Image processing with respect to object recognition in robotics? MATLAB via OpenCV?

The Famous “OpenCV via MATLAB Battle” for Image Processing in robotics

• OpenCV is free and open-source software, MATLAB is licensed software that is pretty costly.
• It offers greater functionality for computer vision compared to Matlab.
• It takes very little memory compared to MATLAB.
• It is portable and it is easy to integrate.
• The C++ OpenCV code runs faster than the Matlab code.
• You can track your robot with OpenCV.

• OpenCV has a lot of bugs for image processing. By the way, the OpenCV modified version has fewer bugs.
• It takes time to get comfortable with.

• MATLAB code is easier to understand for beginners (no import statement and no need to define windows) and it offers a better development environment. For instance, it has several embedded plotting features that help to visualize the algorithm steps or results.
• It is licensed software, so you may count on fewer bugs and better functioning. 
• MATLAB includes many algorithms not found in OpenCV (either implemented in M-code or a lower-level language).
• It’s great for experimenting: it’s easy to use and it’s flexible, plus it is great for rapid prototyping and rather easy to debug.
• Object recognition with MATLAB code has perfect efficiency.

• MATLAB is licensed and it is pretty costly. Yes, that is a double-edged sword: you need to pay to use it and the price is solid.
• Integration with hardware is not so flexible when compared to OpenCV.