Home > Articles > Programming > Java

  • Print
  • + Share This
This chapter is from the book

Implementing a Canvas for Image Manipulation

Our goal is to build an image-viewing application that allows users to interactively manipulate images. The image manipulation functions we intend to implement are pan, zoom, rotate, flip, and shear. Clearly we need a component class or bean that renders images and provides manipulation functionality.

In Chapter 6 we implemented an image-rendering component called ImageCanvas. The image manipulator class, which we'll name ImageCanvas2D, will extend the ImageCanvas class. To implement the requirements described in the preceding section, ImageCanvas2D will implement the ImageManipulator interface (see Figure 7.5).

FIGURE 7.5 The ImageCanvas2D class hierarchy

Besides all the set and get methods for the properties listed in Table 7.1, Image Canvas2D must implement two important methods: paintImage() and apply Transform(). Listing 7.4 shows the implementation.

LISTING 7.4 The ImageCanvas2D class

package com.vistech.imageviewer;
import java.io.*;
import java.awt.*;
import java.awt.event.*;
import java.awt.image.*;

import java.util.*;
import java.beans.*;
import javax.swing.*;
import java.awt.geom.*;
import com.vistech.events.*;

public class ImageCanvas2D extends ImageCanvas implements

  protected AffineTransform atx = new AffineTransform();
  protected AffineTransform dispModeAtx = new AffineTransform();
  protected AffineTransform flipAtx = new AffineTransform();

  // Pan variables
  protected Point panOffset = new Point(0,0);
  // Zoom variables
  protected boolean magOn = true;
  protected double magFactor = 1.0;
  protected int magCenterX = 0;
  protected int magCenterY =0;
  protected Point zoomOffset = new Point(0,0);
  // Rotation variables
  protected double rotationAngle = 0.0;
  protected boolean rotateOn = true;
  protected int rotationCenterX = 0;
  protected int rotationCenterY = 0;
  // Shear variables
  protected boolean shearOn = true;
  protected double shearFactor = 0.0;
  protected double shearX =0.0, shearY=0.0;
  // Off-screen image width and height
  private int bufImageWidth = -1;
  private int bufImageHeight = -1;
  protected int interpolationType = AffineTransformOp.TYPE_NEAREST_NEIGHBOR;

  public ImageCanvas2D(){ init();}

  protected void init(){

  public void setMagFactor(double magFactor){
             new Double(this.magFactor),
             new Double(magFactor));
   this.magFactor = magFactor;
  public double getMagFactor(){ return magFactor;}
  public void setShearFactor(double shearFactor){
             new Double(this.shearFactor),
             new Double(shearFactor));
   this.shearFactor = shearFactor;

  public double getShearFactor(){ return shearFactor; }
  public double getShearFactorX(){ return shearX;}
  public double getShearFactorY(){ return shearY;}

  public void setRotationAngle(double rotationAngle){
             new Double(this.rotationAngle),
             new Double(rotationAngle));
   this.rotationAngle = rotationAngle;
  public double getRotationAngle(){ return rotationAngle; }

  public void setPanOffset(Point panOffset){
   this.panOffset = panOffset;

  public Point getPanOffset(){ return panOffset; }

  public void setInterpolationType(int interType) {interpolationType = interType;}
  public int getInterpolationType() { return interpolationType;}

  public AffineTransform getTransform(){ return atx; }
  public void setTransform(AffineTransform at){ atx = at; }

  public void setMagOn(boolean onOff){ magOn = onOff;}
  public boolean getMagOn(){ return magOn;}

  public synchronized boolean doDisplayModeAndFlip(int imageWidth, int imageHeight){
   width = this.getBounds().width;
   height = this.getBounds().height;
   double magX= (double)width/(double)imageWidth;
   double magY = (double)height/(double)imageHeight;
   int bufferWid = width, bufferHt=height;
   dispModeAtx = new AffineTransform();
    case DisplayMode.ORIG_SIZE:
     bufferWid = imageWidth;
     bufferHt = imageHeight;
    case DisplayMode.SCALED:
     double mag= (magY > magX)? magX:magY;
     dispModeAtx.setToScale(mag, mag);
     bufferWid = (int)(imageWidth*mag);
     bufferHt = (int)(imageHeight*mag);
   case DisplayMode.TO_FIT:
     dispModeAtx.setToScale(magX, magY);
     bufferWid = width;
     bufferHt = height;
   BufferedImage bi = new BufferedImage(bufferWid, bufferHt,imageType);
   Graphics2D bigc = bi.createGraphics();
   if(originalImageType == TYPE_AWT_IMAGE)
     bigc.drawImage(awtImage, dispModeAtx, this);
   else bigc.drawImage(bufferedImage, dispModeAtx, this);

   flipAtx = createFlipTransform(flipMode, bufferWid, bufferHt);
   AffineTransformOp atop = new AffineTransformOp(flipAtx, interpolationType);
   offScrImage = atop.filter(bi, null);
   offScrGc = offScrImage.createGraphics();
   applyTransform(offScrImage, atx);
   return true;

  protected void applyTransform(BufferedImage bi, AffineTransform atx){
   if(offScrImage == null) return;
   if(displayImage == null) createDisplayImage();
   AffineTransformOp atop = new AffineTransformOp(atx, interpolationType);
   dispGc.fillRect(0,0,displayImage.getWidth(), displayImage.getHeight());
   if(clipShape == null) atop.filter(bi, displayImage);
   else dispGc.drawImage(bi,atx,this);

