Due to the desire for real-time information processing of optical imaging system
such as infrared search/track (IRST) systems
we proposed a space-temporal dim target detection method in ego-motion imaging systems.A spatial two-dimensional generalized likelihood ratio test algorithm was employed to extract underlying objects.Then the measurable discriminable feature was combined to a feature vector which was associated with the object.The correct target in image sequence was detected by multistage hypothesis testing(MHT) method using its associated feature vector.The algorithm made a good trade-off between its statistic performance and computing complexity
and had acquired great performance in practical applications.