Covariance.java

public static class Map extends
		Mapper<ImageHeader, FloatImage, IntWritable, FloatImage> {		
	
	public void map(ImageHeader key, FloatImage value, Context context)
			throws IOException, InterruptedException {
		if (value != null && value.getWidth() > 100 && value.getHeight() > 100) {
			float[][] tp = new float[100][32 * 32];
			for (int i = 0; i < 10; i++) {
				int x = (value.getWidth() - 32) * i / 10;
				for (int j = 0; j < 10; j++) {
					int y = (value.getHeight() - 32) * j / 10;
					FloatImage image = value.crop(x, y, 32, 32);
					float[] pels = image.getData();
					for (int iy = 0; iy < 32; iy++)
						for (int ix = 0; ix < 32; ix++)
							tp[i * 10 + j][iy * 32 + ix] = (pels[iy * 32 * 3 + ix * 3] + pels[iy * 32 * 3 + ix * 3 + 1] + pels[iy * 32 * 3 + ix * 3 + 2]) / 3 - 100.0f / 255.0f;
				}
			}
			float[] cov = new float[32 * 32 * 32 * 32];
			for (int i = 0; i < 32 * 32; i++)
				for (int j = 0; j < 32 * 32; j++) {
					cov[i * 32 * 32 + j] = 0;
					for (int k = 0; k < 100; k++)
						cov[i * 32 * 32 + j] += tp[k][i] * tp[k][j];
				}
			context.write(new IntWritable(0), new FloatImage(32 * 32, 32 * 32, 1, cov));
		}
	}
}

public static class Reduce extends
		Reducer<IntWritable, FloatImage, IntWritable, FloatImage> {
	public void reduce(IntWritable key, Iterable<FloatImage> values,
			Context context) throws IOException, InterruptedException {
		float[] cov = new float[32 * 32 * 32 * 32];
		for (int i = 0; i < 32 * 32 * 32 * 32; i++)
			cov[i] = 0;
		for (FloatImage val : values) {
			float[] image = val.getData();
			for (int i = 0; i < 32 * 32 * 32 * 32; i++)
					cov[i] += image[i];
		}
		context.write(key, new FloatImage(32 * 32, 32 * 32, 1, cov));
	}
}

MeanCompute.java

public static class Map extends
		Mapper<ImageHeader, FloatImage, IntWritable, FloatImage> {		
	
	public void map(ImageHeader key, FloatImage value, Context context)
			throws IOException, InterruptedException {
		if (value != null && value.getWidth() > 100 && value.getHeight() > 100) {
			float[] tp = new float[32 * 32];
			for (int i = 0; i < 10; i++) {
				int x = (value.getWidth() - 32) * i / 10;
				for (int j = 0; j < 10; j++) {
					int y = (value.getHeight() - 32) * j / 10;
					FloatImage image = value.crop(x, y, 32, 32);
					float[] pels = image.getData();
					for (int iy = 0; iy < 32; iy++)
						for (int ix = 0; ix < 32; ix++)
							tp[iy * 32 + ix] += (pels[iy * 32 * 3 + ix * 3] + pels[iy * 32 * 3 + ix * 3 + 1] + pels[iy * 32 * 3 + ix * 3 + 2]) / 3;
				}
			}
			for (int i = 0; i < 32; i++)
				for (int j = 0; j < 32; j++)
					tp[i * 32 + j] /= 100;
			context.write(new IntWritable(0), new FloatImage(32, 32, 1, tp));
		}
	}
}

public static class Reduce extends
		Reducer<IntWritable, FloatImage, IntWritable, FloatImage> {
	public void reduce(IntWritable key, Iterable<FloatImage> values,
			Context context) throws IOException, InterruptedException {
		float[] mean = new float[32 * 32];
		int total = 0;
		for (FloatImage val : values) {
			float[] image = val.getData();
			for (int i = 0; i < 32 * 32; i++)
					mean[i] += image[i];
			total++;
		}
		if (total > 0) {
			for (int i = 0; i < 32 * 32; i++)
				mean[i] /= total;
			context.write(key, new FloatImage(32, 32, 1, mean));
		}
	}
}