Microscopic Motion Tracking Systems utilize sub-pixel algorithms to achieve close to nanometer precision motion measurement. This paper characterizes the bias-error of template matching image registration techniques which are commonly used to achieve this level of sub-pixel accuracy. An accurate functional expression for such bias error is derived, which shows how the error depends on image content. Furthermore, this expression is directly utilized in a bias reduction scheme that is shown to significantly reduce the bias error. The newly derived expressions and reduction schemes for bias are applied to a Microscopic Vision System that measures micro-object motion with close to nanometer resolution.