Real-time imaging systems are expected to perform at a rate set by the operating environment. This places additional temporal constraints on the implementation, and has led to advances in parallel processing and optimization. The constraints are also applied, however, to the analysis and design models of the system, an aspect that has received far less consideration, particularly outside of academia. It is well understood that the earlier an error is discovered the cheaper the fix and this is taken to the extent that, if an error can be found during design or even analysis, the repair can be several orders of magnitude cheaper. It is therefore an economic imperative that the analysis and design models are sufficiently expressive such that the flaws in the model can be discovered before they are implemented, and this applies equally to real-time constraints. The contribution of this paper is to make the case for the use of the Unified Modeling Language, extended with a rigorous formal method, the Q-model, for the specification and design of real-time imaging systems. To illustrate its utility the proposed approach is examined in comparison with a more traditional approach. Structured Analysis and Design, by way of a case study.
All Science Journal Classification (ASJC) codes
- Signal Processing
- Computer Vision and Pattern Recognition
- Electrical and Electronic Engineering