This work presents a complex solution for determining the average grain size and additional features in super-alloy (Inconel™ 718) micrographs. A crucial point of each automatic grain size measurement system is a reliable segmentation of the grain boundaries using the methods of image processing. This work introduces a novel method for the marker image calculation, which is an essential part of the grayscale image reconstruction. Unlike the methods of grayscale erosion or image subtraction, our method uses the results of contour classification for the goal-directed calculation of the marker image. The grayscale image reconstruction therefore produces an excellent pre-processed image with removed non-grain objects. In addition, the homogeneity inside the grains increases without losing information about the grain boundaries. When the automated grain size measurement using the presented algorithm is compared to the manually evaluated average grain size, we can confirm the acceptance of the proposed method for application in metallurgical praxis.
[1]
Luc Vincent,et al.
Morphological grayscale reconstruction in image analysis: applications and efficient algorithms
,
1993,
IEEE Trans. Image Process..
[2]
Milan Sonka,et al.
Image Processing, Analysis and Machine Vision
,
1993,
Springer US.
[3]
Luc Vincent,et al.
Watersheds in Digital Spaces: An Efficient Algorithm Based on Immersion Simulations
,
1991,
IEEE Trans. Pattern Anal. Mach. Intell..
[4]
Wanda Benesova,et al.
Twin and scratch detection and removal in micrograph images of Inconel 718
,
2006,
Electronic Imaging.
[5]
Edward R. Dougherty,et al.
Hands-on Morphological Image Processing
,
2003
.