【摘要】 利用威德尔海区域2016年的Sentinel-1ASAR影像数据,采用"双冒泡法"的sigma-on-mu探测器探测冰山边缘区域,并通过对边缘像元进行交换排序和凸显最大像元的方式识别冰山。以人工识别法为基础,通过与自动识别法的对比,定量地分析了"双冒泡法"的识别偏差。研究结果表明,"双冒泡法"识别的冰山线性尺寸和面积等信息中纵横向最大长度分别为24.52km和11.16km;面积为220.833 6km2;单体识别偏差率为2.87%,低于自动识别法(7.5%);平均偏差率为2.48%,亦低于自动识别法(7.27%)。同时,基于"双冒泡法",提出了较小冰山边界的手动分离法(像元≤100),与自动识别方法相比,该方法的手动分离以具体的像元边界为基准,提高了对较小冰山的识别精度。更多还原

【Abstract】 The sigma-on-mu detector of the " Double Bubbling Method" was applied to the Sentinel-1 ASAR image of the Weddell Sea in 2016,to detect the iceberg edge areas,and then the iceberg was identified through exchanging the edge pixels and highlighting the largest pixels.The identification bias was quantitatively analyzed by comparison with the automatic identification method.The results show that the longitudinal and lateral lengths of the iceberg identified by the " Double Bubbling Method" are 24.52 km and11.16 km,respectively,with the area of 220.833 6 km2.The single and average identification deviation rates are 2.87%and 2.48%,respectively,which are lower than those of the automatic identification method(7.5%and 7.27%).Based on the " Double Bubbling Method",a manual separation method for small iceberg boundaries(pixel number less than 100)was proposed,compared with the automatic recognition method,the manual separation of the method is based on the specific pixel boundary,which improves the identification accuracy of the smaller icebergs.更多还原