增加局部纹理聚类

pom.xml

@@ -6,7 +6,7 @@
 
     <groupId>com.wlld</groupId>
     <artifactId>easyAi</artifactId>
-    <version>1.1.2</version>
+    <version>1.0.9</version>
 
     <name>easyAi</name>
     <!-- FIXME change it to the project's website -->

src/main/java/org/wlld/MatrixTools/MatrixOperation.java

@@ -330,6 +330,7 @@ public class MatrixOperation {
             mathMul(myMatrix, def);
             return myMatrix;
         } else {
+            System.out.println(matrixs.getString());
             throw new Exception("this matrixs do not have InverseMatrixs");
         }
     }

src/main/java/org/wlld/config/Kernel.java

@@ -10,7 +10,7 @@ public class Kernel {
     private static final String All_Number2 = "[-1,0,-1]#[0,4,0]#[-1,0,-1]#";
     public static final int Region_Nub = 60;//一张图有多少份
     public static final double th = 0.88;//分水岭灰度阈值
-    public static final double rgbN = 442.0;//RGB范数归一化最大值
+    public static final double rgbN = 442.0;//442.0;//RGB范数归一化最大值
     public static Matrix Vertical;
     public static Matrix Horizontal;
     public static Matrix All;

src/main/java/org/wlld/imageRecognition/Convolution.java

@@ -169,9 +169,9 @@ public class Convolution extends Frequency {
         Matrix matrixR = threeChannelMatrix.getMatrixR();
         Matrix matrixG = threeChannelMatrix.getMatrixG();
         Matrix matrixB = threeChannelMatrix.getMatrixB();
-        matrixR = late(matrixR, poolSize);
+//        matrixR = late(matrixR, poolSize);
-        matrixG = late(matrixG, poolSize);
+//        matrixG = late(matrixG, poolSize);
-        matrixB = late(matrixB, poolSize);
+//        matrixB = late(matrixB, poolSize);
         RGBSort rgbSort = new RGBSort();
         int x = matrixR.getX();
         int y = matrixR.getY();
@@ -187,9 +187,9 @@ public class Convolution extends Frequency {
         Collections.sort(rgbNorms, rgbSort);
         List<Double> features = new ArrayList<>();
         for (int i = 0; i < sqNub; i++) {
-            //double[] rgb = rgbNorms.get(i).getRgb();
+            double[] rgb = rgbNorms.get(i).getRgb();
-            RgbRegression rgbRegression = rgbNorms.get(i).getRgbRegression();
+            // RgbRegression rgbRegression = rgbNorms.get(i).getRgbRegression();
-            double[] rgb = new double[]{rgbRegression.getWr(), rgbRegression.getWg(), rgbRegression.getB()};
+            //double[] rgb = new double[]{rgbRegression.getWr(), rgbRegression.getWg(), rgbRegression.getB()};
             for (int j = 0; j < 3; j++) {
                 features.add(rgb[j]);
             }
@@ -200,44 +200,45 @@ public class Convolution extends Frequency {
         return features;
     }
 
