쿼드트리 분할영상의 국부이진화를 통한 팩아이스 집적도 및 크기 분포도 측정 알고리즘

Abstract

Recently, many research works on the icebreaking vessels have been published as the possibility of passing Arctic routes has been increasing. The model ship test on the pack ice model in the ice basin is actively carried out as a way to investigate the performance of icebreaking vessels. In this test, The concentration as well as the individual area of pack ice are important since they directly affect the performance. However, it is difficult to measure the concentration because not only the pack ice has uneven shape but also it keeps floating around in the basin. In this paper, an algorithm to identify the concentration of pack ice is introduced. From a digital image of pack ice obtained in the ice basin, the goal is to measure the area of pack ice using an image processing technique. Instead of the general global binarization that yields numerical errors in this problem, a local binarization technique, coupled with image subdivision based on the quadtree structure, is developed. Also the individual area and size distribution of each pack ice patch is calculated using the edge detection and the watershed technique. The obtained results are compared with the manually measured data to prove its accuracy.