Update Sobel Edge Detection.cpp

Author: BBuf

Feature Extraction/README.md

@@ -0,0 +1,3 @@
+# Opencv和C++实现图像特征提取的一些算法
+
+- Sobel Edge Detection.cpp C++实现了OpenCV中的Sobel边缘检测算法。算法原理请看：https://blog.csdn.net/just_sort/article/details/85015054

Feature Extraction/Sobel Edge Detection.cpp

@@ -0,0 +1,79 @@
+#include "opencv2/opencv.hpp"
+#include "iostream"
+#include "algorithm"
+#include "vector"
+#include "stdio.h"
+using namespace std;
+using namespace cv;
+
+const int fac[9]={1, 1, 2, 6, 24, 120, 720, 5040, 40320};
+//Sobel平滑算子
+Mat getSmmoothKernel(int ksize){
+    Mat Smooth = Mat::zeros(Size(ksize, 1), CV_32FC1);
+    for(int i = 0; i < ksize; i++){
+        Smooth.at<float>(0, i) = float(fac[ksize-1]/(fac[i] * fac[ksize-1-i]));
+    }
+    return Smooth;
+}
+//Sobel差分算子
+Mat getDiffKernel(int ksize){
+    Mat Diff = Mat::zeros(Size(ksize, 1), CV_32FC1);
+    Mat preDiff = getSmmoothKernel(ksize-1);
+    for(int i = 0; i < ksize; i++){
+        if(i == 0){
+            Diff.at<float>(0, i) = 1;
+        }else if(i == ksize-1){
+            Diff.at<float>(0, i) = -1;
+        }else{
+            Diff.at<float>(0, i) = preDiff.at<float>(0, i) - preDiff.at<float>(0, i-1);
+        }
+    }
+    return Diff;
+}
+//调用filter2D实现卷积
+void conv2D(InputArray src, InputArray kernel, OutputArray dst, int dep, Point anchor=Point(-1,-1), int borderType=BORDER_DEFAULT){
+    Mat kernelFlip;
+    flip(kernel, kernelFlip, -1);
+    filter2D(src, dst, dep, kernelFlip, anchor, 0.0, borderType);
+}
+//先进行垂直方向的卷积，再进行水平方向的卷积
+void sepConv2D_Y_X(InputArray src, OutputArray dst, int dep, InputArray kernelY, InputArray kernelX, Point anchor=Point(-1,-1), int borderType=BORDER_DEFAULT){
+    Mat Y;
+    conv2D(src, kernelY, Y, dep, anchor, borderType);
+    conv2D(Y, kernelX, dst, dep, anchor, borderType);
+}
+//先进行水平方向的卷积，再进行垂直方向的卷积
+void sepConv2D_X_Y(InputArray src, OutputArray dst, int dep, InputArray kernelX, InputArray kernelY, Point anchor=Point(-1,-1), int borderType=BORDER_DEFAULT){
+    Mat X;
+    conv2D(src, kernelX, X, dep, anchor, borderType);
+    conv2D(X, kernelY, dst, dep, anchor, borderType);
+}
+//Sobel算子提取边缘信息
+Mat Sobel(Mat &src, int x_flag, int y_flag, int kSize, int borderType){
+    Mat Smooth = getSmmoothKernel(kSize);
+    Mat Diff = getDiffKernel(kSize);
+    Mat dst;
+    if(x_flag){
+        sepConv2D_Y_X(src, dst, CV_32FC1, Smooth.t(), Diff, Point(-1, -1), borderType);
+    }else if(x_flag == 0 && y_flag){
+        sepConv2D_X_Y(src, dst, CV_32FC1, Smooth, Diff.t(), Point(-1, -1), borderType);
+    }
+    return dst;
+}
+int main(){
+    Mat src = imread("../lena.jpg");
+    Mat gray;
+    cvtColor(src, gray, CV_BGR2GRAY);
+    Mat dst1 = Sobel(gray, 1, 0, 3, BORDER_DEFAULT);
+    Mat dst2 = Sobel(gray, 0, 1, 3, BORDER_DEFAULT);
+    //转8位灰度图显示
+    convertScaleAbs(dst1, dst1);
+    convertScaleAbs(dst2, dst2);
+    imshow("origin", gray);
+    imshow("result-X", dst1);
+    imshow("result-Y", dst2);
+    imwrite("../result.jpg", dst1);
+    imwrite("../result2.jpg", dst2);
+    waitKey(0);
+    return 0;
+}