海岸线遥感光谱角度—距离相似度生长模型自动化提取

发布时间：2018-04-16 06:13

  本文选题：海岸线 + 角度—距离相似度　； 参考：《遥感学报》2017年03期

【摘要】：海岸线变迁是沿海生态系统变化的重要指示因子,是国家海洋经济关注的重要组成部分。本文通过构建光谱角度—距离相似度模型,解决HJ-1B/IRS红外传感器在海岸线自动化提取应用中的可行性问题,以及当前方法应用于不同时相数据过程中的阈值不稳定性问题,拓展IRS传感器的应用领域和价值。光谱角度—距离相似度模型以多光谱像元归一化辐射值为向量元素,度量不同像元在单位空间距离上的角度相似性,以迭代方式分析水体样本像元与周边八邻域相邻像元的角度—距离相似性,通过相似性约束对水体样本进行区域生长以获取水岸分界线。通道辐射归一化分析表明采用反射率和量化等级最大值归一化的通道值能很好地反映地物随季节的变化;样本相似度分析表明以水体和非水体相似度两倍方差(0.01)为误差的生长阈值(0.98)适用于全时相影像水岸线提取,总体精度优于80%。验证数据分析表明,角度—距离相似度模型阈值稳定、不受时相影响。通过与常用的High Pass卷积滤波、Roberts卷积滤波、Sobel卷积滤波、Laplacian卷积滤波、FFT高通变换和Canny提取结果比对分析表明,High Pass、Laplacian和FFT变换无法应用于IRS传感器,Roberts和Sobel相对来说能较好的识别水陆边缘,Canny在正常噪声条件下也能有效识别水陆边缘。但这些算法在识别水陆边缘线的同时,也将内陆地物边缘线进行了识别,如何将内陆地物边缘线从识别结果中有效去除,是这些方面所面临的重要难点。比较而言,角—距相似度模型能很好的应用于IRS传感器的海岸线提取,对传感器的噪声并不敏感,在B4通道非正常水平噪声条件下也能提取出理想结果,而且后续处理简单,不存在内陆边缘线的干扰问题。光谱角度—距离相似度模型对海岸线识别精度较高、模型参数稳定,能有效地提升IRS传感器在海岸线提取方面应用价值。在实际应用过程中需要避免的是,既覆盖陆地又覆盖海洋的云团会遮挡地物光谱信息,造成海岸线无法有效分离,因此需要对影像数据进行有效的筛选。本文基于遥感影像提取的海岸线只是瞬时水边线,需要进一步结合海岸线的类型以及潮位数据和DEM等数据进行修正得到最终的海岸线。
[Abstract]:Coastline change is an important indicator of coastal ecosystem change and an important component of national marine economy.In this paper, a spectral angle-distance similarity model is constructed to solve the feasibility problem of HJ-1B/IRS infrared sensor in the application of coastline automatic extraction, and the threshold instability problem in the process of applying the current method to different phase data.Expand the application field and value of IRS sensor.The spectral angle-distance similarity model takes the normalized radiation value of multi-spectral pixels as vector element to measure the angular similarity of different pixels in unit space distance.The angle-distance similarity between the water sample pixel and the adjacent pixel is analyzed by iterative method, and the water bank boundary is obtained by using the similarity constraint to grow the water sample in the region.The normalized channel radiation analysis shows that the normalized channel values with the maximum reflectivity and quantization grade can well reflect the variation of ground objects with seasons.The sample similarity analysis shows that the growth threshold (0.98), with the error of water body and non-water similarity double variance 0.01), is suitable for extracting the waterfront of full-time image, and the overall accuracy is better than 80%.The analysis of validation data shows that the threshold of angle-distance similarity model is stable and independent of temporal phase.Comparing the results of High Pass convolution filtering with that of High Pass convolutional filtering and Canny extraction results show that the high pass and FFT transform can not be applied to IRS sensors comparatively well.Canny can also be used to identify the sea and land edges effectively under normal noise conditions.However, these algorithms not only recognize the edge line of land and water, but also recognize the edge line of inland object. How to effectively remove the edge line of inland object from the recognition result is an important difficulty in these aspects.In comparison, the angle-distance similarity model can be applied to the coastline extraction of IRS sensor, and it is not sensitive to the noise of the sensor, and can extract the ideal results under the condition of abnormal horizontal noise in the B4 channel, and the subsequent processing is simple.There is no interference of the interior edge line.The spectral angle-distance similarity model has high accuracy for coastline recognition and stable model parameters. It can effectively enhance the application value of IRS sensor in coastline extraction.What needs to be avoided in the practical application is that the cloud covering both land and sea will block the spectral information of ground objects, which makes the coastline can not be separated effectively, so it is necessary to screen the image data effectively.In this paper, the coastline extracted from remote sensing image is only instantaneous waterfront, which needs to be further combined with the type of coastline, tidal level data and DEM data to get the final coastline.
【作者单位】： 江苏省地质调查研究院;中国科学院遥感与数字地球研究所;环境保护部卫星环境应用中心;首都师范大学资源环境与旅游学院;西安欧亚学院建筑工程分院;黑龙江地理信息工程院;
【基金】：国家自然科学基金(编号:41301388) 国家高分辨率对地观测重大专项项目(编号:11-Y20A32-9001-15/17)~~
【分类号】：P715.7;TP79
，

本文编号：1757617