Douglas-Peucker算法(道格拉斯-普克算法)是将曲线近似表示为一系列点,并减少点的数量的一种算法。它的优点是具有平移和旋转不变性,给定曲线与阈值后,抽样结果一定。Douglas—Peucker算法通常用于线状矢量数据压缩、轨迹数据压缩等。
算法步骤
- 连接曲线首尾两点A、B形成一条直线AB;
- 计算曲线上离该直线段距离最大的点C,计算其与AB的距离d;
- 比较该距离与预先给定的阈值threshold的大小,如果小于threshold,则以该直线作为曲线的近似,该段曲线处理完毕。
- 如果距离大于阈值,则用点C将曲线分为两段AC和BC,并分别对两段曲线进行步骤[1~3]的处理。
- 当所有曲线都处理完毕后,依次连接各个分割点形成折线,作为原曲线的近似。
实现代码
Java实现代码如下,代码引用自JTS库。
class DouglasPeuckerLineSimplifier {
private Coordinate[] pts;
private boolean[] usePt;
private double distanceTolerance;
private LineSegment seg = new LineSegment();
public static Coordinate[] simplify(Coordinate[] pts, double distanceTolerance) {
DouglasPeuckerLineSimplifier simp = new DouglasPeuckerLineSimplifier(pts);
simp.setDistanceTolerance(distanceTolerance);
return simp.simplify();
}
public DouglasPeuckerLineSimplifier(Coordinate[] pts) {
this.pts = pts;
}
public void setDistanceTolerance(double distanceTolerance) {
this.distanceTolerance = distanceTolerance;
}
public Coordinate[] simplify() {
this.usePt = new boolean[this.pts.length];
for(int i = 0; i < this.pts.length; ++i) {
this.usePt[i] = true;
}
this.simplifySection(0, this.pts.length - 1);
CoordinateList coordList = new CoordinateList();
for(int i = 0; i < this.pts.length; ++i) {
if(this.usePt[i]) {
coordList.add(new Coordinate(this.pts[i]));
}
}
return coordList.toCoordinateArray();
}
private void simplifySection(int i, int j) {
if(i + 1 != j) {
this.seg.p0 = this.pts[i];
this.seg.p1 = this.pts[j];
double maxDistance = -1.0D;
int maxIndex = i;
int k;
for(k = i + 1; k < j; ++k) {
double distance = this.seg.distance(this.pts[k]);
if(distance > maxDistance) {
maxDistance = distance;
maxIndex = k;
}
}
if(maxDistance <= this.distanceTolerance) {
for(k = i + 1; k < j; ++k) {
this.usePt[k] = false;
}
} else {
this.simplifySection(i, maxIndex);
this.simplifySection(maxIndex, j);
}
}
}
}