摘要:以湖北省應(yīng)城市的ETM+數(shù)據(jù)為例,對3種水體識別方法進(jìn)行了對比研究。結(jié)果表明,單波段閾值法對于較大水體的提取效果較好,而對細(xì)小水體則難以識別;水體指數(shù)法對于細(xì)小水體的提取效果較單波段閾值法好,但是應(yīng)用范圍有局限,針對不同數(shù)據(jù)有不同的指數(shù)模型;K-T變換模型法精度最高,達(dá)到89.5%,對于較大水體和細(xì)小水體提取效果都較好,水體提取結(jié)果比較理想。
關(guān)鍵詞:ETM+影像;水體識別方法;K-T變換模型
中圖分類號:TP753 文獻(xiàn)標(biāo)識碼:A 文章編號:0439-8114(2013)21-5326-03
Comparative Study on the Methods of Water Body Identification Based on ETM + Image
CHENG Wu-xue1,2,LI yue2,LI Hui-jun1,ZHOU Jie-ming1,YANG Cun-jian1
(1. Key Laboratory of the Evaluation and Monitoring of Southwest Land Resources, Ministry of Education, Chengdu 610068, China;
2. Sichuan Forestry Inventory and Planning Institute, Chengdu 610081, China)
Abstract: Taking the ETM+ image of Yingcheng city in Hubei province as an example, the methods of the water body identification were compared. The results showed that the effect was good by using the single-band threshold method for large water body extraction, while it was difficult to identify the small water body; Water index method’s extract effect was better than single-band threshold method, but the range of application had limitation, reguiring different index models for different data; K-T transformation model was the method,with highest indentification accuracy the method with highest identification accuracy, which reached up to 89.5%. The extraction effect of K-T transformation method was good both for large and small water bodies.
Key words: ETM+image; method of water body indentification; K-T transformation model
水體遙感識別的機理是水體與其他地類的光譜特性具有明顯的差異,即水體在近紅外和短(中)波紅外波段的反射能量很少,而植物、土壤在這兩個波段內(nèi)的吸收能量較少,而且又具有較高的反射特性,這就使得水體在這兩個波段上與植被、土壤有明顯的區(qū)別。水體在這兩個波段上反射率低呈現(xiàn)出暗色調(diào),而土壤、植被反射率高則呈現(xiàn)出相對較亮的色調(diào)。在可見光通道波長范圍內(nèi),水體的反射率高于植被,裸土的反射率高于植被;在近紅外通道,裸土的反射率高于水體、低于植被[1,2]。
目前,提取水體的方法主要有單波段閾值法、多波段增強閾值法(即譜間關(guān)系法)、水體指數(shù)提取法、K-T變換綜合提取法等。這些方法各有優(yōu)勢,但對于具體區(qū)域來說,哪一種更適合需要經(jīng)過實踐檢驗,為此以湖北省應(yīng)城市為研究區(qū)域,對比不同水體識別方法的結(jié)果,尋找最佳的方法。
1 研究區(qū)概況及研究數(shù)據(jù)
文中所使用數(shù)據(jù)為湖北省應(yīng)城市的一景ETM+影像,行列號為123/39,成像時間為2002年10月13日。……