论文部分内容阅读
单独利用遥感影像光谱信息进行近海水产养殖信息提取时,养殖水体与自然水体易混淆,而单独利用遥感影像纹理信息提取近海水域水产养殖信息时,单一大块养殖水体与自然水体又难以区分。针对上述问题,利用OLI影像数据,提出了一种综合遥感影像光谱与纹理信息进行水产养殖信息提取的方法。首先,在对研究区养殖水体类型进行光谱特征分析的基础上,采用主成分变换方法,对OLI影像光谱信息进行压缩、挖掘和选取;其次,分析灰度共生矩阵窗口尺寸和纹理特征统计量对研究区水产养殖区域的区分能力,完成纹理特征的选取,并将选取的纹理信息和光谱信息进行特征协同;最后,对特征协同数据进行多尺度分割,根据各水体类型间的光谱特征和纹理特征的差异,构建研究区3种养殖水体的模糊逻辑隶属度函数,实现对研究区水产养殖信息的自动提取。研究结果表明,该方法能较好地提取研究区水产养殖信息,总体分类精度达到97.93%。
When using remote sensing spectral information alone to extract abalone aquaculture information, it is easy to confuse aquaculture water with natural water. However, it is hard to distinguish single bulk aquaculture water with natural water using remote sensing image texture information to extract aquaculture information. In view of the above problems, this paper proposes a method for extracting aquaculture information by combining the spectral and texture information of remote sensing images with OLI image data. Firstly, based on the spectral characteristics analysis of aquaculture water types in the study area, the principal component transformation method was used to compress, mine and select OLI image spectral information. Secondly, the window size and texture feature statistics of gray level co-occurrence matrix According to the spectral characteristics and texture features of each water body type, the water quality of the aquaculture area in the study area can be distinguished from each other, the texture features are selected, and the selected texture information and spectral information are used for feature coordination. Finally, The fuzzy logic membership functions of three aquaculture water bodies in the study area were constructed to realize the automatic extraction of aquaculture information in the study area. The results show that this method can extract the aquaculture information of the study area well and the overall classification accuracy reaches 97.93%.