The input and output arrays

If you look at the OpenCV documentation, you will see that many methods and functions accept parameters of the cv::InputArray type as the input. This type is a simple proxy class introduced to generalize the concept of arrays in OpenCV, and thus, avoid the duplication of several versions of the same method or function with different input parameter types. It basically means that you can supply a cv::Mat object or other compatible types as an argument. This class is just an interface, so you should never declare it explicitly in your code. It is interesting to know that cv::InputArray can also be constructed from the popular std::vector class. This means that such objects can be used as the input to OpenCV methods and functions (as long as it makes sense to do so). Other compatible types are cv::Scalar and cv::Vec; this later structure will be presented in Chapter 2, Manipulating the Pixels. There is also a cv::OutputArray proxy class that is used to designate the arrays returned by some methods or functions.