图像处理之给定任意四点不规则放缩

简介: 图像处理之给定任意四点不规则放缩

基本原理:

计算四个点的增长斜率,使用双线性插值实现像素填充。

废话也懒得说啦,自己看代码吧,我从一个地方抄袭+修改了一下

源来的代码,原因是原来的代码太乱了,也太让人费解了。

运行效果:

1366901518_1725.jpg

滤镜源代码:

package com.gloomyfish.filter.study;
 
import java.awt.Rectangle;
import java.awt.image.BufferedImage;
 
/**
 * 
 * @author gloomy-fish 2013-04-25
 *
 */
public class PerspectiveFilter extends AbstractBufferedImageOp {
  private float x0, y0, x1, y1, x2, y2, x3, y3;
  private float dx1, dy1, dx2, dy2, dx3, dy3;
  private float A, B, C, D, E, F, G, H, I;
  private float a11, a12, a13, a21, a22, a23, a31, a32, a33;
    private boolean scaled;
    private int width;
    private int height;
    
  @Override
  public BufferedImage filter(BufferedImage src, BufferedImage dest) {
        width = src.getWidth();
        height = src.getHeight();
        if ( dest == null )
            dest = createCompatibleDestImage( src, null );
      A = a22*a33 - a32*a23;
      B = a31*a23 - a21*a33;
      C = a21*a32 - a31*a22;
      D = a32*a13 - a12*a33;
      E = a11*a33 - a31*a13;
      F = a31*a12 - a11*a32;
      G = a12*a23 - a22*a13;
      H = a21*a13 - a11*a23;
      I = a11*a22 - a21*a12;
        if ( !scaled ) {
            float invWidth = 1.0f/width;
            float invHeight = 1.0f/height;
 
            A *= invWidth;
            D *= invWidth;
            G *= invWidth;
            B *= invHeight;
            E *= invHeight;
            H *= invHeight;
        }
        
    int[] inPixels = getRGB( src, 0, 0, width, height, null );
 
    int srcWidth = width;
    int srcHeight = height;
    int srcWidth1 = width-1;
    int srcHeight1 = height-1;
    int outX=0, outY=0;
    Rectangle transformedSpace = new Rectangle(0, 0, width, height);
    transformSpace(transformedSpace);
    outX = transformedSpace.x;
    outY = transformedSpace.y;
    int outWidth = transformedSpace.width;
    int outHeight = transformedSpace.height;
    // int index = 0;
    int[] outPixels = new int[transformedSpace.width];
    float[] out = new float[2];
 
    for (int y = 0; y < outHeight; y++) {
      for (int x = 0; x < outWidth; x++) {
        transformInverse(outX+x, outY+y, out);
        int srcX = (int)Math.floor( out[0] );
        int srcY = (int)Math.floor( out[1] );
        float xWeight = out[0]-srcX;
        float yWeight = out[1]-srcY;
        int nw, ne, sw, se;
 
        if ( srcX >= 0 && srcX < srcWidth1 && srcY >= 0 && srcY < srcHeight1) {
          // Easy case, all corners are in the image
          int i = srcWidth*srcY + srcX;
          nw = inPixels[i];
          ne = inPixels[i+1];
          sw = inPixels[i+srcWidth];
          se = inPixels[i+srcWidth+1];
        } else {
          // Some of the corners are off the image
          nw = getPixel( inPixels, srcX, srcY, srcWidth, srcHeight );
          ne = getPixel( inPixels, srcX+1, srcY, srcWidth, srcHeight );
          sw = getPixel( inPixels, srcX, srcY+1, srcWidth, srcHeight );
          se = getPixel( inPixels, srcX+1, srcY+1, srcWidth, srcHeight );
        }
        outPixels[x] = ImageMath.bilinearInterpolate(xWeight, yWeight, nw, ne, sw, se);
      }
      setRGB( dest, 0, y, transformedSpace.width, 1, outPixels );
    }
    
    return dest;
  }
  
  protected void transformSpace( Rectangle rect ) {
    if ( scaled ) {
            rect.x = (int)Math.min( Math.min( x0, x1 ), Math.min( x2, x3 ) );
            rect.y = (int)Math.min( Math.min( y0, y1 ), Math.min( y2, y3 ) );
            rect.width = (int)Math.max( Math.max( x0, x1 ), Math.max( x2, x3 ) ) - rect.x;
            rect.height = (int)Math.max( Math.max( y0, y1 ), Math.max( y2, y3 ) ) - rect.y;
            return;
        }
  }
  final private int getPixel( int[] pixels, int x, int y, int width, int height ) {
    if (x < 0 || x >= width || y < 0 || y >= height) {
      return pixels[(ImageMath.clamp(y, 0, height-1) * width) + ImageMath.clamp(x, 0, width-1)] & 0x00ffffff;
    }
    return pixels[ y*width+x ];
  }
  
