论文部分内容阅读
南极大陆边缘区域浮冰提取对于南极浮冰变化以及全球变化的研究有重要意义,提出一种基于区域增长图像分割技术的南极大陆边缘浮冰信息自动提取方法.结合浮冰的灰度、轮廓、位置关系等信息进行合并和验证,有效解决图像分割过程中的过度分割以及分割不足的问题.还提出一种基于像素检测的小面积浮冰提取算法,有效提取像素个数小于5的浮冰目标.分别选取了LandSatETM+数据、中国环境减灾卫星HJ1BCCD1数据和MODIS1B数据,进行大范围的实验测试,并对不同数据的实验结果进行对比,结果表明三种数据的浮冰面积提取精度均高于81%,其中ETM+数据和HJ1B数据的提取精度大于90%.实验结果说明面向对象的信息提取方法不仅可以得到整个区域浮冰总的面积和数量,还可以比较准确地得到单个浮冰的详细信息.
The extraction of ice floes in the Antarctic continent edge area is of great significance for the research on the changes and global changes of ice floes in the Antarctic. An automatic extraction method of ice floes information of the Antarctic continent edge based on the regional growth image segmentation technology is proposed. Combining with the grayscale, Location and other information merging and verification, effectively solve the problem of over segmentation and segmentation in the image segmentation process.A small area ice extraction algorithm based on pixel detection is also proposed to effectively extract the ice floes target of less than 5 pixels The data of LandSatETM +, HJ1BCCD1 and MODIS1B of China Environmental Disaster Reduction Satellite were selected to conduct a wide range of experimental tests, and the experimental results of different data were compared, the results show that the extraction accuracy of the ice area of the three data are higher than 81% , In which the extraction accuracy of ETM + data and HJ1B data is greater than 90% .The experimental results show that the object-oriented information extraction method can not only get the total area and number of ice floes in the whole area, but also obtain the detailed information of a single ice floe more accurately.