Quantitative Measures for Systematic Optimization, Validation, and Imperfection Compensation in the Holistic Modeling and Parsimonious Design of Application-Specific Vision and Cognition Systems

Intelligent systems, e.g., for vision and cognition tasks, enjoy increasing industrial acceptance and application. Application domains range from optical character and handwriting recognition to biometric systems. The joint exploitation of advanced microelectronics, sensor technology, and intelligent systems provides a tremendous economic potential. Tight application constraints such as, e.g., size, speed, performance, and power consumption give increasing attraction to dedicated integrated system implementations exploiting bio-inspiration and opportunistic design techniques in analog or mixed-signal circuits and systems. However, to achieve a viable design at reasonable cost and time-to-market, an appropriate design methodology is required. This paper presents quantitative measures for system-oriented imperfection compensation, optimization, and validation of dedicated application-specific intelligent systems. These measures serve for the fast and consistent behavioral modeling of an aspired intelligent system, and support the rapid and consistent transformation into a physical design. Further, they provide the basis for ensuing design automation of the process, and contribute to the ongoing vivid activities of method and tool development for system simulation and evaluation.