how to convert cv::Mat from C++ opencv to Mat for C# opencvsharp?

I want to Convert Mat object from  C++  opencv  to C# opencvsharp.but I don’t how to do .

C++/opencv-------Mat------------>DLL------->C#/opencvsharp---------->Mat.

At first,I Wrote one DLL for C++ as following:

//C++ DLL -----unmanaged

TSO.h

extern "C"
{
	__declspec(dllexport)  Mat WINAPI GetImage(char* filename);//The Mat object is C++ object from opencv
}

TSO.cpp

__declspec(dllexport)  Mat WINAPI GetImage(char* filename)
{
	return imread(filename);
}

then I quote this dll in C# ,As the same time I quote opencvsharp

[DllImport("TSO.dll", EntryPoint = "GetImage")]
 public extern static Mat GetImage(string filename);//Mat is from opencvsharp

finally ,I did like this

Mat img=GetImage(@"D:\1.jpg");//C#object form opencvsharp
            pictureBoxIpl1.ImageIpl = img;//pictureBoxIpl1  is control of opencvsharp

but it doesn’t work,So how to convert?

[shimat]

🤔 If your Mat’s Type is 8UC1 or 8UC3, encoding it into an image such as BMP or PNG is one way to do it. If it’s any other type, then copying the pixel values might be a good way.

__declspec(dllexport) std::vector<uchar>* __cdecl vector_uchar_new()
{
    return std::vector<uchar>;
}
__declspec(dllexport) void __cdecl vector_uchar_delete(std::vector<uchar>* vec)
{
    delete vec;
}
__declspec(dllexport) void __cdecl vector_uchar_getPointerAndLength(std::vector<uchar>* vec, uchar** ptr, int* length)
{
    *ptr = &vec[0];
    *length = static_cast<int>(vec->size());
}

__declspec(dllexport) void __cdecl SomeFunction(char* filename, std::vector<uchar> *outVec)
{
    cv::Mat m = cv::imread(filename);
    cv::imencode(m, ".png", *outVec);
}

[DllImport("MyNative.dll", CallingConvention = CallingConvention.Cdecl)]
public extern static IntPtr vector_uchar_new();
[DllImport("MyNative.dll", CallingConvention = CallingConvention.Cdecl)]
public extern static void vector_uchar_delete(IntPtr vec);
[DllImport("MyNative.dll", CallingConvention = CallingConvention.Cdecl)]
public extern static void vector_uchar_getPointerAndLength(IntPtr vec, out IntPtr ptr, out int length);

[DllImport("MyNative.dll", CallingConvention = CallingConvention.Cdecl)]
public extern static void SomeFunction(string fileName, IntPtr outVec);

IntPtr vec = vector_uchar_new();
try
{
    SomeFunction("image.jpg", vec);
    vector_uchar_getPointerAndLength(vec, out IntPtr ptr, out int length);

    byte[] pngImageBytes = new byte[length];
    Marshal.Copy(ptr, pngImageBytes , 0, length);

    using var mat = OpenCvSharp.Mat.FromImageData(pngImageBytes);
}
finally
{
    vector_uchar_delete(vec);
}

However, if you are using C++ in the first place, I would suggest that you don’t use OpenCvSharp and write all image processing consistently in C++.

[tlhcelik]

it doesn’t work 👎