海量扫测数据特征点及边界提取方法研究

The Research of Extracting Feature Points and Boundary from Massive Data

针对"水声呐"水下扫测设备测量所得的数据量大,精度高,分布规则,但同时数据的冗余度大的特点,为了真实反映水下地形,利用四叉树的数据分块思想对海量数据进行预处理,再对处理后的分块数据采用绝对高程差,高程加权平均等算法判断不同分块窗口的地形的起伏状态,然后用Jenson & Domingue等算法提取特征点,同时把图像处理边缘提取方法——梯度算子应用到河道扫测边界提取中。

The results captured by AquaticSonar have these characters., large volume of data, high precision and regular distribution. However, the data has much redundancy. In order to express the under-water terrain, quadtree and data segmentation algorithm are adopted to preprocess the massive data. Then based on the segmented data, absolute value of height difference, weighted average method and Jenson ＆ Domingue algorithm are used to estimate status of topographic relief of various regions and extract feature points, the survey boundary will be extracted by a method of the image edge extraction -- gradient operator.