Method of normal-direction flow information-based target detection under dynamic environment

The invention provides a method of normal-direction flow information-based target detection under a dynamic environment. The method mainly includes the three major steps of: step 1, estimating FOE bya voting method; step 2, generating an initial labeling field; and step 3, generating an optimal labeling field, and thus realizing detection of a target. The greatest advantages of the method are that an algorithm is lower in sensitivity to noises, has no requirement for image texture and image features of shooting scenes, is thus wider in applicable scenes, and has broad prospects.