import org.bytedeco.javacpp.Pointer; import org.nd4j.linalg.api.buffer.DataBuffer; import org.nd4j.linalg.api.ndarray.INDArray; import org.nd4j.linalg.api.ops.impl.indexaccum.IAMax; import org.nd4j.linalg.api.ops.impl.indexaccum.IAMin; import org.nd4j.linalg.factory.Nd4j; import org.nd4j.linalg.indexing.NDArrayIndex; import javax.imageio.ImageIO; import java.awt.*; import java.awt.image.BufferedImage; import java.io.*; import java.util.*; import java.util.List; /** * Created by Administrator on 2017/9/21 0021. */ public class ImageUtil { public static Image readImage(String path) { Image image = null; File imagePath = new File(path); try { image = ImageIO.read(imagePath); return image; } catch (IOException e) { e.printStackTrace(); return null; } } public static BufferedImage readFileImage(String path){ File file = new File(path); BufferedImage image = null; try { image = ImageIO.read(file); return image; } catch (IOException e) { e.printStackTrace(); return image ; } } //BufferedImage转成数组 public static INDArray convertImageToArray(BufferedImage image){ int width = image.getWidth() ; int height = image.getHeight() ; Nd4j.factory().setDType(DataBuffer.Type.DOUBLE); INDArray rgbArray = Nd4j.zeros(width,height,3); for(int j=0;j<height;j++){ for(int i=0;i<width;i++){ // int rgb = image.getRGB(i,j); Object data = image.getRaster().getDataElements(i, j, null);//获取该点像素，并以object类型表示 int R = image.getColorModel().getRed(data); int G = image.getColorModel().getBlue(data); int B = image.getColorModel().getGreen(data); // Color color = new Color(rgb); rgbArray.putScalar(i,j,0,R); rgbArray.putScalar(i,j,1,G); rgbArray.putScalar(i,j,2,B); //System.out.println(rgbArray[j][i]); } } return rgbArray ; } //BufferedImage转成数组 /** * 将图片转成INDArray，图片对象识别，INDArray[1,3,w,h] * @param image 图片 * @param w 图片宽度 * @param h 图片高 * @param channel 图片图片颜色通道 * */ public static INDArray imageToArray(BufferedImage image,int w,int h,int channel){ int width = image.getWidth() ; int height = image.getHeight() ; BufferedImage resizeImag = new BufferedImage(w,h,BufferedImage.TYPE_INT_RGB); resizeImag.getGraphics().drawImage(image,0,0,w,h,null); Nd4j.factory().setDType(DataBuffer.Type.DOUBLE); INDArray rgbArray = Nd4j.zeros(1,channel,w,h); for(int j=0;j<h;j++){ for(int i=0;i<w;i++){ // int rgb = image.getRGB(i,j); Object data = resizeImag.getRaster().getDataElements(i, j, null);//获取该点像素，并以object类型表示 int R = resizeImag.getColorModel().getRed(data); int G = resizeImag.getColorModel().getGreen(data); int B = resizeImag.getColorModel().getBlue(data); // Color color = new Color(rgb); rgbArray.putScalar(0,0,i,j,R); rgbArray.putScalar(0,1,i,j,G); rgbArray.putScalar(0,2,i,j,B); //System.out.println(rgbArray[j][i]); } } return rgbArray ; } /** **数据归一化 */ public static INDArray getImageNormalization(INDArray imageArr) { // int maxValueIndex = 255//Nd4j.getExecutioner().execAndReturn(new IAMax(imageArr)).getFinalResult(); // int minValueIndex = //Nd4j.getExecutioner().execAndReturn(new IAMin(imageArr)).getFinalResult(); double maxValue = 255;//imageArr.getDouble(maxValueIndex); double minValue = 0;//imageArr.getDouble(minValueIndex); try { for (int i = 0; i < imageArr.size(0); i++) { for (int j = 0; j < imageArr.size(1); j++) { double r = (imageArr.getDouble(i,j,0) - minValue) / (maxValue - minValue+1); double g = (imageArr.getDouble(i,j,1) - minValue) / (maxValue - minValue+1); double b = (imageArr.getDouble(i,j,2) - minValue) / (maxValue - minValue+1); imageArr.putScalar(i,j,0, r); imageArr.putScalar(i,j,1, g); imageArr.putScalar(i,j,2, b); } } } catch (Exception e) { e.printStackTrace(); } return imageArr; } /** **数据归一化 */ public static INDArray getNormalization(INDArray imageArr) { int maxValueIndex = Nd4j.getExecutioner().execAndReturn(new IAMax(imageArr)).getFinalResult(); int minValueIndex = Nd4j.getExecutioner().execAndReturn(new IAMin(imageArr)).getFinalResult(); INDArray normlization = Nd4j.zeros(1,3,imageArr.size(2),imageArr.size(3)); double maxValue = imageArr.getDouble(maxValueIndex); double minValue = imageArr.getDouble(minValueIndex); try { for (int i = 0; i < imageArr.size(2); i++) { for (int j = 0; j < imageArr.size(3); j++) { double r = (imageArr.getDouble(0,0,i,j) - minValue) / (maxValue - minValue+1); double g = (imageArr.getDouble(0,1,i,j) - minValue) / (maxValue - minValue+1); double b = (imageArr.getDouble(0,2,i,j) - minValue) / (maxValue - minValue+1); normlization.putScalar(0,0,i,j, r); normlization.putScalar(0,1,i,j, g); normlization.putScalar(0,2,i,j, b); } } } catch (Exception e) { e.printStackTrace(); } return normlization; } public static INDArray kmean(INDArray imageArr,int number){ int rows = imageArr.size(0); int columns = imageArr.size(1); INDArray centers = Nd4j.zeros(number,3); INDArray result = Nd4j.zeros(rows,columns); for (int i = 0;i < number;i++){ Random random = new Random(); int row = random.nextInt(rows); int column = random.nextInt(columns); centers.putScalar(i,0,imageArr.getDouble(row,column,0)); centers.putScalar(i,1,imageArr.getDouble(row,column,1)); centers.putScalar(i,2,imageArr.getDouble(row,column,2)); } double threshold = 0.1; double distance = 1; Map<Integer,List<INDArray>> cluster = null ; while(distance > threshold){ cluster = new HashMap<>() ; for(int i=0;i<rows;i++) { for (int j = 0; j < columns; j++) { INDArray rgb = imageArr.get(NDArrayIndex.point(i), NDArrayIndex.point(j), NDArrayIndex.all()); INDArray colorCenter = centers; INDArray color = rgb.reshape(1,3); int label = getClassCode(colorCenter,color); if(cluster.containsKey(label)){ List<INDArray> list = cluster.get(label); list.add(color); cluster.put(label,list); } else { List<INDArray> listColor = new ArrayList<>(); listColor.add(color); cluster.put(label,listColor); } result.putScalar(i,j,label); } } INDArray newCenters = getCenters(cluster,number); distance = getDisTance(newCenters,centers); centers = newCenters; System.out.println(distance); } System.out.println(centers); return result; } public static INDArray kmeanByCenter(INDArray imageArr,int number,INDArray centers){ int rows = imageArr.size(0); int columns = imageArr.size(1); INDArray result = Nd4j.zeros(rows,columns); double threshold = 0.000000001; double distance = 1; Map<Integer,List<INDArray>> cluster = new HashMap<>() ; for(int i=0;i<rows;i++) { for (int j = 0; j < columns; j++) { INDArray rgb = imageArr.get(NDArrayIndex.point(i), NDArrayIndex.point(j), NDArrayIndex.all()); INDArray colorCenter = centers; INDArray color = rgb.reshape(1,3); int label = getClassCode(colorCenter,color); if(cluster.containsKey(label)){ List<INDArray> list = cluster.get(label); list.add(color); cluster.put(label,list); } else { List<INDArray> listColor = new