10:45 〜 12:15
[SGC38-P04] 全自動粒子画像解析と計算シミュレーションを用いた土壌粒子の形態分析における評価粒子数の妥当性に関する調査(3)
キーワード：粒子径、粒子形状、画像解析
[Introduction]
Soil particle morphology aspects, such as particle shape and size are well known possible to suggest for prediction of various bulk characteristics on mechanics behavior of the grounds as microscopic insight. To investigate this matter, Image Analysis (IA) using a manually microscopic technic is commonly used. However, primary challenges of a conventional IA is the near impossible of measuring a statistically significant number of particles. Typically, realistic particle morphology in natural field should has aspects as broadly distribution. Regarding ISO and JIS statements suggested to measuring over than tens of thousands of particles. Furthermore, reason of importance to concern of a deviation model is as well as supported from that typical model of particle size distribution should follow with a lognormal distribution. Recent advances in computer technology have allowed for the development of a new automated particle image analysis (APIA) approach that uses digital imaging technology. This technology is based on allowed acquisition of a calculation of binarized (2D) particle projection images on each individual particle from on over than ten thousand particles within at least few hours. Hence, these approaches are allowed for the calculation of the various morphological parameters using graphical comparison, such as various distribution graph, such as frequency curve and cumulant curve. However, even if recently progress as mentioned above, it is still only limited suggestion for enough particles to determine distribution on realistic particles which has various aspects.
Our group has been continuously investigate for number of particle on IA . This report will discuss focused on shape aspect especially by APIA combined with using statistically modeling and simulation.
[Method]
Milled Silica sand were used as model samples. APIA analyses were conducted on a Morphologi G3SE as automated image analysis system (Malvern Instruments, Worcestershire, UK). Sample was subsequently dispersed with an SDU using a short duration pulse of compressed air. Measurements were collected automatically using standard operating procedures (SOPs), which clearly define the software and hardware settings used during the measurement process. The measurement sample was dispersed on a glass plate, which was used as a sample carrier to minimize environmental exposure within the enclosed sample chamber unit.
A computer simulation was carried out mainly Scilab platform with implemented own programming to calculate various numerical model, MS Excel used as supportive.
[Simulation Design]
In this study, we adopted a method of comparing the circularity obtained by directly changing the shape using a simulation with the measured value (Fig.1). Specifically, a polar coordinate function is used to give roughness to the end face with random numbers. The number of vertices can be defined computationally. The particle shape distribution and each characteristic value were compared with the number of occurrences of figures as the number of particles. To minimize the influence of the particle size, the actual measurement data was verified by limiting the particle size.
Soil particle morphology aspects, such as particle shape and size are well known possible to suggest for prediction of various bulk characteristics on mechanics behavior of the grounds as microscopic insight. To investigate this matter, Image Analysis (IA) using a manually microscopic technic is commonly used. However, primary challenges of a conventional IA is the near impossible of measuring a statistically significant number of particles. Typically, realistic particle morphology in natural field should has aspects as broadly distribution. Regarding ISO and JIS statements suggested to measuring over than tens of thousands of particles. Furthermore, reason of importance to concern of a deviation model is as well as supported from that typical model of particle size distribution should follow with a lognormal distribution. Recent advances in computer technology have allowed for the development of a new automated particle image analysis (APIA) approach that uses digital imaging technology. This technology is based on allowed acquisition of a calculation of binarized (2D) particle projection images on each individual particle from on over than ten thousand particles within at least few hours. Hence, these approaches are allowed for the calculation of the various morphological parameters using graphical comparison, such as various distribution graph, such as frequency curve and cumulant curve. However, even if recently progress as mentioned above, it is still only limited suggestion for enough particles to determine distribution on realistic particles which has various aspects.
Our group has been continuously investigate for number of particle on IA . This report will discuss focused on shape aspect especially by APIA combined with using statistically modeling and simulation.
[Method]
Milled Silica sand were used as model samples. APIA analyses were conducted on a Morphologi G3SE as automated image analysis system (Malvern Instruments, Worcestershire, UK). Sample was subsequently dispersed with an SDU using a short duration pulse of compressed air. Measurements were collected automatically using standard operating procedures (SOPs), which clearly define the software and hardware settings used during the measurement process. The measurement sample was dispersed on a glass plate, which was used as a sample carrier to minimize environmental exposure within the enclosed sample chamber unit.
A computer simulation was carried out mainly Scilab platform with implemented own programming to calculate various numerical model, MS Excel used as supportive.
[Simulation Design]
In this study, we adopted a method of comparing the circularity obtained by directly changing the shape using a simulation with the measured value (Fig.1). Specifically, a polar coordinate function is used to give roughness to the end face with random numbers. The number of vertices can be defined computationally. The particle shape distribution and each characteristic value were compared with the number of occurrences of figures as the number of particles. To minimize the influence of the particle size, the actual measurement data was verified by limiting the particle size.