-    private void regression(XYBody xyBody) {
+    public List<Double> getCenterTexture(ThreeChannelMatrix threeChannelMatrix, int size, int poolSize, TempleConfig templeConfig
-        //计算当前图形的线性回归
+            , int sqNub) throws Exception {
-        RegressionBody regressionBody = new RegressionBody();
+        RGBSort rgbSort = new RGBSort();
-        regressionBody.lineRegression(xyBody.getY(), xyBody.getX(), this);
+        MeanClustering meanClustering = new MeanClustering(sqNub, templeConfig);
-        xyBody.setRegressionBody(regressionBody);
+        Matrix matrixR = threeChannelMatrix.getMatrixR();
-    }
+        Matrix matrixG = threeChannelMatrix.getMatrixG();
-
+        Matrix matrixB = threeChannelMatrix.getMatrixB();
-    public XYBody imageTrance(Matrix matrix, int size) throws Exception {//矩阵和卷积核大小
+        int xn = matrixR.getX();
-        int xn = matrix.getX();
+        int yn = matrixR.getY();
-        int yn = matrix.getY();
+        for (int i = 0; i <= xn - size; i += size) {
-        int xSize = xn / size;//求导后矩阵的行数
+            for (int j = 0; j <= yn - size; j += size) {
-        int ySize = yn / size;//求导后矩阵的列数
+                Matrix sonR = late(matrixR.getSonOfMatrix(i, j, size, size), poolSize);
-        double[] Y = new double[xSize * ySize];
+                Matrix sonG = late(matrixG.getSonOfMatrix(i, j, size, size), poolSize);
-        double[] X = new double[xSize * ySize];
+                Matrix sonB = late(matrixB.getSonOfMatrix(i, j, size, size), poolSize);
-        double rgbN = Kernel.rgbN;
+                int tSize = sonR.getX();
-        for (int i = 0; i < xn - size; i += size) {
+                int kSize = sonR.getY();
-            for (int j = 0; j < yn - size; j += size) {
+                double[] rgb = new double[tSize * kSize * 3];
-                Matrix matrix1 = matrix.getSonOfMatrix(i, j, size, size);
+                for (int t = 0; t < tSize; t++) {
-                double[] nubs = new double[size * size];//平均值数组
+                    for (int k = 0; k < kSize; k++) {
-                for (int t = 0; t < size; t++) {
+                        int index = t * kSize + k;
-                    for (int k = 0; k < size; k++) {
+                        rgb[index] = sonR.getNumber(t, k);
-                        double nub = matrix1.getNumber(t, k) / rgbN;
+                        rgb[tSize * kSize + index] = sonG.getNumber(t, k);
-                        nubs[t * size + k] = nub;
+                        rgb[tSize * kSize * 2 + index] = sonB.getNumber(t, k);
                     }
                 }
-                double avg = average(nubs);//平均值
+                meanClustering.setColor(rgb);
-                //double dc = frequency.dcByAvg(nubs, avg);//当前离散系数
-                double va = varianceByAve(nubs, avg);//方差
-                //离散系数作为X，AVG作为Y
-                int t = i / size * ySize + j / size;
-                Y[t] = avg;
-                X[t] = va;
             }
         }
-        XYBody xyBody = new XYBody();
+        meanClustering.start();//开始聚类
-        xyBody.setX(X);
+        List<RGBNorm> rgbNorms = meanClustering.getMatrices();
-        xyBody.setY(Y);
+        Collections.sort(rgbNorms, rgbSort);
-        return xyBody;
+        List<Double> features = new ArrayList<>();
+        for (int i = 0; i < sqNub; i++) {
+            double[] rgb = rgbNorms.get(i).getRgb();
+            for (int j = 0; j < rgb.length; j++) {
+                features.add(rgb[j]);
+            }
+        }
+        return features;
     }
 
 
@@ -260,11 +261,13 @@ public class Convolution extends Frequency {
         Matrix matrixR = threeChannelMatrix.getMatrixR().getSonOfMatrix(x, y, xSize, ySize);
         Matrix matrixG = threeChannelMatrix.getMatrixG().getSonOfMatrix(x, y, xSize, ySize);
         Matrix matrixB = threeChannelMatrix.getMatrixB().getSonOfMatrix(x, y, xSize, ySize);
+        Matrix matrixH = threeChannelMatrix.getH().getSonOfMatrix(x, y, xSize, ySize);
+        Matrix matrixRGB = threeChannelMatrix.getMatrixRGB().getSonOfMatrix(x, y, xSize, ySize);
         threeChannelMatrix1.setMatrixR(matrixR);
         threeChannelMatrix1.setMatrixG(matrixG);
         threeChannelMatrix1.setMatrixB(matrixB);
-        threeChannelMatrix1.setH(threeChannelMatrix.getH());
+        threeChannelMatrix1.setH(matrixH);
-        threeChannelMatrix1.setMatrixRGB(threeChannelMatrix.getMatrixRGB());
+        threeChannelMatrix1.setMatrixRGB(matrixRGB);
         return threeChannelMatrix1;
     }
 
