An object-oriented framework for the Cloner software prototyping environment

Cloner is a library-based prototyping environment for computer vision and image processing algorithms and tasks. It is being designed to allow users to take advantage of the computing power provided by parallel processing systems without requiring an extensive understanding of the underlying architecture. At the heart of Cloner is a library organized as families of algorithms, where an algorithm family is defined in terms of functionality. The individual algorithms that comprise a family correspond to different algorithms that reflect the effects of architecture and data set parameters. Parallel implementation decisions based on user or system-provided information are embedded in the library. Cloner helps in selecting the best algorithm and generates optimized code for the specified target parallel architecture, based on factors that include problem size, machine size, communications topology, and data characteristics. In this paper we present an object-oriented framework based on HPC++ for the Cloner library. We discuss the use of inheritance, polymorphism, function overloading and the C++ template facility to facilitate the library building and algorithm selection process.