Abstract: Micro-expression is a kind of short-duration subtle expression which is not controlled by
the autonomic nervous system. Micro-expression appears when a person is attempting to conceal his
true emotion. Micro-expression detection boasts great application prospects in many fields, such as
public security, investigation and interrogation as well as clinical medicine due to its close
relationship with lie detection. Automatic detection of micro-expressions has come to the fore in
research, because it is of great difficulty to artificially identify micro-expression . This paper proposes
an improved algorithm based on the Horn-Schunck (HS) optical flow for automatic micro-expression
detection. In this study, the pre-conditioned Gauss-Seidel iterative method is employed to improve the
HS optical flow method, which accelerates the convergence rate. Experiments in the spontaneous
micro-expression database CASME show that the propounded method exerts an excellent effect on
the detection of micro-expression.

Key words: micro-expression detection, optical flow, preconditioned iteration