Segmentation Performance Analysis of the Otsu Algorithm for Spent Nuclear Fuel Cladding Image According to Morphological Operations

Jee A Baik, Jun Won Choi, Jung Jin Kim^*

Keimyung University, 1095, Dalgubeol-daero, Dalseo-gu, Daegu 42601, Republic of Korea

Received 2024-03-27 ; Revised 2024-06-18 ; Approved 2024-07-30

Abstract

Hydride analysis is required to assess the mechanical integrity of spent nuclear fuel cladding. Image segmentation, which is a hydride analysis method, is a technique that can analyze the orientation and distribution of hydrides in cladding images of spent nuclear fuels. However, the segmentation results varied according to the image preprocessing. Inaccurate segmentation results can make hydride difficult to analyze. This study aims to analyze the segmentation performance of the Otsu algorithm according to the morphological operations of cladding images. Morphological operations were applied to four different cladding images, and segmentation performance was quantitatively compared using a histogram, between-class variance, and radial hydride fraction. As a result, this study found that morphological operations can induce errors in cladding images and that appropriate combinations of morphological operations can maximize segmentation performance. This study emphasizes the importance of image preprocessing methods, suggesting that they can enhance the accuracy of hydride analysis. These findings are expected to contribute to the advancements in integrity assessment of spent nuclear fuel cladding.

Key Words : Hydride, Image segmentation, Morphological operations, Otsu algorithm, Spent nuclear fuel