總結(jié)一下OpenCV遍歷圖像的幾種方法

在圖形處理中，遍歷每個(gè)像素點(diǎn)是最基本的功能，是做算法的基礎(chǔ)，這篇文章來總結(jié)一下OpenCV遍歷圖像的幾種方法。

本文章參考文檔OpenCV tutorials的how_to_scan_images.cpp例子。

最有效率--指針

用c語言直接訪問是最有效率的，最快的，下面是簡單的示例。

int scan_image_c(Mat &I)
{
    int channels = I.channels();
    if (channels != 3)
    {
        printf("test support only three channel.\\n");
        return -1;
    }


    for (int i = 0; i < I.rows; i++)
    {
        Vec3b *ptr = I.ptr

最安全--迭代器

迭代器是C++中的一個(gè)概念，因?yàn)榈鲝挠脩羰种薪庸芰艘恍┕ぷ?，它?huì)保證訪問的安全，所以必然會(huì)導(dǎo)致一些性能上的降低，簡單例子如下。

int scan_image_iterator(Mat &I)
{
    int channels = I.channels();
    if (channels != 3)
    {
        printf("test support only three channel.\\n");
        return -1;
    }


    MatIterator_

最便捷--at方法

OpenCV的Mat類中有一個(gè)at方法，它可以直接返回某個(gè)像素點(diǎn)，示例如下。

int scan_image_random(Mat &I)
{
    int channels = I.channels();
    if (channels != 3)
    {
        printf("test support only three channel.\\n");
        return -1;
    }


    for( int i = 0; i < I.rows; ++i)
    {
        for( int j = 0; j < I.cols; ++j 
        {
            I.at

完整例子

#include 
#include 


using namespace std;
using namespace cv;


int scan_image_c(Mat &I);
int scan_image_iterator(Mat &I);
int scan_image_random(Mat &I);


int main( int argc, char* argv[])
{
    if (argc != 2)
    {
        cout << "input parameters failed!" << endl;
        return -1;
    }


    Mat I;
    I = imread(argv[1], IMREAD_COLOR);


    if (I.empty())
    {
        cout << "The image" << argv[1] << " could not be loaded." << endl;
        return -1;
    }


    const int times = 100;
    double t = 0;


    t = (double)getTickCount();


    for (int i = 0; i < times; ++i)
    {
        cv::Mat clone_i = I.clone();
        scan_image_c(clone_i);
    }


    t = 1000*((double)getTickCount() - t)/getTickFrequency();
    t /= times;


    cout << "Time of scan_image_c        (averaged for "
         << times << " runs): " << t << " ms."<< endl;


    t = (double)getTickCount();


    for (int i = 0; i < times; ++i)
    {
        cv::Mat clone_i = I.clone();
        scan_image_iterator(clone_i);
    }


    t = 1000*((double)getTickCount() - t)/getTickFrequency();
    t /= times;


    cout << "Time of scan_image_iterator (averaged for "
        << times << " runs): " << t << " ms."<< endl;


    t = (double)getTickCount();


    for (int i = 0; i < times; ++i)
    {
        cv::Mat clone_i = I.clone();
        scan_image_random(clone_i);
    }


    t = 1000*((double)getTickCount() - t)/getTickFrequency();
    t /= times;


    cout << "Time of scan_image_random   (averaged for "
        << times << " runs): " << t << " ms."<< endl;


    return 0;
}


int scan_image_c(Mat &I)
{
    int channels = I.channels();
    if (channels != 3)
    {
        printf("test support only three channel.\\n");
        return -1;
    }


    for (int i = 0; i < I.rows; i++)
    {
        Vec3b *ptr = I.ptr

運(yùn)行結(jié)果如下：

Time of scan_image_c        (averaged for 100 runs): 2.04884 ms.
Time of scan_image_iterator (averaged for 100 runs): 4.77701 ms.
Time of scan_image_random   (averaged for 100 runs): 3.64237 ms.

從數(shù)據(jù)上看，c語言的方法確實(shí)是最快的，和其他兩種方式拉開了一定的差距。而at遍歷比迭代器遍歷快了不少。

在平常使用中，我們可以根據(jù)每個(gè)方法的優(yōu)點(diǎn)去選擇不同的方法。

審核編輯：劉清