【摘要】：選用2013年7月23日-10月27日期間5期分辨率為5.2 m×7.6 m的Radarsat-2影像為數(shù)據(jù),采用支持向量機(jī)法(SVM)和最大似然法(MLC)分別對各時(shí)相水稻種植面積進(jìn)行提取,并以地面實(shí)測GPS水稻樣方進(jìn)行精度驗(yàn)證。結(jié)果表明SVM和MLC方法的水稻面積提取精度均在9月9日達(dá)到最高,所以選擇在9月9日的水稻面積提取結(jié)果上研究耕地地塊優(yōu)化和碎小圖斑去除對精度的影響。通過耕地地塊優(yōu)化和碎小圖斑去除處理,水稻面積提取精度顯著提高,SVM法由原先的72.876%提高到95.482%,MLC法由74.224%提高到91.792%。
[Abstract]:From July 23 to October 27, 2013, five Radarsat-2 images with a resolution of 5.2 m 脳 7.6 m were selected as data. The rice planting area was extracted by using support vector machine (SVM) and maximum likelihood method (MLC), respectively. The accuracy was verified by ground measured GPS rice square. The results showed that the precision of rice area extraction by SVM and MLC methods reached the highest on September 9. Therefore, the effect of farmland optimization and small patch removal on the precision was studied based on the results of rice area extraction on September 9th. The precision of rice area extraction was improved significantly through the optimization of cultivated land and small patch removal, and the SVM method increased from 72.876% to 95.482MLC from 74.224% to 91.792%.
【作者單位】： 江蘇省農(nóng)業(yè)科學(xué)院農(nóng)業(yè)經(jīng)濟(jì)與信息研究所;
【基金】：國家科技重大專項(xiàng)課題(09-Y30B03-9001-13/15-4) 江蘇省農(nóng)業(yè)科學(xué)院基本科研業(yè)務(wù)專項(xiàng)課題(ZX-15-3003);江蘇省農(nóng)業(yè)科學(xué)院基金項(xiàng)目(6111651;6111650) 農(nóng)業(yè)部遙感應(yīng)用中心技術(shù)創(chuàng)新課題(2911660)
【分類號】：S127;S511

【相似文獻(xiàn)】

相關(guān)期刊論文 前1條

1 化國強(qiáng);李晨;楊沈斌;葛道闊;李秉柏;;利用Radarsat-2數(shù)據(jù)基于比值檢測的水稻制圖[J];江蘇農(nóng)業(yè)學(xué)報(bào);2012年06期

相關(guān)碩士學(xué)位論文 前3條

1 周志全;基于RADARSAT-2干涉數(shù)據(jù)的礦區(qū)地表形變提取研究[D];北京交通大學(xué);2016年

2 景卓鑫;基于神經(jīng)網(wǎng)絡(luò)方法與RADARSAT-2雷達(dá)遙感數(shù)據(jù)的水稻參數(shù)反演研究[D];華東師范大學(xué);2014年

3 匡薇;Radarsat-2數(shù)據(jù)處理方法研究及其在地面沉降中的應(yīng)用[D];中國地質(zhì)大學(xué)（北京）;2014年

，

本文編號：2315879