  public void applyTransform(AffineTransform atx){
   applyTransform(offScrImage, atx);
   this.atx = atx;

  public void reset(){
    panOffset = new Point(0,0);
    magCenterX = 0; magCenterY =0;
    magFactor = 1.0;
    rotationAngle = 0.0;
    shearX = 0.0; shearY = 0.0;
    atx = new AffineTransform();

  public void resetManipulation(){
    panOffset = new Point(0,0);
    magCenterX = 0; magCenterY =0;
    magFactor = 1.0;
    shearX = 0.0; shearY = 0.0;
    rotationAngle = 0.0;
    atx = new AffineTransform();


The ImageCanvas2D class holds the current value of the transform property in atx, which is concatenated whenever the current image is manipulated.

Recall from Chapter 6 how images are painted (see Figure 6.4). When the setAWTImage() or setBufferedImage() method is called, the original image is loaded. When the setDisplayMode() or setFlipMode() method is called, an offScreenImage object is created by application of the current displayMode and flipMode properties. The offScrImage variable holds this image.

Any manipulation function will either concatenate atx or create a new instance of atx and apply it to offScrImage. The applyTransform() method discussed earlier (see Listings 7.1 and 7.2) does this. The applyTransform() method creates another BufferedImage object, which is saved in the displayImage variable. You may recall that offScrImage and displayImage are defined in the superclass ImageCanvas. Having two copies of the same image may seem wasteful, but they are needed for performance.

Using the Affine Transformation to Set Display and Flip Modes

The doDisplayModeAndFlip() method overrides the doDisplayModeAndFlip() method in the ImageCanvas class. As far as the display mode is concerned, we need to apply just one type of transformation: scaling. On the basis of the requirements listed in Chapter 6, when the display modes are set, the viewport is reset. This means that we need to reset the underlying AffineTransform object.

The doDisplayModeAndFlip() method in Listing 7.4 first creates an Affine Transform object—dispModeAtx—and then checks for display mode type, which can be any one of ORIG_SIZE, SCALED, or TO_FIT. In the case of ORIG_SIZE, the affine transformation is the identity transformation itself because there is no scaling. In the case of SCALED, the scale factor is calculated in both the x and the y directions, and the larger factor is chosen. The doDisplayModeAndFlip() method then calls the setToScale() method to create a transformation with a specified scale factor. In this case the scale factor is the same in both the x and the y directions because of the need to preserve the aspect ratio of the image. In the TO_FIT case, the scale factors in both directions are taken into consideration, and an AffineTransform object is created.

Next the doDisplayModeAndFlip() method creates a BufferedImage instance that is as big as the image to be rendered. The original image is drawn on this BufferedImage object by the drawImage() method. The dispModeAtx object is passed as a parameter to drawImage(), thus applying the desired display mode to the original image. The image thus created may need to be flipped, depending on the flip mode selected.

The doDisplayModeAndFlip() method then creates a flip transformation. It uses the filter() method of the AffineTransformOp class to apply the flip transformation to the current image and creates another BufferedImage object. Although this image is ready for rendering, the doDisplayModeAndFlip() method performs one more transformation because we may use this image for image manipulation purposes. The BufferedImage object obtained after the flip mode operation is called offScrImage, and we'll make this image the base for all other image manipulation operations.

The doDisplayModeAndFlip() method transforms the image through the applyTransform() method, which we described earlier. The resulting image—displayImage—is the one that is rendered (see Figure 6.4). So there are two instance variables: atx and displayImage. The variable atx will be concatenated when an image is manipulated, and it will be reset when the display or flip modes are set.


The AffineTransform class does not have direct methods for creating a transformation matrix for flip. We'll use the reflection transformation to create a flip matrix (see Chapter 4). You may recall that the reflection transformation creates a mirror image rather than a flipped image. To create a flip transformation, all we need to do is to translate the reflection matrix to the same quadrant as the image. Listing 7.5 shows how to do this.

LISTING 7.5 Creating a flip transformation

static public AffineTransform createFlipTransform(int mode,
                          int imageWid,
                          int imageHt){
   AffineTransform at = new AffineTransform();
      case FlipMode.NORMAL:
      case FlipMode.TOP_BOTTOM:
       at = new AffineTransform(new double[] {1.0,0.0,0.0,-1.0});
       at.translate(0.0, -imageHt);
      case FlipMode.LEFT_RIGHT :
       at = new AffineTransform(new double[] {-1.0,0.0,0.0,1.0});
       at.translate(-imageWid, 0.0);
      case FlipMode.TOP_BOTTOM_LEFT_RIGHT:
       at = new AffineTransform(new double[] {-1.0,0.0,0.0,-1.0});
       at.translate(-imageWid, -imageHt);
   return at;

As Listing 7.5 shows, the createFlipTransform() method constructs an AffineTransform object for reflection by passing a flat matrix array. This array contains the elements of a top left-hand 2 3 2 matrix, which is different for each of the flip mode cases. Once the reflection transformation has been constructed, it is translated to the original image location itself.

In the sections that follow we'll look at the different types of manipulation, starting with pan.

  • + Share This
  • 🔖 Save To Your Account