previous Workmentioning

confidence: 99%

Section: Previous Workmentioning

confidence: 99%

Section: Previous Workmentioning

confidence: 99%

See 1 more Smart Citation

An End-to-End Computer Vision Methodology for Quantitative Metallography

Rusanovsky¹,

Beeri²,

Oren³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…The mentioned-above improvements allow our model to focus on the lost high-frequency information when transferring high-resolution information across the network, and enhance feature interpretability in decoders with the aid of spatial-channel attentions. (2) We propose SASAPD based on SAPD to detect constituents in multi-phase metallographic images. It improves soft-weighting scheme by reranking anchor points with powerful feature representation, and self-adaptively selects the reasonable features for each instance from attention-aware pyramid levels.…”

Section: Introductionmentioning

confidence: 99%

“…For image segmentation, the models roughly range from early rule-based and learning-based methods to recent deep-learning methods. The rule-based methods could offer accurate segmentation results, but often involve the prior rules, which greatly limit the generality in other applications [ 1 , 2 ]. The learning-based methods work based on handcrafted features, but they always suffer from the sensitively to constructed features for metallographic images with complex features [ 3 , 4 , 5 ].…”

Section: Introductionmentioning

confidence: 99%

Quantitative Analysis of Metallographic Image Using Attention-Aware Deep Neural Networks

Zhang

et al. 2020

Sensors

View full text Add to dashboard Cite

As a detection tool to identify metal or alloy, metallographic quantitative analysis has received increasing attention for its ability to evaluate quality control and reveal mechanical properties. The detection procedure is mainly operated manually to locate and characterize the constitution in metallographic images. The automatic detection is still a challenge even with the emergence of several excellent models. Benefiting from the development of deep learning, with regard to two different metallurgical structural steel image datasets, we propose two attention-aware deep neural networks, Modified Attention U-Net (MAUNet) and Self-adaptive Attention-aware Soft Anchor-Point Detector (SASAPD), to identify structures and evaluate their performance. Specifically, in the case of analyzing single-phase metallographic image, MAUNet investigates the difference between low-frequency and high-frequency and prevents duplication of low-resolution information in skip connection used in an U-Net like structure, and incorporates spatial-channel attention module with the decoder to enhance interpretability of features. In the case of analyzing multi-phase metallographic image, SASAPD explores and ranks the importance of anchor points, forming soft-weighted samples in subsequent loss design, and self-adaptively evaluates the contributions of attention-aware pyramid features to assist in detecting elements in different sizes. Extensive experiments on the above two datasets demonstrate the superiority and effectiveness of our two deep neural networks compared to state-of-the-art models on different metrics.

show abstract