@@ -359,7 +362,7 @@ public class Convolution extends Frequency {
         return myMatrix;
     }
 
-    protected Matrix late(Matrix matrix, int size) throws Exception {//池化处理
+    public Matrix late(Matrix matrix, int size) throws Exception {//池化处理
         int xn = matrix.getX();
         int yn = matrix.getY();
         int x = xn / size;//求导后矩阵的行数
@@ -380,7 +383,7 @@ public class Convolution extends Frequency {
                         }
                     }
                 }
-                maxNub = ArithUtil.div(sigma, n);
+                maxNub = sigma / n;
                 //迟化的最大值是 MAXNUB
                 myMatrix.setNub(i / size, j / size, maxNub);
             }

src/main/java/org/wlld/imageRecognition/MeanClustering.java

@@ -96,7 +96,7 @@ public class MeanClustering {
             for (int i = 0; i < speciesQuantity; i++) {//初始化均值向量
                 int index = random.nextInt(matrixList.size());
                 double[] rgb = matrixList.get(index);
-                RGBNorm rgbNorm = new RGBNorm(rgb);
+                RGBNorm rgbNorm = new RGBNorm(rgb, length);
                 //要进行深度克隆
                 matrices.add(rgbNorm);
             }
@@ -111,7 +111,7 @@ public class MeanClustering {
                     break;
                 }
             }
-            startRegression();//开始进行回归
+            // startRegression();//开始进行回归
         } else {
             throw new Exception("matrixList number less than 2");
         }

src/main/java/org/wlld/imageRecognition/Operation.java

@@ -7,7 +7,6 @@ import org.wlld.config.Classifier;
 import org.wlld.config.StudyPattern;
 import org.wlld.i.OutBack;
 import org.wlld.imageRecognition.border.*;
-import org.wlld.imageRecognition.modelEntity.TrayBody;
 import org.wlld.imageRecognition.segmentation.RegionBody;
 import org.wlld.imageRecognition.segmentation.RgbRegression;
 import org.wlld.imageRecognition.segmentation.Specifications;
@@ -64,10 +63,10 @@ public class Operation {//进行计算
         Matrix matrixR = threeChannelMatrix.getMatrixR();
         Matrix matrixG = threeChannelMatrix.getMatrixG();
         Matrix matrixB = threeChannelMatrix.getMatrixB();
-        Matrix matrixRGB = threeChannelMatrix.getMatrixRGB();
+        //Matrix matrixRGB = threeChannelMatrix.getMatrixRGB();
         Random random = new Random();
-        int x = matrixRGB.getX();
+        int x = matrixR.getX();
-        int y = matrixRGB.getY();
+        int y = matrixR.getY();
         int size = templeConfig.getFood().getRegressionNub();
         RgbRegression rgbRegression = new RgbRegression(size);
         for (int i = 0; i < size; i++) {
@@ -88,11 +87,9 @@ public class Operation {//进行计算
         Matrix matrixR = threeChannelMatrix.getMatrixR();
         Matrix matrixG = threeChannelMatrix.getMatrixG();
         Matrix matrixB = threeChannelMatrix.getMatrixB();
-        Matrix matrixRGB = threeChannelMatrix.getMatrixRGB();
         threeChannelMatrix.setMatrixR(matrixR.getSonOfMatrix(x, y, xSize, ySize));
         threeChannelMatrix.setMatrixG(matrixG.getSonOfMatrix(x, y, xSize, ySize));
         threeChannelMatrix.setMatrixB(matrixB.getSonOfMatrix(x, y, xSize, ySize));
-        threeChannelMatrix.setMatrixRGB(matrixRGB.getSonOfMatrix(x, y, xSize, ySize));
     }
 
