Skip to Main Content
A novel complex background suppression based dim point target detection algorithm is presented in this study. Comparing with existing spatial-filter-based background suppression algorithms, the nucleus similarity degree (NSD) of each pixel is analyzed first, then the complex background is predicted by fusing results of two different predictors, and the complex background is removed by subtracting the fused prediction result. A thresholding process is performed on the remnant image, and the target trajectories are detected in binary image sequences to confirm the potential targets. Experimental results indicated that the method can predict the heavily cluttered background accurately as well as enhance the point target, and detect target trajectories effectively. Moreover, the pixel-level processing which is time-consuming in the algorithm is suitable for hardware implementation, this makes a feasible way to achieve real time processing in Automatic Target Recognition (ATR) weapon system.