``Transparent Migration of Off-Line Frame Rate Vision Systems to Real-Time''
A. Tyrrell, J. LaPre, C. D. Carothers, B. Roysam and C. V. Stewart
IEEE Transactions on Information Technology in Biomedicine, Volume 8, Number 2, June 2004.
ABSTRACT
This paper presents a collection of techniques and lessons learned in
the context of efficient migration of a large and complex computer
vision code base developed off-line into an equivalent real-time
implementation under a standard open-source operation system (Linux).
Using creative linking strategies, it is possible to create a
robust environment based on loadable kernel modeuls that enables
simultaneous realization of real-time and off-line frame rate computer
vision systems from a single code based.
Using our approach we demonstrate hard real-time predictability of a complex
frame-rate vision system using commericial-off-the-self (COTS) hardware
and a standard uni-processor Linux OS. This approach is simple; one can
maximize systemic predictability by simply placing time-critical components
of a user-level executable directly into the kernel as a virtual device
driver. This effecitively emulates a single process space model that is
non-preemptable, non-pageable and that has direct access to a powerful
set of system level services.
Experiments on a frame-rate vision system designed for computer-assisted
laser retinal surgery shows that this kernel method reduces the variance
of observed per frame CPU cycle counts by two orders of magnitude. The
conclusion is that when predictable off-line algorithms are used, it is
possible to efficiently migrate to a predictable frame-rate computer
vision system.