A complete computer vision system contains components ranging from low-level image processing functions to high-level AI reasoning. Implementing such a heterogeneous system requires a general-purpose language, which allows the programmer to switch style (e.g. between functional and imperative) as appropriate. Implementations of the language must be reliable and readily available.
Although we have chosen to extend Common Lisp, similar primitives could be added to any one of several modern high-level languages. These include:
Other high-level languages might also be worth examining.
To achieve suitable efficiency, many computer vision researchers resort to using low-level languages. C is traditional. Some recent projects have used C++. The recent language Java is similar: it looks like it might be safer and more portable, but possibly less efficient. These languages lack many useful abstract features found in high-level languages.
Ownership, Maintenance and Disclaimers
Last modified