Inter-frame Correlation Based on Moving Vehicle Target Detection in Infrared Image Sequences

Abstract

Moving vehicle targets detection in infrared image sequences is playing a more and more important role in infrared search and track systems. This paper presents a novel method based on inter-frame correlation to detect moving vehicle target in infrared image sequences reliably. Firstly, for a single frame, the image respectively is sharpened and enhanced after image denoising, and then generating the preprocessed image. Secondly, the vehicle targets in infrared image sequences are detected by a saliency based target detection algorithm. For consecutive frames, features of motion between real vehicle targets and false ones are different, then the inter-frame correlation is operated to suppress the false alarm, making the vehicle targets detection more accurate. The experiments on the image sequences demonstrate that the proposed method has ensure detection effectiveness and robustness for vehicle detection in infrared image sequences under complex backgrounds, and it also improve the reliability.