A real-time system for detecting moving point target in low SNR image sequences

Concerned with the huge amount of information and complicated combinations of trajectories, the real-time automatic detection of moving point targets in low SNR image sequences belongs to the difficult area of multidimensional image processing. The design and implementation of a real-time system for detecting point targets is presented. The algorithm is under the framework of sequential hypothesis testing based on stage integration, and a new method of stage integration is put forward, which efficiently combines summation projection with maximum value projection. The algorithm has clear hierarchies and only uses some simple operations in the whole procedure. So a real-time approach has been easily implemented on a simple pipeline-based IPM (information processing machine)