Abstract: Most of traditional visual indoor navigation algorithms
and methods only consider the localization in ordinary daytime, while
we focus on the indoor re-localization in low light in the paper. As
RGB images are degraded in low light, less discriminative infrared
and depth image pairs are taken, as the input, by RGB-D cameras, the
most similar candidates, as the output, are searched from databases
which is built in the bag-of-word framework. Epipolar constraints can
be used to relocalize the query infrared and depth image sequence.
We evaluate our method in two datasets captured by Kinect2. The
results demonstrate very promising re-localization results for indoor
navigation system in low light environments.
Abstract: In this paper, some relative efficiency have been
discussed, including the LSE estimate with respect to BLUE in curve
model. Four new kinds of relative efficiency have defined, and their
upper bounds have been discussed.
Abstract: A new relative efficiency in linear model in reference is
instructed into the linear weighted regression, and its upper and lower
bound are proposed. In the linear weighted regression model, for the
best linear unbiased estimation of mean matrix respect to the
least-squares estimation, two new relative efficiencies are given, and
their upper and lower bounds are also studied.