与 OpenCV 3.0 链接时出现“未解析的外部符号"错误 [英] 'unresolved external symbol' error when linking with OpenCV 3.0

问题描述

我从源代码构建 OpenCV 3.0.以及一些额外模块的 contrib repo.我手动设置了我的 VS2013 项目以使用生成的 lib 文件.

我的代码现在很简单:

#include "opencv2corecore.hpp"#include "opencv2imgcodecsimgcodecs.hpp"int _tmain(int argc, _TCHAR* argv[]){cv::Mat image = cv::imread("img.jpg");返回0；}

但是在 VS 2013 社区版中它给了我这些错误:

我看到下载

提取

让我们将包含以下内容的目录称为 OPENCV_DIR:

• 构建
• 来源

创建一个空项目:

• 新建项目 -> Visual C++ -> 空项目

添加一个 cpp 文件(比如 Start.cpp)，其中将包含您的主要功能(例如上面的代码片段)

配置调试

添加include和lib目录:

• 配置属性 -> VC++ 目录
• 包含目录:添加 OPENCV_DIRuildinclude
• 库目录:添加 OPENCV_DIRuildx86vc12staticlib

添加所需的库(以下对这个简单的例子很有用，如果你需要其他功能，你应该添加更多):

• opencv_core300d.lib
• opencv_highgui300d.lib
• opencv_imgproc300d.lib
• opencv_hal300d.lib
• opencv_imgcodecs300d.lib
• libjpegd.lib;
• libpngd.lib
• libjasperd.lib
• IlmImfd.lib
• libtiffd.lib
• libwebpd.lib
• zlibd.lib
• ippicvmt.lib
• %(AdditionalDependencies)

配置属性 -> C/C++ -> 代码生成 -> 运行时库

• 设置为多线程调试 (/MTd)

对于 RELEASE 版本，您需要在发布模式下执行从 6 到 9 的步骤，在步骤 8 中添加不带尾随d"的库，并在步骤 9 中添加多线程 (/MT).

享受吧！

作为奖励，我还建议为 Visual Studio 安装 Image Watch 扩展.对调试您的 Mat 非常有用！

I build the OpenCV 3.0 from source. And with the contrib repo for some extra modules. And I manually setup my VS2013 project to use the generated lib files.

My code is simple for now:

#include "opencv2corecore.hpp"
#include "opencv2imgcodecsimgcodecs.hpp"

int _tmain(int argc, _TCHAR* argv[])
{
    cv::Mat image = cv::imread("img.jpg");
    return 0;
}

But it gave me these errors when in VS 2013 community version:

I see similar thread, they said it is caused by the x86/x64 issue. But My project is already x86. And the OpenCV I built is also targeting x86 (see below). What reason could it be?

ADD 1

The OpenCV 3.0 INSTALL project generates the following lib directory:

So it is indeed a x86 one.

My lib path is configured to the above path:

And I added all the *d.lib files.

And below is my VC project's Configuration Manager

So it is indeed x86, too.

Where could be wrong?

ADD 2

I manually searched for the fastFree() function. It is defined in the opencv_core300d.lib file. I use the dumpbin /symbols command to check the symbols. And I find its name is mangled exactly as fastFree@cv@@YAXPAX@Z. So why it cannot be found?

解决方案

Here the steps to use OpenCV 3.0.0 with precompiled libs, for a C++ project that links OpenCV statically, in Windows (tested with Windows 8.1) and Visual Studio (tested with Visual Studio 2013) to run this program:

#include <opencv2opencv.hpp>
using namespace cv;

int main()
{
    Mat3b img = imread("path_to_image");
    imshow("img", img);
    waitKey();
    return 0;
}

1. Download from http://opencv.org/downloads.html
2. Extract
3. Let's call OPENCV_DIR the dir containing:
   • build
   • source
4. Create an empty project:
   • New Project -> Visual C++ -> Empty Project
5. Add a cpp file (say Start.cpp) that will contain your main function (e.g. the snippet above)
6. Configuration DEBUG
7. Add include and lib directories:
   • Configuration Properties -> VC++ Directories
   • Include Directories: add OPENCV_DIRuildinclude
   • Library Directories: add OPENCV_DIRuildx86vc12staticlib

8. Add required libs (the following are good for this simple example, you should add more if you need other functionalities):

   • opencv_core300d.lib
   • opencv_highgui300d.lib
   • opencv_imgproc300d.lib
   • opencv_hal300d.lib
   • opencv_imgcodecs300d.lib
   • libjpegd.lib;
   • libpngd.lib
   • libjasperd.lib
   • IlmImfd.lib
   • libtiffd.lib
   • libwebpd.lib
   • zlibd.lib
   • ippicvmt.lib
   • %(AdditionalDependencies)

9. Configuration Properties -> C/C++ -> Code Generation -> Runtime Library

   • Set to Multi-threaded Debug (/MTd)

10. For a RELEASE build, you need to do steps from 6 to 9 in release mode, adding libs without the trailing "d" in step 8, and Multi threaded (/MT) in step 9.

11. Enjoy!

As a bonus, I also recommend to install Image Watch extension for Visual Studio. Very very useful for debugging your Mats!

这篇关于与 OpenCV 3.0 链接时出现“未解析的外部符号"错误的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！