scholarly journals Machine Learning Versus Human-Developed Algorithms in Image Analysis of Microstructures

2019 ◽  
Vol 1 (1) ◽  
pp. 412-416
Author(s):  
Adam Piwowarczyk ◽  
Leszek Wojnar
Keyword(s):  
Machine Learning ◽  
Grain Size ◽  
Quality Control ◽  
Image Analysis ◽  
Standard Method ◽  
The Other ◽  
Metallic Materials ◽  
Automatic Image Analysis ◽  
Factors Affecting ◽  
Precise Tool

Abstract Automatic image analysis is nowadays a standard method in quality control of metallic materials, especially in grain size, graphite shape and non-metallic content evaluation. Automatically prepared solutions, based on machine learning, constitute an effective and sufficiently precise tool for classification. Human-developed algorithms, on the other hand, require much more experience in preparation, but allow better control of factors affecting the final result. Both attempts were described and compared.

Effects of resolution on the measurement of grain ‘size’

1985 ◽  
Vol 49 (353) ◽  
pp. 539-546 ◽  
Author(s):  
R. Dearnley
Keyword(s):  
Grain Size ◽  
Image Analysis ◽  
Grain Boundary ◽  
Surface Area ◽  
Unit Volume ◽  
Fractal Dimensions ◽  
Size Analysis ◽  
Automatic Image Analysis ◽  
Fine Grained ◽  
Kinetics Of

AbstractMeasurements of fine-grained dolerites by optical automatic image analysis are used to illustrate the effects of magnification and resolution on the values obtained for grain ‘size’, grain boundary length, surface area per unit volume, and other parameters. Within the measured range of optical magnifications (× 26 to × 3571) and resolutions (1.20 × 10−3 cm to 8.50 × 10−6 cm), it is found that the values of all grain parameters estimated by chord size analysis vary with magnification. These results are interpreted in terms of the concepts of ‘fractal dimensions’ introduced by Mandelbrot (1967, 1977). For some comparative purposes the fractal relationships may be of little significance as relative changes of size, surface area, and other parameters can be expressed adequately at given magnification(s). But for many studies, for instance in kinetics of grain growth, the actual diameter or surface area per unit volume is an important dimension. The consequences are disconcerting and suggest that it may be difficult in some instances to specify the ‘true’ measurements of various characteristics of fine-grained aggregates.

scholarly journals ASTM Standards in Microscopy

2005 ◽  
Vol 13 (5) ◽  
pp. 40-43
Author(s):  
John J. Friel
Keyword(s):  
Grain Size ◽  
Image Analysis ◽  
Quantitative Analysis ◽  
Test Methods ◽  
Standard Test ◽  
Automatic Image Analysis ◽  
Reporting Method

ASTM International publishes many standards specifically about microscopes and using microscopy. The value of these standards falls into three distinct categories. First, they can be a quick tutorial on how to perform some operation. E 1508, "Standard Guide to Quantitative Analysis by EDS" is only eight pages long and falls into this category. Second, they can be used to standardize a test and reporting method. The methods described in E 1382, "Standard Test Methods for Determining Grain Size by Semiautomatic and Automatic Image Analysis" are examples of procedures that have been agreed upon for many years.

scholarly journals Optical granulometry of coarse aggregates and sediments

2019 ◽  
pp. 59-71 ◽  
Author(s):  
Maja Matič ◽  
Nejc Bezak ◽  
Mikoš Matjaž
Keyword(s):  
Grain Size ◽  
Image Analysis ◽  
The Other ◽  
Sieve Analysis ◽  
Practical Applications ◽  
Coarse Aggregates ◽  
Other Hand ◽  
Significant Difference ◽  
Gravel Bar ◽  
Relative Differences

In order to determine the granularity of the material, image analysis may be used instead of traditional methods such as sieving. Data on granulometry is important in the water management sector for several practical applications, such as calculation of sediment transport capacity in watercourses, design of hydraulic structures, or modeling of debris flows. The WipFrag and Basegrain programs were tested on the case study of the Belca rockfall and the Sava Dolinka River’s gravel bar near the village of Mojstrana. In both cases, multiple images were taken from different heights and sieve analysis was performed. The results of both programs were compared with the results of the sieving analysis. The results showed that WipFrag yielded more comparable results with the sieving analysis than Basegrain. WipFrag gave slightly better results in the case of the Belca rockfall than in the case of the river gravel bar. The Basegrain program, on the other hand, produced similarly comparable results in both case studies. In most cases, both programs underestimated the grain size compared to the results of the sieving analysis. In some cases, the relative differences were close to 100%. On the other hand, the selected statistical test did not show a statistically significant difference between the results of the sieving analysis and the results of the image analysis in both of the programs. Optimal image analysis results were obtained from images taken from the height of about 1.5-2 m, which means that we suggest that photos of aggregates, fluvial sediments, and erosion material should be taken from a height of approximately ten times the maximum grain size.

Sign in / Sign up

Export Citation Format

Share Document