001/** 002 * Copyright (c) 2011, The University of Southampton and the individual contributors. 003 * All rights reserved. 004 * 005 * Redistribution and use in source and binary forms, with or without modification, 006 * are permitted provided that the following conditions are met: 007 * 008 * * Redistributions of source code must retain the above copyright notice, 009 * this list of conditions and the following disclaimer. 010 * 011 * * Redistributions in binary form must reproduce the above copyright notice, 012 * this list of conditions and the following disclaimer in the documentation 013 * and/or other materials provided with the distribution. 014 * 015 * * Neither the name of the University of Southampton nor the names of its 016 * contributors may be used to endorse or promote products derived from this 017 * software without specific prior written permission. 018 * 019 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND 020 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 021 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 022 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR 023 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 024 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 025 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON 026 * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 027 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 028 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 029 */ 030package org.openimaj.image.processing.convolution; 031 032import org.openimaj.image.FImage; 033import org.openimaj.image.processor.ImageProcessor; 034 035/** 036 * Image convolution with a triangular filter. Implementation is based on 037 * repeated convolution with rectangular kernels, which is done efficiently 038 * using a integral image style approach. Overall complexity is independent of 039 * filter size and linear in the number of pixels. 040 * 041 * @author Jonathon Hare (jsh2@ecs.soton.ac.uk) 042 * 043 */ 044public class FTriangleFilter implements ImageProcessor<FImage> { 045 private boolean zeropad; 046 private int filterHeight; 047 private int filterWidth; 048 049 /** 050 * Construct with the given dimensions. 051 * 052 * @param filterWidth 053 * width of filter 054 * @param filterHeight 055 * height of filter 056 * @param zeropad 057 * zero-pad off the edge of the image if true; duplicate edge 058 * value otherwise. 059 */ 060 public FTriangleFilter(int filterWidth, int filterHeight, boolean zeropad) { 061 super(); 062 this.filterWidth = filterWidth; 063 this.filterHeight = filterHeight; 064 this.zeropad = zeropad; 065 } 066 067 /** 068 * Construct with the given dimensions. Edge effects are handled by 069 * duplicating the edge pixels. 070 * 071 * @param filterWidth 072 * half-width of filter 073 * @param filterHeight 074 * half-height of filter 075 */ 076 public FTriangleFilter(int filterWidth, int filterHeight) { 077 this(filterWidth, filterHeight, false); 078 } 079 080 @Override 081 public void processImage(FImage image) { 082 convolve(image, filterWidth, filterHeight, zeropad); 083 } 084 085 /** 086 * Construct a triangular kernel of the given size. The kernel will have 087 * 2*width - 1 elements. 088 * 089 * @param width 090 * the kernel half-width 091 * @return the triangular kernel 092 */ 093 public static float[] createKernel1D(int width) { 094 final float[] kernel = new float[width * 2 - 1]; 095 final float invNorm = 1f / (width * width); 096 097 kernel[width - 1] = width * invNorm; 098 for (int i = 0; i < width - 1; i++) { 099 kernel[i] = (i + 1) * invNorm; 100 kernel[kernel.length - i - 1] = kernel[i]; 101 } 102 return kernel; 103 } 104 105 static void convolve(FImage image, int filterWidth, int filterHeight, boolean zeropad) { 106 convolveVertical(image, image, filterHeight, zeropad); 107 convolveHorizontal(image, image, filterWidth, zeropad); 108 } 109 110 static void convolveVertical(FImage dest, FImage image, int filterSize, boolean zeropad) { 111 if (image.height == 0) { 112 return; 113 } 114 115 final float scale = (float) (1.0 / ((double) filterSize * (double) filterSize)); 116 final float[] buffer = new float[image.height + filterSize]; 117 final int bufferOffset = filterSize; 118 119 for (int x = 0; x < image.width; x++) { 120 // integrate backward 121 buffer[bufferOffset + image.height - 1] = image.pixels[image.height - 1][x]; 122 int y; 123 for (y = image.height - 2; y >= 0; --y) { 124 buffer[bufferOffset + y] = buffer[bufferOffset + y + 1] + image.pixels[y][x]; 125 } 126 if (zeropad) { 127 for (; y >= -filterSize; y--) { 128 buffer[bufferOffset + y] = buffer[bufferOffset + y + 1]; 129 } 130 } else { 131 for (; y >= -filterSize; y--) { 132 buffer[bufferOffset + y] = buffer[bufferOffset + y + 1] + image.pixels[0][x]; 133 } 134 } 135 136 // filter forward 137 for (y = -filterSize; y < image.height - filterSize; y++) { 138 buffer[bufferOffset + y] = buffer[bufferOffset + y] - buffer[bufferOffset + y + filterSize]; 139 } 140 if (!zeropad) { 141 for (y = image.height - filterSize; y < image.height; ++y) { 142 buffer[bufferOffset + y] = 143 buffer[bufferOffset + y] - buffer[bufferOffset + image.height - 1] 144 * (image.height - filterSize - y); 145 } 146 } 147 148 // integrate forward 149 for (y = -filterSize + 1; y < image.height; y++) { 150 buffer[bufferOffset + y] += buffer[bufferOffset + y - 1]; 151 } 152 153 // filter backward 154 for (y = dest.height - 1; y >= 0; y--) { 155 dest.pixels[y][x] = scale * (buffer[bufferOffset + y] - buffer[bufferOffset + y - filterSize]); 156 } 157 } 158 } 159 160 static void convolveHorizontal(FImage dest, FImage image, int filterSize, boolean zeropad) { 161 if (image.width == 0) { 162 return; 163 } 164 165 final float scale = (float) (1.0 / ((double) filterSize * (double) filterSize)); 166 final float[] buffer = new float[image.width + filterSize]; 167 final int bufferOffset = filterSize; 168 169 for (int y = 0; y < image.height; y++) { 170 // integrate backward 171 buffer[bufferOffset + image.width - 1] = image.pixels[y][image.width - 1]; 172 int x; 173 for (x = image.width - 2; x >= 0; --x) { 174 buffer[bufferOffset + x] = buffer[bufferOffset + x + 1] + image.pixels[y][x]; 175 } 176 if (zeropad) { 177 for (; x >= -filterSize; x--) { 178 buffer[bufferOffset + x] = buffer[bufferOffset + x + 1]; 179 } 180 } else { 181 for (; x >= -filterSize; x--) { 182 buffer[bufferOffset + x] = buffer[bufferOffset + x + 1] + image.pixels[y][0]; 183 } 184 } 185 186 // filter forward 187 for (x = -filterSize; x < image.width - filterSize; x++) { 188 buffer[bufferOffset + x] = buffer[bufferOffset + x] - buffer[bufferOffset + x + filterSize]; 189 } 190 if (!zeropad) { 191 for (x = image.width - filterSize; x < image.width; ++x) { 192 buffer[bufferOffset + x] = 193 buffer[bufferOffset + x] - buffer[bufferOffset + image.width - 1] 194 * (image.width - filterSize - x); 195 } 196 } 197 198 // integrate forward 199 for (x = -filterSize + 1; x < image.width; x++) { 200 buffer[bufferOffset + x] += buffer[bufferOffset + x - 1]; 201 } 202 203 // filter backward 204 for (x = dest.width - 1; x >= 0; x--) { 205 dest.pixels[y][x] = scale * (buffer[bufferOffset + x] - buffer[bufferOffset + x - filterSize]); 206 } 207 } 208 } 209}