抄録
This paper proposes a velocity field sensing system using a conventional color TV camera and a trichrome stroboscope that sequentially flashes three different color lights within a frame interval. The key consideration is the distinct separation of three sequential images from a mixed RGB image. The three sequential images can then be processed for velocity field extraction. We assume that any targets or flow markers have a uniform color. A procedure for determining mixture matrix A, which plays a crucial role in the separation, is presented and some requirements on A for inversion are discussed. Experimental results show that this method successfully decomposed an RGB image of a fast rotating object and extracted the velocity field from the separated image sequence. Practical issues-automatic adjustment of the flash interval to the object's speed and applicability of this method-are also discussed.