     public RegionBody colorStudy(ThreeChannelMatrix threeChannelMatrix, int tag, List<Specifications> specificationsList) throws Exception {
@@ -107,15 +104,15 @@ public class Operation {//进行计算
             int xSize = maxX - minX;
             int ySize = maxY - minY;
             ThreeChannelMatrix threeChannelMatrix1 = convolution.getRegionMatrix(threeChannelMatrix, minX, minY, xSize, ySize);
-            //convolution.filtering(threeChannelMatrix1);//光照过滤
             int times = templeConfig.getFood().getTimes();
             for (int i = 0; i < times; i++) {
-                List<Double> feature = convolution.getCenterColor(threeChannelMatrix1, templeConfig.getPoolSize(),
+//                List<Double> feature = convolution.getCenterColor(threeChannelMatrix1, templeConfig.getPoolSize(),
-                        templeConfig.getFeatureNub(), templeConfig);
+//                        templeConfig.getFeatureNub(), templeConfig);
+                List<Double> feature = convolution.getCenterTexture(threeChannelMatrix1, templeConfig.getFood().getRegionSize(),
+                        templeConfig.getPoolSize(), templeConfig, templeConfig.getFeatureNub());
                 if (templeConfig.isShowLog()) {
                     System.out.println(tag + ":" + feature);
                 }
-                //System.out.println("=====================================");
                 int classifier = templeConfig.getClassifier();
                 switch (classifier) {
                     case Classifier.DNN:
@@ -177,9 +174,10 @@ public class Operation {//进行计算
             int xSize = maxX - minX;
             int ySize = maxY - minY;
             ThreeChannelMatrix threeChannelMatrix1 = convolution.getRegionMatrix(threeChannelMatrix, minX, minY, xSize, ySize);
-            //convolution.filtering(threeChannelMatrix1);//光照过滤
+//            List<Double> feature = convolution.getCenterColor(threeChannelMatrix1, templeConfig.getPoolSize(),
-            List<Double> feature = convolution.getCenterColor(threeChannelMatrix1, templeConfig.getPoolSize(),
+//                    templeConfig.getFeatureNub(), templeConfig);
-                    templeConfig.getFeatureNub(), templeConfig);
+            List<Double> feature = convolution.getCenterTexture(threeChannelMatrix1, templeConfig.getFood().getRegionSize(),
+                    templeConfig.getPoolSize(), templeConfig, templeConfig.getFeatureNub());
             if (templeConfig.isShowLog()) {
                 System.out.println(feature);
             }

src/main/java/org/wlld/imageRecognition/RGBNorm.java

@@ -7,13 +7,14 @@ import java.util.ArrayList;
 import java.util.List;
 
 public class RGBNorm {
-    private double[] rgbAll = new double[3];
+    private double[] rgbAll;
     private double norm;
     private int nub;
-    private double[] rgb = new double[3];
+    private double[] rgb;
     private double[] rgbUp;
     private List<double[]> rgbs = new ArrayList<>();
     private RgbRegression rgbRegression;
+    private int len;
 
     public List<double[]> getRgbs() {
         return rgbs;
@@ -27,7 +28,10 @@ public class RGBNorm {
         this.rgbRegression = rgbRegression;
     }
 
-    RGBNorm(double[] rgb) {
+    RGBNorm(double[] rgb, int len) {
+        this.len = len;
+        rgbAll = new double[len];
+        this.rgb = new double[len];
         this.rgbUp = rgb;
     }
 
@@ -36,12 +40,11 @@ public class RGBNorm {
     }
 
     public void clear() {
-        rgbAll = new double[3];
+        rgbAll = new double[len];
         nub = 0;
         for (int i = 0; i < rgb.length; i++) {
             rgbUp[i] = rgb[i];
         }
-        rgbs.clear();
         //System.out.println("clear==" + Arrays.toString(rgbUp));
     }
 
