Declarative Video Processing for Decomposable Algorithm Simulation

Video signal processing algorithms which are to be implemented in hardware typically have severe constraints on their memory usage and window of data access. When implemented in software as a chain of processes, there are nontrivial data dependencies between output samples and input samples. We demonstrate that simulating such a chain of processes using a lazily evaluated, higher order functional language can lead to significant benefits during development. We show why modularity is increased, the gap between the implementation and specification reduced, and memory usage minimised. Keywords: Video signal processing, Software engineering, Software prototyping, Functional programming