基于TM影像的水体信息提取算法研究

以郑洛地区为研究区域,利用LandsatTM影像进行波谱再分析,对谱间关系法进行改进,设计了更适合于郑洛地区水体信息提取的算法。通过综合比较和分析单波段阈值法、谱间关系法、归一化水体指数法和IHS(intensity hue saturation)变换法,借助数学形态学、中值滤波等进行优化处理,进而提出了改进的谱间关系法。实验证明该改进算法具有有效性,提取效果良好,能够提取到较细小的河流等水体信息。河流细节丰盈,山体阴影较少,总体精度较高,与实际情况一致性较好。

Research on Water Information Extraction Algorithm based on TM Images

Based on analyzing the Landsat TM image spectrum, the relations between the spectrum was improved, and then creating a new model on the basis of band-relation based algorithm to avoid the disadvantages of the existing methods. The new method is suitable for Zhengzhou and Luoyang. Through comparing four different methods, such as single-band threshold method, band-relation algorithm, normal differential water index and the IHS transform method. The Paper did some mathematical morphology processing, and filter optimization about the water information, then selected Zhengzhou and Luoyang as experimental area . Experiments show that the improved algorithm is effective and extract effect is good. The new algorithm is effective: Small river can be found out, less shadow of hills, the overall accuracy is higher and consistent with the actual situation.