Abstract:The motion target detection algorithm is easily affected by noise and changes in brightness, leading to the presence of false targets, making it challenging to meet the requirements of accuracy and real-time performance in practical applications. To address these issues, this paper proposes a real-time multi-object motion detection algorithm that incorporates edge information and constructs a video image processing system on the FPGA platform. The algorithm first utilizes a Gaussian filtering enhancement algorithm to improve noise resistance. Subsequently, the three-frame difference method is employed to reduce the probability of false targets. Simultaneously, edge detection is applied to the images, and the edge information is fused with the results of the frame difference method. Finally, a multi-object recognition algorithm is used to identify and label the moving targets. The algorithm fully utilizes the parallel computation and pipeline operations of FPGA. Experimental tests demonstrate that the detection and recognition system can achieve real-time detection of up to 16 moving targets in 1280×720@60Hz video scenes.
