Example Convolution

Example of how to convolve 1D and 2D convolution kernels across an image. Besides providing the kernel, how the border is handled needs to be specified. A normalized kernel will renormalize the

Example Code:

• ExampleConvolution.java

Concepts:

• Convolution
• Spacial filtering

Example Code

/**
 * Several examples demonstrating convolution.
 *
 * @author Peter Abeles
 */
public class ExampleConvolution {

	public static void main(String[] args) {
		BufferedImage image = UtilImageIO.loadImage(UtilIO.pathExample("sunflowers.jpg"));

		ImageUInt8 gray = ConvertBufferedImage.convertFromSingle(image, null, ImageUInt8.class);

		convolve1D(gray);
		convolve2D(gray);
		normalize2D(gray);
	}

	/**
	 * Convolves a 1D kernel horizontally and vertically
	 */
	private static void convolve1D(ImageUInt8 gray) {
		ImageBorder<ImageUInt8> border = FactoryImageBorder.single(gray, BorderType.EXTENDED);
		Kernel1D_I32 kernel = new Kernel1D_I32(2);
		kernel.offset = 1; // specify the kernel's origin
		kernel.data[0] = 1;
		kernel.data[1] = -1;

		ImageSInt16 output = new ImageSInt16(gray.width,gray.height);

		GConvolveImageOps.horizontal(kernel, gray, output, border);
		ShowImages.showWindow(VisualizeImageData.standard(output, null), "1D Horizontal");

		GConvolveImageOps.vertical(kernel, gray, output, border);
		ShowImages.showWindow(VisualizeImageData.standard(output, null), "1D Vertical");
	}

	/**
	 * Convolves a 2D kernel
	 */
	private static void convolve2D(ImageUInt8 gray) {
		// By default 2D kernels will be centered around width/2
		Kernel2D_I32 kernel = new Kernel2D_I32(3);
		kernel.set(1,0,2);
		kernel.set(2,1,2);
		kernel.set(0,1,-2);
		kernel.set(1,2,-2);

		// Output needs to handle the increased domain after convolution.  Can't be 8bit
		ImageSInt16 output = new ImageSInt16(gray.width,gray.height);
		ImageBorder<ImageUInt8> border = FactoryImageBorder.single(gray, BorderType.EXTENDED);

		GConvolveImageOps.convolve(kernel, gray, output, border);
		ShowImages.showWindow(VisualizeImageData.standard(output, null), "2D Kernel");
	}

	/**
	 * Convolves a 2D normalized kernel.  This kernel is divided by its sum after computation.
	 */
	private static void normalize2D(ImageUInt8 gray) {
		// Create a Gaussian kernel with radius of 3
		Kernel2D_I32 kernel = FactoryKernelGaussian.gaussian2D(ImageUInt8.class, -1, 3);
		// Note that there is a more efficient way to compute this convolution since it is a separable kernel
		// just use BlurImageOps instead.

		// Since it's normalized it can be saved inside an 8bit image
		ImageUInt8 output = new ImageUInt8(gray.width,gray.height);

		GConvolveImageOps.convolveNormalized(kernel, gray, output);
		ShowImages.showWindow(VisualizeImageData.standard(output, null), "2D Normalized Kernel");
	}
}