Abstract Numerous studies on icebreaking vessels have been reported recently owing to a potential increase in the number of passable Arctic routes. Model ship tests of pack ice models in ice basins have been actively conducted to investigate the performance of different icebreakers. The estimation of the pack ice concentration is critical as pack ice directly affects the performance of such experiment. In this paper, an algorithm that automatically determines the concentration of pack ice is proposed. This study aimed to measure the area of pack ice, based on digital images captured in an ice basin, using an image processing technique. A local binarization technique, coupled with image subdivision based on a quadtree structure, was developed. Independent thresholds were calculated for the subdivided images and rapid image separation was achieved. To demonstrate the accuracy of the proposed method, the concentration results obtained with the developed algorithm were compared with the manually measured data, with both results and data showing good agreement. The main advantage of the developed algorithm is that it facilitates the near-real-time estimation of the pack ice concentration from captured digital images.

中文翻译：

---

使用直方图峰值分析和图像细分估计浮冰浓度

摘要 由于可通行的北极航线数量的潜在增加，最近报道了大量关于破冰船的研究。已积极开展冰盆浮冰模型的模型船试验，以研究不同破冰船的性能。浮冰浓度的估计至关重要，因为浮冰直接影响此类实验的性能。在本文中，提出了一种自动确定浮冰浓度的算法。本研究旨在根据在冰盆中捕获的数字图像，使用图像处理技术测量浮冰的面积。开发了一种局部二值化技术，结合基于四叉树结构的图像细分。计算细分图像的独立阈值并实现快速图像分离。为了证明所提出方法的准确性，将使用开发算法获得的浓度结果与手动测量的数据进行了比较，结果和数据均显示出良好的一致性。所开发算法的主要优点是它有助于从捕获的数字图像中近乎实时地估计浮冰浓度。