Text this: Scale-invariant and adaptive-search template matching for monocular visual odometry in low-textured environment