I have an expanding image processing project which relies heavily on the OpenCV library for much of its functionality, although I do also use a few boost functions as well.

I'd like to start using smart pointers to replace some raw pointers which are beginning to cause problems. My question is on which type of smart pointer to use, with my main choices (I think) being the OpenCV cv::Ptr or one of the boost variants.

I realise there are a number of questions explaining the different between each of the boost pointers, but I hoped somebody could offer an explanation of how cv::Ptr compares with them and make any recommendations of one or the other?

EDIT - I've noticed from the OpenCV docs that Ptr is similar to boost shared_ptr, is the essential difference just which library/include files are required?

I have an expanding image processing project which relies heavily on the OpenCV library for much of its functionality, although I do also use a few boost functions as well.

I'd like to start using smart pointers to replace some raw pointers which are beginning to cause problems. My question is on which type of smart pointer to use, with my main choices (I think) being the OpenCV cv::Ptr or one of the boost variants.

I realise there are a number of questions explaining the different between each of the boost pointers, but I hoped somebody could offer an explanation of how cv::Ptr compares with them and make any recommendations of one or the other?

EDIT - I've noticed from the OpenCV docs that Ptr is similar to boost shared_ptr, is the essential difference just which library/include files are required?