何湘, 王杰, 舒成强. 面向对象化分割方法的 TM 影像主河道信息提取—以嘉陵江为例[J]. 内江师范学院学报, 2017, (4): 78-82. DOI: 10.13603/j.cnki.51-1621/z.2017.04.013
    引用本文: 何湘, 王杰, 舒成强. 面向对象化分割方法的 TM 影像主河道信息提取—以嘉陵江为例[J]. 内江师范学院学报, 2017, (4): 78-82. DOI: 10.13603/j.cnki.51-1621/z.2017.04.013
    HE Xiang, WANG Jie, SHU Chengqiang. Main Channel Information Extraction Based on Object Oriented Segment Algorithm in TM Image: A Case Study of Jialing River[J]. Journal of Neijiang Normal University, 2017, (4): 78-82. DOI: 10.13603/j.cnki.51-1621/z.2017.04.013
    Citation: HE Xiang, WANG Jie, SHU Chengqiang. Main Channel Information Extraction Based on Object Oriented Segment Algorithm in TM Image: A Case Study of Jialing River[J]. Journal of Neijiang Normal University, 2017, (4): 78-82. DOI: 10.13603/j.cnki.51-1621/z.2017.04.013

    面向对象化分割方法的 TM 影像主河道信息提取—以嘉陵江为例

    Main Channel Information Extraction Based on Object Oriented Segment Algorithm in TM Image: A Case Study of Jialing River

    • 摘要: 为了同时利用 TM影像的空间与光谱特征信息,克服水体指数阈值法提取的缺点,利用面向对象图像分割技术,提取了四个时间段的嘉陵江主河道轮廓,然后使用 RIVWIDTH工具自动计算河道中心线与河道宽度. 以 2007年河道提取影像为例,采用四个指标对精度进行验证,分别是制图精度、用户精度、 Kappa系数和完整度,其计算结果依次为84 %、90%、0. 91、86 %.同时, RIVWIDTH工具自动河道提取工具运行速度较快,自动化程度较高,通过目视判读发现,当相对误差设定为小于25 %时,RIVWIDTH工具能够达到88 %的精度.结果表明:此方法能够快速、准确、完整提取河道轮廓,并有效消除微小的水域图斑.

       

      Abstract: Remote sensing technology had been widely applied in the river information extraction, and the water index and threshold algorithm, a kind of spectrum analysis way, were the dominant method. In spite of fast speed, the water index and threshold method could hardly consider effects of adjacent pixels, which may result in many tiny rivers channel and frag- mentation water, and then increases the workload of the subsequent processing. In order to simultaneously exploit spatial and spectral information of TM images, and overcome the shortcomings of the water index threshold, object oriented image seg- mentation technology was used to extract Jialing River channel of four time intervals, then RIVWIDTH tool was used to calcu- late the river central line and river width. Four indicators are used to verify the accuracy of the river extraction images in2007, which were mapping accuracy, user accuracy, Kappa coefficient, the complete degree, and the values for them were84%, 90%, 0. 91, 86%, respectively. At the same time, RIVWIDTH automatic channel extraction tool had high operational effi- ciency, high degree of automation. It was found that the accuracy of RIVWIDTH tool attained88% on the condition of relative error less than25%. The results showed that the model could accurately extract the channel of Jialing River, and removed most of the small patches.

       

    /

    返回文章
    返回