目录
前言
之前跟大家介绍过,通过其他软件绘制样本,然后导入到PIE-Engine studio平台上作为监督分类的训练样本,这次跟大家分享,直接在平台上进行样本的绘制。
操作步骤
导入研究区域和影像
代码如下(示例):
//1、获取采样区域范围
var geometry = pie.FeatureCollection("NGCC/CHINA_CITY_BOUNDARY")
.filter(pie.Filter.eq("name", "邯郸市"))
.first()
.geometry();
Map.centerObject(geometry, 6);
Map.addLayer(geometry, { color: 'FF0000', fillColor: '00000000', width: 1 }, "邯郸");
//2.调用哨兵-2 TOA数据集,设定筛选条件:日期、区域、云量
var img = pie.ImageCollection("S2/L2A")
.filterBounds(geometry)
.filterDate("2020-05-01", "2020-10-01")
.filter(pie.Filter.lte("cloudyPixelPercentage", 0.5))
.select(["B2", "B3", "B4","B5", "B6", "B7", "B8"])
.max()
.clip(geometry);
//设置影像参数与波段组合并加载
Map.addLayer(img, { min: 0, max: 3000, bands: ["B4", "B3", "B2"] }, "img", true);
print(img)
当然你也可以直接用平台自带的底图影像进行绘制
放大后也比较清晰
样本点绘制
设置样本点的类别,按照:水体- 30 绿地-40 建筑物-50依次进行设置。
生成代码并导入的编辑器中。
//==========分类样本文件手动绘制===========
var geometry0 = pie.FeatureCollection([pie.Feature(pie.Geometry.Point([
113.88455664035098,
36.85198010528727
], null),{
"class": "30"
}),pie.Feature(pie.Geometry.Point([
113.91687615181024,
36.850909664234905
], null),{
"class": "30"
}),pie.Feature(pie.Geometry.Point([
113.99465186475379,
36.850909664234905
], null),{
"class": "30"
}),pie.Feature(pie.Geometry.Point([
114.04783672729502,
36.85029950462521
], null),{
"class": "30"
}),pie.Feature(pie.Geometry.Point([
113.91115734939018,
36.8333656328858
], null),{
"class": "30"
}),pie.Feature(pie.Geometry.Point([
113.8954653072646,
36.811426469193705
], null),{
"class": "30"
}),pie.Feature(pie.Geometry.Point([
113.85764556577652,
36.778957478869756
], null),{
"class": "30"
}),pie.Feature(pie.Geometry.Point([
113.97278412121665,
36.77071216726111
], null),{
"class": "30"
}),pie.Feature(pie.Geometry.Point([
114.02101268831251,
36.77880479597407
], null),{
"class": "30"
}),pie.Feature(pie.Geometry.Point([
113.78844651503044,
3