Abstract
Purpose - This paper aims at providing crystal size distribution (CSD) using photogrammetric and image analysis techniques. A new algorithm is proposed to detect CSDs and a comparison is carried out with conventional watershed segmentation algorithm.Design/methodology/approach - Polished granite plates were prepared to designate the metrics of CSD measurements. There are many important metrics for measurements on CSD. Some of them are orientation, size, position, area, aspect ratio, convexity, circularity, perimeter, convex hull, bounding box, eccentricity, shape, max-min length of CSD’s fitted and corrected ellipse, and population density in a per unit area. Prior to image processing stage, camera calibration was performed to remove the image distortion errors. Image processing techniques were applied to corrected images for detecting the CSD parameters.Findings - The proposed algorithm performed better than watershed segmentation in determination accuracy of size and shape characteristics of the crystal material. Proposed algorithm revealed promising results in identifying CSDs more easily and efficiently.Originality/value - This paper describes CSD of granitic rocks by using automated grain boundary detection methods in polished plate images. Some metrics of CSDs were detected by employing a new procedure. A computer based image analysis technique was developed to measure the CSDs on the granitic rock plates. A validation is done by superimposing digitally detected CSD metrics to original samples.
Purpose - This paper aims at providing crystal size distribution (CSD) using photogrammetric and image analysis techniques. A new algorithm is proposed to detect CSDs and a comparison is carried out with conventional watershed segmentation algorithm.Design/methodology/approach - Polished granite plates were prepared to designate the metrics of CSD measurements. There are many important metrics for measurements on CSD. Some of them are orientation, size, position, area, aspect ratio, convexity, circularity, perimeter, convex hull, bounding box, eccentricity, shape, max-min length of CSD’s fitted and corrected ellipse, and population density in a per unit area. Prior to image processing stage, camera calibration was performed to remove the image distortion errors. Image processing techniques were applied to corrected images for detecting the CSD parameters.Findings - The proposed algorithm performed better than watershed segmentation in determination accuracy of size and shape characteristics of the crystal material. Proposed algorithm revealed promising results in identifying CSDs more easily and efficiently.Originality/value - This paper describes CSD of granitic rocks by using automated grain boundary detection methods in polished plate images. Some metrics of CSDs were detected by employing a new procedure. A computer based image analysis technique was developed to measure the CSDs on the granitic rock plates. A validation is done by superimposing digitally detected CSD metrics to original samples.