  public PerspectiveFilter() {
    this( 0, 0, 1, 0, 1, 1, 0, 1);
  }
    
  public PerspectiveFilter(float x0, float y0, float x1, float y1, float x2, float y2, float x3, float y3) {
    unitSquareToQuad(x0, y0, x1, y1, x2, y2, x3, y3);
  }
 
  protected void transformInverse( int x, int y, float[] out ) {
    out[0] = width * (A*x+B*y+C)/(G*x+H*y+I);
    out[1] = height * (D*x+E*y+F)/(G*x+H*y+I);
  }
  
  public void unitSquareToQuad( float x0, float y0, float x1, float y1, float x2, float y2, float x3, float y3 ) {
    this.x0 = x0;
    this.y0 = y0;
    this.x1 = x1;
    this.y1 = y1;
    this.x2 = x2;
    this.y2 = y2;
    this.x3 = x3;
    this.y3 = y3;
    
    dx1 = x1-x2;
    dy1 = y1-y2;
    dx2 = x3-x2;
    dy2 = y3-y2;
    dx3 = x0-x1+x2-x3;
    dy3 = y0-y1+y2-y3;
    
    if (dx3 == 0 && dy3 == 0) {
      a11 = x1-x0;
      a21 = x2-x1;
      a31 = x0;
      a12 = y1-y0;
      a22 = y2-y1;
      a32 = y0;
      a13 = a23 = 0;
    } else {
      a13 = (dx3*dy2-dx2*dy3)/(dx1*dy2-dy1*dx2);
      a23 = (dx1*dy3-dy1*dx3)/(dx1*dy2-dy1*dx2);
      a11 = x1-x0+a13*x1;
      a21 = x3-x0+a23*x3;
      a31 = x0;
      a12 = y1-y0+a13*y1;
      a22 = y3-y0+a23*y3;
      a32 = y0;
    }
        a33 = 1;
        scaled = false;
  }
  
  public void setCorners(float x0, float y0, float x1, float y1, float x2, float y2, float x3, float y3) {
    unitSquareToQuad( x0, y0, x1, y1, x2, y2, x3, y3 );
        scaled = true;
  }
    
}

转载请务必注明

相关文章
|
8月前
|
监控 算法 安全
基于伽马变换自适应修正的全景首尾融合算法
基于伽马变换自适应修正的全景首尾融合算法
|
22天前
|
算法 BI 计算机视觉
图像处理之积分图应用一(半径无关的快速模糊算法)
图像处理之积分图应用一(半径无关的快速模糊算法)
14 0
|
1月前
|
算法 计算机视觉
OpenCV(四十四):亚像素级别角点位置优化
OpenCV(四十四):亚像素级别角点位置优化
75 0
|
1月前
[Halcon&几何] 线段中点、端点和角度的计算
[Halcon&几何] 线段中点、端点和角度的计算
104 0
|
1月前
|
机器学习/深度学习 算法 C#
C# | 凸包算法之Jarvis,寻找一组点的边界/轮廓
这篇关于凸包算法的文章,本文使用C#和Jarvis算法来实现凸包算法。 首先消除两个最基本的问题: 什么是凸包呢? 凸包是一个包围一组点的凸多边形。凸多边形是指多边形中的每个内角都小于180度的多边形。 凸包算法有什么用呢? 凸包算法的作用是找到这个凸多边形,并且使用最少的点来绘制出它的轮廓。凸包算法在计算机图形学、计算几何和机器学习等领域中有着广泛的应用。
51 0
射线法——判断一个点是否在多边形内部(适用于凸多边形和凹多边形)【关键原理解释+文字伪代码】
射线法——判断一个点是否在多边形内部(适用于凸多边形和凹多边形)【关键原理解释+文字伪代码】
437 0
三维之外的更高维度,数学家发现了无限可能的黑洞形状
三维之外的更高维度,数学家发现了无限可能的黑洞形状
110 0
【双目视觉】 理想条件下计算物体距离
【双目视觉】 理想条件下计算物体距离
51 0
|
数据采集 文字识别 算法
二值图像分析:案例实战(文本分离+硬币计数)
二值图像分析:案例实战(文本分离+硬币计数)
261 0
二值图像分析:案例实战(文本分离+硬币计数)