ArrayList<>(); listColor.add(color); cluster.put(label,listColor); } result.putScalar(i,j,label); } } INDArray newCenters = getCenters(cluster,number); distance = getDisTance(newCenters,centers); System.out.println(distance); return result; } public static BufferedImage arrayToGreyImage(INDArray sourceArray,int number){ int width = sourceArray.rows(); int height = sourceArray.columns(); BufferedImage targetImage = new BufferedImage(width, height, BufferedImage.TYPE_INT_RGB); Integer[] colors = new Integer[number]; for(int i = 0;i<number;i++){ Random random = new Random(); Integer red = random.nextInt(256); Integer green = random.nextInt(256); Integer blue = random.nextInt(256); Integer rgb = (65536 * blue) + (256 * green) + (red); colors[i] = rgb; } for(int i=0;i < width ;i++) { for (int j = 0;j < height ;j++){ int index = sourceArray.getInt(i,j)-1; targetImage.setRGB(i, j, colors[index]); } } return targetImage; } public static INDArray getRGBTOHSV(INDArray image){ int height = image.size(0); int width = image.size(1); INDArray imageHSV = Nd4j.zeros(height, width,3); for(int i = 0;i<height;i++){ for(int j=0;j<width;j++){ INDArray rgb = image.get(NDArrayIndex.point(i), NDArrayIndex.point(j), NDArrayIndex.all()); INDArray row = rgb.reshape(1,3); int maxIndex = Nd4j.getExecutioner().execAndReturn(new IAMax(row)).getFinalResult(); int minIndex = Nd4j.getExecutioner().execAndReturn(new IAMin(row)).getFinalResult(); int max = row.getInt(maxIndex); int min = row.getInt(minIndex); int R = row.getInt(0); int G = row.getInt(1); int B = row.getInt(2); float delta = max - min; float H,S,V; if(delta > 0){ if(R == max){ H = (G-B)/delta; } else if(G == max){ H = 2f + (B-R)/(delta); } else { H = 4f + (R-G)/delta; } H = H*60f; if(H < 0){ H += 360f; } S = delta/max; } else { H = 0f; S = 0f; } V = max*1.0f/255f; imageHSV.putScalar(i,j,0,(double)H); imageHSV.putScalar(i,j,1,(double)V); imageHSV.putScalar(i,j,2,(double)S); } } return imageHSV; } private static int getClassCode(INDArray X,INDArray Y){ INDArray dXY = X.subRowVector(Y); double minValue = 999; int index = 0; for (int i = 0;i<dXY.size(0);i++) { double value = dXY.getDouble(i,0)*dXY.getDouble(i,0)+ dXY.getDouble(i,1)*dXY.getDouble(i,1)+dXY.getDouble(i,2)*dXY.getDouble(i,2); double delta = Math.sqrt(value); if(minValue>delta){ minValue = delta; index = i; } } return index+1; } private static double getDisTance(INDArray X,INDArray Y){ INDArray dXY = X.sub(Y); double minValue = 999; int index = 0; for(int i=0;i<dXY.size(0);i++){ double value = dXY.getDouble(i,0)*dXY.getDouble(i,0)+ dXY.getDouble(i,1)*dXY.getDouble(i,1)+dXY.getDouble(i,2)*dXY.getDouble(i,2); double delta = Math.sqrt(value); if(minValue>delta){ minValue = delta; } } return minValue; } private static INDArray getCenters(Map<Integer,List<INDArray>> clusters,int number){ INDArray centers = Nd4j.zeros(number,3); int label = 1; while (label <= number) { List<INDArray> listValue = clusters.get(label); INDArray sumColor = Nd4j.zeros(1,3); for(INDArray color:listValue){ sumColor.addi(color); } double count = listValue.size(); double r = sumColor.getDouble(0,0)/count; double g = sumColor.getDouble(0,1)/count; double b = sumColor.getDouble(0,2)/count; int row = label - 1; centers.putScalar(row,0,r); centers.putScalar(row,1,g); centers.putScalar(row,2,b); label++; } return centers; } }
1万+
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