@@ -76,7 +79,6 @@ public class RGBNorm {
         for (int i = 0; i < rgb.length; i++) {
             rgbAll[i] = rgbAll[i] + rgb[i];
         }
-        rgbs.add(rgb);
         nub++;
     }
 

src/main/java/org/wlld/imageRecognition/border/Knn.java

@@ -85,7 +85,7 @@ public class Knn {//KNN分类器
                 compare(dists, types, dist, type);
             }
         }
-        System.out.println(Arrays.toString(types));
+        //System.out.println(Arrays.toString(types));
         Map<Integer, Integer> map = new HashMap<>();
         for (int i = 0; i < nub; i++) {
             int type = types[i];

src/main/java/org/wlld/imageRecognition/segmentation/Watershed.java

@@ -277,13 +277,13 @@ public class Watershed {
                 regionBodies.add(regionBody);
             }
         }
-        for (RegionBody regionBody : regionBodies) {
+//        for (RegionBody regionBody : regionBodies) {
-            int minX = regionBody.getMinX();
+//            int minX = regionBody.getMinX();
-            int maxX = regionBody.getMaxX();
+//            int maxX = regionBody.getMaxX();
-            int minY = regionBody.getMinY();
+//            int minY = regionBody.getMinY();
-            int maxY = regionBody.getMaxY();
+//            int maxY = regionBody.getMaxY();
-            System.out.println("minX==" + minX + ",minY==" + minY + ",maxX==" + maxX + ",maxY==" + maxY);
+//            System.out.println("minX==" + minX + ",minY==" + minY + ",maxX==" + maxX + ",maxY==" + maxY);
-        }
+//        }
         return regionBodies;
     }
 
@@ -399,8 +399,8 @@ public class Watershed {
     }
 
     private void sigmaPixel() throws Exception {//生成降雨密度图
-        int x = matrix.getX();
+//        int x = matrix.getX();
-        int y = matrix.getY();
+//        int y = matrix.getY();
         int size = xSize * ySize;
         for (int i = 0; i < xSize * regionNub; i += xSize) {
             for (int j = 0; j < ySize * regionNub; j += ySize) {
@@ -429,7 +429,7 @@ public class Watershed {
         int minIdx = 0;
         for (int i = 0; i < array.length; i++) {
             double nub = array[i];
-            if (nub > -1 && nub < mySelf - rainTh && nub < maxRain) {
+            if (nub > -1 && nub < mySelf) {
                 minIdx = minIdx | (1 << i);
             }
         }

src/main/java/org/wlld/param/Food.java

@@ -19,6 +19,15 @@ public class Food {
     private List<RgbRegression> trayBody = new ArrayList<>();//托盘实体参数
     private int regressionNub = 10000;//回归次数
     private double trayTh = 0.1;//托盘回归阈值
+    private int regionSize = 10;//纹理区域大小
+
+    public int getRegionSize() {
+        return regionSize;
+    }
+
+    public void setRegionSize(int regionSize) {
+        this.regionSize = regionSize;
+    }
 
     public double getTrayTh() {
         return trayTh;

src/test/java/coverTest/FoodTest.java

@@ -16,35 +16,15 @@ import org.wlld.param.Food;
 import org.wlld.tools.ArithUtil;
 
 import java.util.ArrayList;
+import java.util.Arrays;
 import java.util.List;
 
 public class FoodTest {
 
     public static void main(String[] args) throws Exception {
-//        ModelParameter parameter = JSON.parseObject(ModelData.DATA, ModelParameter.class);
-//        if (parameter.getKnnVector() != null) {
-//            System.out.println("空的");
-//        } else {
-//            System.out.println("不是空===");
-//        }
         test();
     }
 
