The large sky area multi-object fiber spectroscopic telescope (LAMOST) is an innovative reflecting schmidt
telescope, promising a very high spectrum acquiring rate of several ten-thousands of spectra per night. By using the
parallel controllable fiber positioning technique, LAMOST makes reconfiguration of fibers accurately according to
the positions of objects in minutes and fine adjusting the fibers. As a key problem, High precision positioning
detection of LAMOST fiber positioning unit has always been highly regarded and some detection schemes have
been proposed. Among these, active detection method, which determines the final accurate position of optical fiber
end with the help of lighting the fiber, has been most widely researched, but this kind of method could not be applied
in LAMOST real-time observation because it needs projecting light into fiber. A novel detection idea exploiting the
technique of template matching is presented in this paper. As we know, final position of a specific fiber end can be
easily inferred by its corresponding revolving angles of the central revolving axle and bias revolving axle in double
revolving style, so the key point in this problem is converted to the accurate determination of these revolving angles.
Template matching technique are explored to acquire the matching parameters for its real-time collected imagery,
and thus determine the corresponding revolving angle of the central revolving axle and bias revolving axle
respectively. Experiments results obtained with data acquired from LAMOST site are used to verify the feasibility
and effectiveness of this novel method.