添加多种类干食切图

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

@@ -102,7 +102,6 @@ public class Convolution extends Frequency {
         if (foodTypes == null) {
             foods = new int[]{type};
         } else {
-            foods = new int[foodTypes.length + 1];
             boolean isLife = false;
             for (int myType : foodTypes) {
                 if (type == myType) {
@@ -111,9 +110,14 @@ public class Convolution extends Frequency {
                 }
             }
             if (!isLife) {
+                foods = new int[foodTypes.length + 1];
+            } else {
+                foods = new int[foodTypes.length];
+            }
             for (int i = 0; i < foodTypes.length; i++) {
                 foods[i] = foodTypes[i];
             }
+            if (!isLife) {
                 foods[foodTypes.length] = type;
             }
         }

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

@@ -56,7 +56,7 @@ public class CutFood {
         return sigma / (x * y);
     }
 
-    private boolean getAvgPro(ThreeChannelMatrix threeChannelMatrix, GMClustering gm) throws Exception {
+    private void getAvgPro(ThreeChannelMatrix threeChannelMatrix) throws Exception {
         int size = templeConfig.getFood().getRegionSize();
         Matrix matrixR = threeChannelMatrix.getMatrixR();
         Matrix matrixG = threeChannelMatrix.getMatrixG();
@@ -64,35 +64,25 @@ public class CutFood {
         int x = matrixR.getX();
         int y = matrixR.getY();
         regionMap = new Matrix(x / size, y / size);
-        double sigmaPro = 0;
-        double sigmaOther = 0;
         for (int i = 0; i <= x - size; i += size) {
             for (int j = 0; j <= y - size; j += size) {
                 double r = getAvg(matrixR.getSonOfMatrix(i, j, size, size));
                 double g = getAvg(matrixG.getSonOfMatrix(i, j, size, size));
                 double b = getAvg(matrixB.getSonOfMatrix(i, j, size, size));
                 double[] rgb = new double[]{r / 255, g / 255, b / 255};
-                TypePro typePro = getType(rgb);
-                int index = typePro.index;//该区域所属类别
-                sigmaPro = sigmaPro + typePro.pro;
-                if (gm != null) {
-                    sigmaOther = sigmaOther + gm.getProbabilityDensity(rgb);//概率密度
-                }
+                int index = getType(rgb);
                 regionMap.setNub(i / size, j / size, index);
             }
         }
-        return sigmaPro > sigmaOther;
     }
 
-    public Map<Integer, Integer> getTypeNub(ThreeChannelMatrix threeChannelMatrix,
-                                            GMClustering gm) throws Exception {
-        if (getAvgPro(threeChannelMatrix, gm)) {//进行二次判定
-            return getTypeNub();
-        }
-        return null;
+    public Map<Integer, Integer> getTypeNub(ThreeChannelMatrix threeChannelMatrix
+            , int myType) throws Exception {
+        getAvgPro(threeChannelMatrix);
+        return getTypeNub(myType);
     }
 
-    private Map<Integer, Integer> getTypeNub() throws Exception {
+    private Map<Integer, Integer> getTypeNub(int myType) throws Exception {
         int size = templeConfig.getFood().getRegionSize();
         double s = Math.pow(size, 2);
         int xr = regionMap.getX();
@@ -113,7 +103,7 @@ public class CutFood {
             int type = gmBody.getType();
             if (type != 1) {//背景直接过滤
                 double oneSize = meanMap.get(type).getRegionSize();
-                int nub = (int) (regionSize / oneSize);
+                int nub = (int) (regionSize / oneSize);//数量
                 if (nub > 0) {
                     if (gmTypeNub.containsKey(type)) {
                         int myNub = gmTypeNub.get(type);
@@ -126,6 +116,9 @@ public class CutFood {
                 }
             }
         }
+        if (gmTypeNub.size() == 0) {
+            gmTypeNub.put(myType, 1);
+        }
         return gmTypeNub;
     }
 
@@ -140,17 +133,7 @@ public class CutFood {
         return isInsert;
     }
 
-    class TypePro {
-        private int index;
-        private double pro;//概率密度
-
-        TypePro(int index, double pro) {
-            this.index = index;
-            this.pro = pro;
-        }
-    }
-
-    private TypePro getType(double[] rgb) throws Exception {
+    private int getType(double[] rgb) throws Exception {
         int index = 0;
         double max = 0;
         for (Map.Entry<Integer, GMClustering> entry : meanMap.entrySet()) {
@@ -163,10 +146,8 @@ public class CutFood {
         }
         if (max < 2) {
             index = 1;
-            max = 0;
         }
-
-        return new TypePro(index, max);
+        return index;
     }
 
     public void study(int type, ThreeChannelMatrix threeChannelMatrix) throws Exception {

