如何用OpenCV在任意电脑显示图片并嵌入程序，解决路径依赖问题

Got it, let's fix this path dependency issue so your image loads reliably no matter where the exe is placed. Here are two solid methods tailored to your Windows + OpenCV setup:

Method 1: Convert Image to C++ Header File (Cross-Platform Friendly) #

This approach turns your image into a byte array stored in a header file, which gets compiled directly into your executable. It works on any platform that supports C++ and OpenCV.

Step 1: Generate the Image Header File #

First, create a small helper program to convert d.png into a C++ array:

#include <opencv2/opencv.hpp>
#include <fstream>
using namespace cv;

int main() {
    // Read the image with all channels (including transparency if present)
    Mat img = imread("d.png", IMREAD_UNCHANGED);
    if (img.empty()) {
        cerr << "Failed to read d.png!" << endl;
        return 1;
    }

    std::ofstream out("image_data.h");
    out << "#pragma once\n\n";
    out << "unsigned char img_data[] = {\n";
    
    // Write each byte of the image data
    for (int i = 0; i < img.total() * img.elemSize(); ++i) {
        out << static_cast<int>(img.data[i]) << ",";
        // Add line breaks for readability
        if (i % 16 == 15) out << "\n";
    }
    out << "};\n\n";
    out << "const int img_rows = " << img.rows << ";\n";
    out << "const int img_cols = " << img.cols << ";\n";
    out << "const int img_type = " << img.type() << ";\n";
    
    return 0;
}

Compile and run this program—you'll get image_data.h in the same directory.

Step 2: Modify Your Main Program #

Update your original code to load the image from the embedded array instead of a file:

#include <iostream>
#include <Windows.h>
#include <opencv2/opencv.hpp>
#include "opencv2/highgui.hpp"
#include "image_data.h" // Include the generated header
using namespace std;
using namespace cv;

int main() {
    // Create Mat from the embedded byte array
    Mat img(img_rows, img_cols, img_type, img_data);
    if (img.empty()) {
        cerr << "Failed to load embedded image!" << endl;
        return 1;
    }

    namedWindow("Image");
    imshow("Image", img);
    waitKey(0);
    
    cout << "h";
    int i;
    cin >> i;
    return 0;
}

Method 2: Use Windows Resource Files (Windows-Native) #

If you're only targeting Windows, using the native resource system is clean and avoids generating extra header files.

Step 1: Create a Resource File #

Create a new file named resource.rc with this content:

// Define a resource ID and link it to your image
MY_IMAGE PNG "d.png"

Add this .rc file to your Visual Studio project (right-click project > Add > Existing Item).

Step 2: Load the Resource in Your Program #

Update your main code to fetch the embedded resource from the exe and decode it into an OpenCV Mat:

#include <iostream>
#include <Windows.h>
#include <opencv2/opencv.hpp>
#include "opencv2/highgui.hpp"
using namespace std;
using namespace cv;

// Define the resource ID (matches what's in resource.rc)
#define MY_IMAGE 101

int main() {
    // Locate and load the embedded PNG resource
    HRSRC hRes = FindResource(NULL, MAKEINTRESOURCE(MY_IMAGE), L"PNG");
    if (!hRes) {
        cerr << "Could not find embedded image resource!" << endl;
        return 1;
    }

    HGLOBAL hData = LoadResource(NULL, hRes);
    if (!hData) {
        cerr << "Failed to load image resource!" << endl;
        return 1;
    }

    DWORD imgSize = SizeofResource(NULL, hRes);
    unsigned char* imgBytes = static_cast<unsigned char*>(LockResource(hData));

    // Decode the raw bytes into an OpenCV Mat
    Mat img = imdecode(Mat(1, imgSize, CV_8UC1, imgBytes), IMREAD_UNCHANGED);
    if (img.empty()) {
        cerr << "Failed to decode embedded image!" << endl;
        return 1;
    }

    // Clean up resource handles
    UnlockResource(hData);
    FreeResource(hData);

    namedWindow("Image");
    imshow("Image", img);
    waitKey(0);
    
    cout << "h";
    int i;
    cin >> i;
    return 0;
}

Quick Notes #

• If your image has transparency (like PNGs), use IMREAD_UNCHANGED to preserve the alpha channel—otherwise OpenCV will strip it out.
• For Method 1, if you update the image, you'll need to re-run the helper program to regenerate image_data.h.
• For Method 2, make sure the resource ID in your code matches the one in the .rc file (I used 101 as an example).

内容的提问来源于stack exchange，提问作者Programmer