《Mastering Opencv读书笔记》第一章 实现图像卡通效果

这本书和配套代码网上都有得下载。

要实现书中的效果，只要三步：1.使用拉普拉斯算子提取轮廓  2.使用双边滤波器对图像进行平滑 3.根据第一步得到的轮廓模版图，将第二步的结果拷贝过去【填充轮廓图中全白的部分】

由于我的笔记本摄像头坏了，故我的程序读取的是手机拍摄的视频。

下面给出我整理的两段代码：

1.边缘提取：

// GetMySketch.cpp : 定义控制台应用程序的入口点。
//

#include "stdafx.h"
#include<iostream>
#include <opencv2/core/core.hpp>  
#include <opencv2/highgui/highgui.hpp>  
#include <opencv2/imgproc/imgproc.hpp>  
using namespace std;
using namespace cv;

int _tmain(int argc, _TCHAR* argv[])
{
	VideoCapture m_caputre("test2.avi");
	Mat m_frame,gray,edges,masks;
	const int MEDIAN_BLUR_FILTER_SIZE = 7;
	const int LAPLACIAN_FILTER_SIZE = 5;
	const int EDGES_THRESHOLD = 80;

	while (true)
	{
		m_caputre>>m_frame;
		if (m_frame.empty()) {
			std::cerr << "ERROR: Couldn't grab a video frame." <<
				std::endl;
			exit(1);
		}

		cvtColor(m_frame,gray,CV_BGR2GRAY);

		medianBlur(gray,gray,MEDIAN_BLUR_FILTER_SIZE);
		
		Laplacian(gray, edges, CV_8U, LAPLACIAN_FILTER_SIZE);

		threshold(edges, masks, EDGES_THRESHOLD, 255, THRESH_BINARY_INV);

		// Display the processed image onto the screen.
		imshow("keep smile :)", masks);
		char keypress = cv::waitKey(20); // Need this to see anything!
		if (keypress==27)
		{
			break;
		}
	}
	return 0;
}

边缘检测结果：

2.最终卡通效果：

// ColorPainting_Cartoon.cpp : 定义控制台应用程序的入口点。
//

#include "stdafx.h"
#include<iostream>
#include <opencv2/core/core.hpp>  
#include <opencv2/highgui/highgui.hpp>  
#include <opencv2/imgproc/imgproc.hpp>  
using namespace std;
using namespace cv;

int _tmain(int argc, _TCHAR* argv[])
{
	VideoCapture m_capture("test2.avi");
	Mat m_frame,smallImg,tmp,bigImg,gray,edges,masks,dst;
	int repetitions = 7; // Repetitions for strong cartoon effect.
	const int MEDIAN_BLUR_FILTER_SIZE = 7;
	const int LAPLACIAN_FILTER_SIZE = 5;
	const int EDGES_THRESHOLD = 80;

	m_capture>>m_frame;
	Size size = m_frame.size();
	Size smallSize;
	smallSize.width = size.width/2;
	smallSize.height = size.height/2;
	smallImg = Mat(smallSize, CV_8UC3);
	tmp = Mat(smallSize, CV_8UC3);
	dst= Mat(size,CV_8UC3);

	while (true)
	{
		m_capture>>m_frame;
		if (m_frame.empty()) {
			std::cerr << "ERROR: Couldn't grab a video frame." <<
				std::endl;
			exit(1);
		}

		cvtColor(m_frame,gray,CV_BGR2GRAY);

		medianBlur(gray,gray,MEDIAN_BLUR_FILTER_SIZE);

		Laplacian(gray, edges, CV_8U, LAPLACIAN_FILTER_SIZE);

		threshold(edges, masks, EDGES_THRESHOLD, 255, THRESH_BINARY_INV);

		resize(m_frame, smallImg, smallSize, 0,0, INTER_LINEAR);
		
		for (int i=0; i<repetitions; i++) {
			int ksize = 9; // Filter size. Has a large effect on speed.
			double sigmaColor = 9; // Filter color strength.
			double sigmaSpace = 7; // Spatial strength. Affects speed.
			bilateralFilter(smallImg, tmp, ksize, sigmaColor, sigmaSpace);
			bilateralFilter(tmp, smallImg, ksize, sigmaColor, sigmaSpace);
		}

		resize(smallImg, bigImg, size, 0,0, INTER_LINEAR);

		dst.setTo(0);

		//! copies those matrix elements to "m" that are marked with non-zero mask elements.
		bigImg.copyTo(dst,masks);

		// Display the processed image onto the screen.
		imshow("keep smile :)", dst);
		char keypress = cv::waitKey(20); // Need this to see anything!
		if (keypress==27)
		{
			break;
		}
	}
	return 0;
}

最终效果图：

第一章很简单，后面的肤色模型很水，就没去弄。下一篇，为实现如何将上述代码移植到android平台上。