-    public static void one(double[] test, double[] right, double[] wrong) {
-        int nub = 0;
-        for (int i = 0; i < test.length; i++) {
-            double test1 = test[i];
-            double right1 = right[i];
-            double wrong1 = wrong[i];
-            double sub1 = Math.abs(ArithUtil.sub(test1, right1));
-            double sub2 = Math.abs(ArithUtil.sub(test1, wrong1));
-            if (sub1 > sub2) {
-                nub++;
-            }
-        }
-        System.out.println(nub);
-    }
-
     public static void test2(TempleConfig templeConfig) throws Exception {
         if (templeConfig == null) {
             ModelParameter parameter = JSON.parseObject(ModelData.DATA, ModelParameter.class);
@@ -53,22 +33,24 @@ public class FoodTest {
         Picture picture = new Picture();
         List<Specifications> specificationsList = new ArrayList<>();
         Specifications specifications = new Specifications();
-        specifications.setMinWidth(300);
+        specifications.setMinWidth(100);
-        specifications.setMinHeight(300);
+        specifications.setMinHeight(100);
         specifications.setMaxWidth(950);
         specifications.setMaxHeight(950);
         specificationsList.add(specifications);
         Operation operation = new Operation(templeConfig);
-        for (int i = 1; i <= 28; i++) {
+        long a = System.currentTimeMillis();
-            ThreeChannelMatrix threeChannelMatrix1 = picture.getThreeMatrix("/Users/lidapeng/Desktop/foodModel/test1/g" + i + ".jpg");
+        for (int i = 1; i <= 1; i++) {
+            ThreeChannelMatrix threeChannelMatrix1 = picture.getThreeMatrix("/Users/lidapeng/Desktop/test/d.jpg");
             List<RegionBody> regionBody = operation.colorLook(threeChannelMatrix1, specificationsList);
+            long b = System.currentTimeMillis() - a;
+            System.out.println(b);
             for (int j = 0; j < regionBody.size(); j++) {
                 RegionBody regionBody1 = regionBody.get(j);
                 System.out.println("minX==" + regionBody1.getMinX() + ",minY==" + regionBody1.getMinY()
                         + ",maxX==" + regionBody1.getMaxX() + ",maxY==" + regionBody1.getMaxY());
-                System.out.println(regionBody.get(j).getType());
+                System.out.println("type==" + regionBody.get(j).getType());
             }
-            System.out.println("===================================" + i);
         }
     }
 
@@ -87,10 +69,10 @@ public class FoodTest {
         //池化比例
         templeConfig.setPoolSize(2);//缩小比例
         //聚类
-        templeConfig.setFeatureNub(3);//聚类特征数量
+        templeConfig.setFeatureNub(5);//聚类特征数量
         //菜品识别实体类
         food.setShrink(20);//缩紧像素
-        food.setTimes(1);//聚类数据增强
+        food.setTimes(2);//聚类数据增强
         food.setRowMark(0.1);//0.12
         food.setColumnMark(0.1);//0.25
         food.setRegressionNub(20000);
@@ -109,21 +91,26 @@ public class FoodTest {
         Operation operation = new Operation(templeConfig);
         List<Specifications> specificationsList = new ArrayList<>();
         Specifications specifications = new Specifications();
-        specifications.setMinWidth(300);
+        specifications.setMinWidth(100);
-        specifications.setMinHeight(300);
+        specifications.setMinHeight(100);
         specifications.setMaxWidth(1000);
         specifications.setMaxHeight(1000);
         specificationsList.add(specifications);
         ThreeChannelMatrix threeChannelMatrix = picture.getThreeMatrix("/Users/lidapeng/Desktop/myDocument/d.jpg");
         operation.setTray(threeChannelMatrix);
         for (int i = 1; i <= 1; i++) {
-            ThreeChannelMatrix threeChannelMatrix1 = picture.getThreeMatrix("/Users/lidapeng/Desktop/myDocument/m.jpg");
+            ThreeChannelMatrix threeChannelMatrix1 = picture.getThreeMatrix("/Users/lidapeng/Desktop/test/a1.jpg");
+            ThreeChannelMatrix threeChannelMatrix2 = picture.getThreeMatrix("/Users/lidapeng/Desktop/test/b.jpg");
+            ThreeChannelMatrix threeChannelMatrix3 = picture.getThreeMatrix("/Users/lidapeng/Desktop/test/c.jpg");
             operation.colorStudy(threeChannelMatrix1, 1, specificationsList);
-            // System.out.println("=======================================" + i);
+            operation.colorStudy(threeChannelMatrix2, 2, specificationsList);
+            operation.colorStudy(threeChannelMatrix3, 3, specificationsList);
+
         }
+
 //        minX==301,minY==430,maxX==854,maxY==920
 //        minX==497,minY==1090,maxX==994,maxY==1520
-        // test2(templeConfig);
+        test2(templeConfig);
     }
 
     public static void study() throws Exception {