Page 232 - The Definitive Guide to Building Java Robots

Preston_5564C06.fm Page 213 Friday, September 23, 2005 5:13 AM

CHAPTER 6 ■ VISION 213

• ColorGram: A data structure that represents the color ratios as a linear equation
• ColorGramCalibration: A program that creates a ColorGram from a thumbnail of the
desired image

Next, I’ll describe image processing using Java’s Advanced Imaging API.

6.4 Advanced Image Processing

To find the edge of an image requires area-level processing. Area-level processing is the process
of looking at a specific pixel in relation to the pixels around it. This is done by applying a kernel
to an image. Some operations we will show are low-pass filters or smoothing, high-pass filters
or sharpening, and line-finding filters known as Sobel Gradient masks.

Code Objective

The following objectives are handled in this example:
• Smooth an image by passing it through a low-pass filter.

• Sharpen an image by passing it through a high-pass filter.
• Get the edges of an image by passing it through a Sobel Gradient filter.

Code Discussion

Smoothing involves a kernel containing values of 1/9 in a 3×3 matrix. When applying this
kernel to the image, you get the results shown in Figure 6-21.
Sharpening involves a kernel of 0, –1, 0, –1, 5, –1, 0, –1, 0 in a 3×3 matrix. When applying
this kernel to the image, you get the results shown in Figure 6-22.
The Sobel Gradient filter involves taking the gradient of neighboring pixels to find
edges. When applying this kernel to an image, you get the results shown in Figure 6-23. (See
Examples 6-24 through 6-27.)

Example 6-24. bufferedToPlanar() and planarToBuffered()
private PlanarImage bufferedToPlanar(BufferedImage bImg) {
Image awtImg = Toolkit.getDefaultToolkit().createImage(bImg.getSource());
return JAI.create("awtimage", awtImg);
}

private BufferedImage planarToBuffered(PlanarImage pImg) {
return pImg.getAsBufferedImage();
}

97022d2480fe4a63cfdfa123a6e70098