How to generate mex for openCV's 'imdecode' for CPU and GPU?

I want to generate mex funtion for 'imdecode'.

First for CPU and then for GPU(GPU, if possible).

I have already downloaded 'OpenCV Interface Support'. But it lacks in some libraries. E.g. mexOpenCV.m shows..

%     Libraries that are not included:
%     opencv_highgui, opencv_imgcodecs, opencv_videoio, opencv_cudacodec

And I think, openCV's 'imdecode' contains in "opencv_highgui, opencv_imgcodecs".

How can this libraries be added permanently in MATLAB?

and

Is it possible to avail GPU performance benefits for this particular purpose?

Thank you

Answer 1

[Hitesh Kumar Dasika]

Some libraries are not included because of dependency on other third party libraries. The functionality offered by these libraries can also be found in Image processing toolbox as well. You can make use of that.

Image Processing Toolbox

[JAI PRAKASH]

I want to decode jpg data.

Jpg data which are in 'uint8' bits form.

In openCV imdecode is designed for this purpose. But I am searching for a function which can give better performnace than imdecode.

I searched 'MANGO' library is fastest.

Is there any way I can use mango library through MATLAB?

[Hitesh Kumar Dasika]

You can make use of the below mentioned MATLAB function to load a C library into MATLAB and use it.

Load Library

Answer 2

[Joss Knight]

You can just download and use OpenCV yourself, and link in whatever libraries you need on the compile line. This blog post gives an example; it's a bit out of date because it uses MEX instead of MEXCUDA (so now you don't need to copy across the mex options file) but it's a workflow you can follow.

[JAI PRAKASH]

How can I use nvJPEG libraries to decode image from jpeg's buffer?

e.g. nvjpegDecode(); and so on....??

I can do this easily using OpenCV lib, below is my 'imdecode.cpp'

#include "opencvmex.hpp"
#include "opencv2/imgcodecs.hpp"
#define _DO_NOT_EXPORT
#if defined(_DO_NOT_EXPORT)
#define DllExport  
#else
#define DllExport __declspec(dllexport)
#endif
using namespace cv;
using namespace std;

void mexFunction(int nlhs, mxArray *plhs[], int nrhs, const mxArray *prhs[])
  {
      cv::Ptr<cv::Mat> rawData = ocvMxArrayToImage_uint8(prhs[0]);
      Mat frame = imdecode(*rawData, -1);
      plhs[0] = ocvMxArrayFromImage_uint8(frame);
  }

An then compiling and calling

mexOpenCV imdecode.cpp
im=imdecode(bufferData); % bufferData is bits in uint8

[xingxingcui]

Which static library file do I need to specify when I use ocvMxArrayToImage_uint8()? libmat.lib, libmx.lib, libmex.lib, libeng.lib? These 4 library files are not enough, these libraries are located in "C:\Program Files\MATLAB\R2020b\extern\lib\win64\microsoft", there are so many I don’t know which one to choose? thanks!

Similar questions refer to this link.