加入其他几种特征，主要是对比度有用

ImageDetect/CMakeLists.txt

@@ -40,7 +40,7 @@ endif (USE_CROSS)
 project (ImageDetect)
 
 # The version number.
-set (ImageDetect_VERSION_MAJOR 5)
+set (ImageDetect_VERSION_MAJOR 6)
 set (ImageDetect_VERSION_MINOR 0)
 # configure a header file to pass some of the CMake settings
 # to the source code

ImageDetect/ImageDetect.cxx

@@ -57,12 +57,84 @@ void detect_saturation(cv::Mat input_img, float& satu)
     satu = mean.val[0];  
 }
 
+void detect_graydiff(cv::Mat img, float& CON, float& ASM, float& ENT, float& MEAN)
+{
+	Mat mi;
+	int row = img.rows;
+	int col = img.cols;
+	//CON=i*i*P(I)
+	//计算i
+	mi.create(row, col, img.type());
+	for (int i = 0; i < row-1; i++)
+		for (int j = 0; j < col; j++)
+		{   
+			//cout << "BEF" << int(img.ptr<uchar>(i)[j]) << endl;
+			mi.at<Vec3b>(i, j) = img.at<Vec3b>(i + 1, j) - img.at<Vec3b>(i, j);
+			//cout << "IMF:" << int(mi.ptr<uchar>(i)[j]) << endl;
+		}
+	//计算p(I)  差值的概率
+	int new_cols = mi.cols;
+	int new_rows = mi.rows;
+	int gray[256] = { 0 };
+	double gray_prob[256] = { 0 };
+	int num = 0;// 像素的总个数
+	//统计直方图各个像素灰度值
+	for (int i = 0; i < mi.rows; i++)
+	{
+		uchar *p = mi.ptr<uchar>(i);
+		for (int j = 0; j < mi.cols; j++)
+		{
+			int value = p[j];
+			gray[value]++;
+			num++;
+		}
+	}
+	//计算直方图概率分布
+	for (int i = 0; i < 256; i++)
+	{
+		gray_prob[i] = ((double)gray[i] / num);
+	}
+	CON = 0.0;
+	ASM = 0.0;
+	ENT = 0.0;
+	MEAN = 0.0;
+	double v = 1.0/ 255.0;
+	for (int i = 0; i <= 255; i++)
+	{
+		CON += i*i*gray_prob[i];//计算CON
+		ASM += gray_prob[i] * gray_prob[i];//计算角度方向二阶矩
+		if (gray_prob[i]!=0)
+		ENT += (0-gray_prob[i] * log10(gray_prob[i]));//计算熵
+		MEAN += v*i*gray_prob[i];//计算平均值
+	}
+	return; 
+}
+
+void detect_definition(cv::Mat input_img, float& sb, float& lap, float& dev)
+{
+    Mat imgGray, imgSobel, imgLap, imgDev, meanValueImage;
+    cvtColor(input_img, imgGray, CV_BGR2GRAY);
+    Sobel(imgGray, imgSobel, CV_16U, 1, 1);
+    Laplacian(imgGray, imgLap, CV_16U);
+    meanStdDev(imgGray,meanValueImage,imgDev);
+    sb = mean(imgSobel)[0];
+    lap = mean(imgLap)[0];
+    dev = mean(imgDev)[0];
+}
+
+
 int main(int argc, char *argv[])
 {
     Mat img = imread(argv[1]);
-    float cast, da, satu;
-    detect_brightness(img, cast, da);
-    detect_saturation(img, satu);
-    cout<<cast<<"||"<<da<<"||"<<satu;
+    float cast, da, satu, CON, ASM, ENT, MEAN, sb, lap, dev;
+    Mat img_brightness = img.clone();
+    detect_brightness(img_brightness, cast, da);
+    Mat img_saturation = img.clone();
+    detect_saturation(img_saturation, satu);
+    Mat img_graydiff = img.clone();
+    detect_graydiff(img_graydiff, CON, ASM, ENT, MEAN);
+    Mat img_definition = img.clone();
+    detect_definition(img_definition, sb, lap, dev);
+    cout<<cast<<"||"<<da<<"||"<<satu<<"||"<<CON<<"||"<<ASM<<"||"<<ENT<<"||"<<MEAN<<"||"<<sb<<"||"<<lap<<"||"<<dev;
     return 0;
 }

learn/views.py

@@ -61,8 +61,15 @@ def upload(request):
             cast = result[1].split("||")[0]
             da = result[1].split("||")[1]
             satu = result[1].split("||")[2]
+            CON = result[1].split("||")[3]
+            ASM = result[1].split("||")[4]
+            ENT = result[1].split("||")[5]
+            MEAN = result[1].split("||")[6]
+            sb = result[1].split("||")[7]
+            lap = result[1].split("||")[8]
+            dev = result[1].split("||")[9]
         print(cast, da, satu)
-        mesg = "颜色分布(0-1合理):{}   分布方差(负数偏暗):{}   色彩饱和度(零为补光):{}".format(cast, da, satu)
+        mesg = "颜色分布(0-1合理):{}&nbsp;&nbsp;&nbsp;分布方差(负数偏暗):{}&nbsp;&nbsp;&nbsp;色彩饱和度(零为补光):{}<br/>对比度:{}&nbsp;&nbsp;&nbsp;角度方向二阶矩:{}&nbsp;&nbsp;&nbsp;熵:{}&nbsp;&nbsp;&nbsp;平均值:{}&nbsp;&nbsp;&nbsp;Sobel:{}&nbsp;&nbsp;&nbsp;Lap:{}&nbsp;&nbsp;&nbsp;dev:{}".format(cast, da, satu, CON, ASM, ENT, MEAN,sb,lap,dev)
         if(int(m_class)==1):
             if(int(m_model)==1): #小模型
                 RR = DL.run(weights=(curPath+"/yolov5/v3s.pt"), source=img_path, project=(curPath+"/static/detected"))
@@ -103,8 +110,15 @@ def dlurl(request):
             cast = result[1].split("||")[0]
             da = result[1].split("||")[1]
             satu = result[1].split("||")[2]
+            CON = result[1].split("||")[3]
+            ASM = result[1].split("||")[4]
+            ENT = result[1].split("||")[5]
+            MEAN = result[1].split("||")[6]
+            sb = result[1].split("||")[7]
+            lap = result[1].split("||")[8]
+            dev = result[1].split("||")[9]
         print(cast, da, satu)
-        mesg = "颜色分布(0-1合理):{}&nbsp;&nbsp;&nbsp;分布方差(负数偏暗):{}&nbsp;&nbsp;&nbsp;色彩饱和度(零为补光):{}".format(cast, da, satu)
+        mesg = "颜色分布(0-1合理):{}&nbsp;&nbsp;&nbsp;分布方差(负数偏暗):{}&nbsp;&nbsp;&nbsp;色彩饱和度(零为补光):{}<br/>对比度:{}&nbsp;&nbsp;&nbsp;角度方向二阶矩:{}&nbsp;&nbsp;&nbsp;熵:{}&nbsp;&nbsp;&nbsp;平均值:{}&nbsp;&nbsp;&nbsp;Sobel:{}&nbsp;&nbsp;&nbsp;Lap:{}&nbsp;&nbsp;&nbsp;dev:{}".format(cast, da, satu, CON, ASM, ENT, MEAN,sb,lap,dev)
         if(int(m_class)==1):
             if(int(m_model)==1): #小模型
                 RR = DL.run(weights=(curPath+"/yolov5/v3s.pt"), source=img_path, project=(curPath+"/static/detected"))