I would like to know how the functions within the Matlab computer vision toolbox work? (like functions which operate Feature Detection, Extraction, and Matching)
Are they written originally in Matlab language, or calling opencv libraries ?
The functionality in Computer Vision System Toolbox is available as MATLAB functions, System Objects, and Simulink blocks.
It's implemented in a mixture of MATLAB code and C code, and as far as I know does not rely on OpenCV at all.
However, it also supports Code Generation using MATLAB Coder and Simulink Coder - in other words, once you have prototyped and developed a system using MATLAB, Computer Vision System Toolbox and maybe Simulink, you can automatically generate much faster C code from it (which, if you want, could be integrated with functionality from OpenCV).