Point spread character based low SNR single pixel infrared target detection

• Published in: Journal of engineering (Stevenage, England) Vol. 2019; no. 20; pp. 7042 - 7046
• Main Authors: Nan, Jinghong, Zhao, Baojun, Tang, Linbo, Zhang, Zengshuo, Tang, Wei
• Format: Journal Article
• Language: English
• Published: The Institution of Engineering and Technology 01.10.2019; Wiley
• Subjects: adaptive segmentation method; background suppression; detection accuracy; feature extraction; high boost filter; IET International Radar Conference (IRC 2018); image denoising; image enhancement; image filtering; image segmentation; infrared imaging; IR searching and tracking system; low signal-to-noise ratio single pixel ir target detection; low SNR single pixel infrared target detection; median filter; median filters; noise disturbance; noise suppression; object detection; point-spread character; point-spread indicator; point-spread local contrast method; potential target protection; target enhancement; target extraction; two-step enhancement model; point-spread indicator; noise suppression; point-spread local contrast method; image filtering; object detection; IR searching and tracking system; image enhancement; image segmentation; feature extraction; detection accuracy; potential target protection; target extraction; low SNR single pixel infrared target detection; noise disturbance; adaptive segmentation method; median filter; median filters; high boost filter; image denoising; background suppression; target enhancement; point-spread character; infrared imaging; two-step enhancement model; low signal-to-noise ratio single pixel IR target detection
• ISSN: 2051-3305; 2051-3305
• DOI: 10.1049/joe.2019.0323

Robust and efficient low signal-to-noise ratio (SNR) single pixel infrared (IR) target detection has been a key technique in IR searching and tracking system. However, there exist two critical difficulties, i.e. noise disturbance and low contrast between target and background, influencing detection accuracy. A novel IR target detection method is proposed to overcome those difficulties, by utilising the point-spread character. Specifically, a point-spread indicator is advocated to protect the potential target with the median filter to suppress the noise. Afterwards, a two-step enhancement model is built to increase the contrast value between target region and surrounding background effectively. In the first step, a point-spread local contrast method is proposed to enhance targets and suppress background. In the second step, a high boost filter is introduced to further enhance the target. In addition, an adaptive segmentation method is exploited to extract targets from the enhanced image. Experimental results on testing sequences demonstrate that this method outperforms the other state-of-the-art methods in single pixel IR target detection with higher detection rate, lower false alarm rate and less time consuming.