Research On Small Target Detection In Passive Millimeter-wave Images

Passive Millimeter-wave imaging is a kind of original detection way. It can be applied to detecting aerial targets, and have many valuable and desired features, including concealment and survival ability, illumination and weather independence, battlefield applicability, and strong ability of detecting invisible targets. But the technology of small target detection based on Passive Millimeter-wave images is rarely reported. Theoretically, Passive Millimeter-wave imaging is similar with Infrared imaging, all by receiving radiation signals of targets for imaging. Therefore, the theory and methods of small target detection in Infrared image can be used for reference, and then designed and ameliorated on the characteristics of Passive Millimeter-wave image. This paper will focus on the background suppression of Passive Millimeter-wave images, and small target detection methods based on single frame image.First of all, we introduce the theory of Passive Millimeter-wave imaging and small targets detection. Calculation methods of the detection probabilities and the false-alarm probabilities of two detection ways are discussed, including detection on single frame and image sequences. And then the characteristics of small target, background and noise in Passive Millimeter-wave image are analyzed. Suppression of image background is a crucial step in small target detection. For the limitation of traditional background suppression arithmetic, this paper designs an improved TDLMS filter based on consecutive frames. The simulation result demonstrates the new method can wipe off background effectively, and enhance the SNR of images. In single frame image, multi-orientation gradient and singularity are the biggest differences between small targets and background. So the improved arithmetic of multi-orientation gradient and wavelet transform are used to detect small targets in Passive Millimeter-wave image. To the real images given by the Passive Millimeter-wave radiation system, the result demonstrates that the two methods can both detect small target effectively.