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

@@ -167,11 +167,13 @@ public class Operation {//进行计算
                     }
                 }
             }
-            CutFood cutFood = new CutFood(templeConfig, food.getFoodMeanMap());
-            if (isFood) {//一次判定就属于干食，则无需进行二次判定
-                regionBody.setTypeNub(cutFood.getTypeNub(threeChannelMatrix1, null));
-            } else {//一次判定属于非干食，则进行二次判定
-                regionBody.setTypeNub(cutFood.getTypeNub(threeChannelMatrix1, food.getNotFoodMeanMap().get(id)));
+            if (isFood) {//判断类别，获取积分
+                Map<Integer, GMClustering> foods = food.getFoodMeanMap();
+                Map<Integer, GMClustering> test = new HashMap<>();
+                test.put(1, foods.get(1));
+                test.put(id, foods.get(id));
+                CutFood cutFood = new CutFood(templeConfig, test);
+                regionBody.setTypeNub(cutFood.getTypeNub(threeChannelMatrix1, id));
             }
             regionBody.setType(id);
             System.out.println("类别" + id);

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

@@ -22,6 +22,18 @@ public class RGBNorm {
     private double gmParameter;//gm混合系数
     private double probabilitySigma = 0;//后验概率求和
 
+    public double getVarAll() {
+        return varAll;
+    }
+
+    public Matrix getAvgMatrix() {
+        return avgMatrix;
+    }
+
+    public Matrix getVarMatrix() {
+        return varMatrix;
+    }
+
     public List<double[]> getRgbs() {
         return rgbs;
     }

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

@@ -707,7 +707,6 @@ public class TempleConfig {
                                 food.getNotFoodMeanMap().put(key, gmClustering);
                             }
                         }
-                        //
                         DimensionMappingStudy dimensionAll = new DimensionMappingStudy(this);
                         dimensionAll.start();//生成映射层，并将已经保存的knn特征进行映射
                         food.setDimensionMappingStudy(dimensionAll);

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

@@ -53,7 +53,7 @@ public class GMClustering extends MeanClustering {
             RGBNorm rgbNorm = new RGBNorm();
             matrices.add(rgbNorm);
             for (int j = i; j < i + size; j++) {
-                feature[j - i] = matrix.getNumber(0, i);
+                feature[j - i] = matrix.getNumber(0, j);
             }
             rgbNorm.insertFeature(feature);
         }

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

@@ -48,7 +48,7 @@ public class FeatureMapping extends Frequency implements PsoFunction {
     @Override
     public double getResult(double[] parameter, int id) throws Exception {
 
-        return featuresMapping2(parameter);
+        return featuresMapping(parameter);
     }
 
     private double[] mapping(Matrix matrix, double[] parameter) throws Exception {

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

@@ -1,6 +1,5 @@
 package org.wlld.param;
 
-import org.wlld.imageRecognition.CutFood;
 import org.wlld.imageRecognition.border.GMClustering;
 import org.wlld.imageRecognition.segmentation.DimensionMappingStudy;
 import org.wlld.imageRecognition.segmentation.RgbRegression;

src/test/java/coverTest/FoodTest.java

@@ -80,7 +80,6 @@ public class FoodTest {
         ThreeChannelMatrix threeChannelMatrix1 = picture.getThreeMatrix(a);
         Watershed watershed = new Watershed(threeChannelMatrix1, specificationsList, templeConfig);
         List<RegionBody> regionList = watershed.rainfall();
-        System.out.println("==");
     }
 
     private static void look(ThreeChannelMatrix threeChannelMatrix, TempleConfig templeConfig, List<Specifications> specifications,
@@ -95,9 +94,6 @@ public class FoodTest {
             int maxY = regionBody.getMaxY();
             int xSize = maxX - minX;
             int ySize = maxY - minY;
-            ThreeChannelMatrix threeChannelMatrix1 = convolution.getRegionMatrix(threeChannelMatrix, minX, minY, xSize, ySize);
-            cutFood.getTypeNub(threeChannelMatrix1, null);
-            System.out.println("结束===================");
         }
     }