boofcv.alg.background.stationary.BackgroundStationaryGaussian_MS Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of feature Show documentation
BoofCV is an open source Java library for real-time computer vision and robotics applications.
There is a newer version: 0.26
/*
 * Copyright (c) 2011-2015, Peter Abeles. All Rights Reserved.
 *
 * This file is part of BoofCV (http://boofcv.org).
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *   http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package boofcv.alg.background.stationary;

import boofcv.alg.InputSanityCheck;
import boofcv.alg.misc.GImageMiscOps;
import boofcv.alg.misc.ImageMiscOps;
import boofcv.core.image.FactoryGImageMultiBand;
import boofcv.core.image.GConvertImage;
import boofcv.core.image.GImageMultiBand;
import boofcv.struct.image.*;

/**
 * Implementation of {@link BackgroundStationaryGaussian} for {@link MultiSpectral}.
 *
 * @author Peter Abeles
 */
public class BackgroundStationaryGaussian_MS
		extends BackgroundStationaryGaussian>
{

	// wrappers which provide abstraction across image types
	protected GImageMultiBand inputWrapper;
	protected GImageMultiBand bgWrapper;

	// storage for multi-band pixel values
	protected float[] inputPixel;

	// background is composed of bands*2 channels.  even = mean, odd = variance
	MultiSpectral background;

	/**
	 * Configurations background removal.
	 *
	 * @param learnRate Specifies how quickly the background is updated.  0 = static  1.0 = instant.  Try 0.05
	 * @param threshold Threshold for background.  Consult a chi-square table for reasonably values.
	 *                  10 to 16 for 1 to 3 bands.
	 * @param imageType Type of input image.
	 */
	public BackgroundStationaryGaussian_MS(float learnRate, float threshold,
										   ImageType> imageType)
	{
		super(learnRate, threshold, imageType);

		int numBands = imageType.getNumBands();

		background = new MultiSpectral(ImageFloat32.class,1,1,2*numBands);
		bgWrapper = FactoryGImageMultiBand.create(background.getImageType());
		bgWrapper.wrap(background);

		inputWrapper = FactoryGImageMultiBand.create(imageType);

		inputPixel = new float[numBands];
	}

	@Override
	public void reset() {
		background.reshape(1,1);
	}

	@Override
	public void updateBackground( MultiSpectral frame) {
		if( background.width == 1 ) {
			background.reshape(frame.width, frame.height);
			// initialize the mean to the current image and the initial variance is whatever it is set to
			for (int band = 0; band < background.getNumBands(); band += 2) {
				GConvertImage.convert(frame.getBand(band / 2), background.getBand(band));
				GImageMiscOps.fill(background.getBand(band + 1), initialVariance);
			}
			return;
		} else {
			InputSanityCheck.checkSameShape(background, frame);
		}

		inputWrapper.wrap(frame);

		int numBands = background.getNumBands()/2;
		float minusLearn = 1.0f - learnRate;


		int indexBG = 0;
		for (int y = 0; y < background.height; y++) {
			int indexInput = frame.startIndex + y*frame.stride;
			int end = indexInput + frame.width;
			while( indexInput < end ) {
				inputWrapper.getF(indexInput, inputPixel);

				for (int band = 0; band < numBands; band++) {
					ImageFloat32 backgroundMean = background.getBand(band*2);
					ImageFloat32 backgroundVar = background.getBand(band*2+1);

					float inputValue = inputPixel[band];
					float meanBG = backgroundMean.data[indexBG];
					float varianceBG = backgroundVar.data[indexBG];

					float diff = meanBG-inputValue;
					backgroundMean.data[indexBG] = minusLearn*meanBG + learnRate*inputValue;
					backgroundVar.data[indexBG] = minusLearn*varianceBG + learnRate*diff*diff;
				}

				indexInput++;
				indexBG++;
			}
		}
	}

	@Override
	public void segment(MultiSpectral frame, ImageUInt8 segmented) {
		if( background.width == 1 ) {
			ImageMiscOps.fill(segmented, unknownValue);
			return;
		}
		inputWrapper.wrap(frame);

		final int numBands = background.getNumBands()/2;

		float adjustedMinimumDifference = minimumDifference*numBands;

		int indexBG = 0;
		for (int y = 0; y < frame.height; y++) {
			int indexInput = frame.startIndex + y*frame.stride;
			int indexSegmented = segmented.startIndex + y*segmented.stride;

			int end = indexInput + frame.width;
			while( indexInput < end ) {
				inputWrapper.getF(indexInput, inputPixel);

				float mahalanobis = 0;
				for (int band = 0; band < numBands; band++) {

					ImageFloat32 backgroundMean = background.getBand(band*2);
					ImageFloat32 backgroundVar = background.getBand(band*2+1);

					float meanBG = backgroundMean.data[indexBG];
					float varBG = backgroundVar.data[indexBG];

					float diff = meanBG - inputPixel[band];
					mahalanobis += diff * diff / varBG;
				}

				if (mahalanobis <= threshold) {
					segmented.data[indexSegmented] = 0;
				} else {
					if( minimumDifference == 0) {
						segmented.data[indexSegmented] = 1;
					} else {
						float sumAbsDiff = 0;
						for (int band = 0; band < numBands; band++) {
							ImageFloat32 backgroundMean = background.getBand(band*2);
							sumAbsDiff += Math.abs(backgroundMean.data[indexBG] - inputPixel[band]);
						}
						if (sumAbsDiff >= adjustedMinimumDifference)
							segmented.data[indexSegmented] = 1;
						else
							segmented.data[indexSegmented] = 0;
					}
				}

				indexInput++;
				indexSegmented++;
				indexBG++;
			}
		}
	}
}