Infrared small target detection method and system based on FPGA

Abstract

The invention provides an infrared small target detection method and system based on an FPGA. Acquired images are preprocessed to be denoised and processed through blind pixel removal; a head image and a tail image are extracted; clutter edge filtering operation is carried out on the head image and the tail image; threshold segmentation operation is carried out on the head image and the tail image so that the images can be binary ; morphological expansion is carried out on the target in the head image processed through threshold segmentation and the target in the tail image processed through threshold segmentation, and targets with large contours are obtained; the head target processed through morphological expansion and the tail target processed through morphological expansion are synthesized so that a complete target can be obtained. According to the infrared small target detection method and system based on the FPGA, the signal-to-noise ratio can be increased by accumulating the superposition portions of a previous frame and a next frame, the interference of a background cloud layer can be well removed while the algorithm complexity is reduced, and the real time performance of the detection system can be improved while accuracy is guaranteed.

Description

technical field [0001] The invention belongs to the field of infrared early warning systems, and in particular relates to an FPGA-based infrared small target detection method and system. Background technique [0002] In the infrared early warning system, the commonly used small target detection method is the point target detection technology. Point target detection technology is aimed at situations where the position of the thermal imaging camera is fixed. Now, a new detection technology has been proposed at home and abroad, that is, to place the infrared thermal imager on a rotating platform, and use the tailing phenomenon generated by the small target during the rotation to convert the small target detection into line target detection, which greatly reduces the detection cost. difficulty. [0003] Traditional detection algorithms include frequency-domain high-pass filtering algorithm and frequency-domain low-pass filtering algorithm. These two algorithms are better for g...

AI Technical Summary

Problems solved by technology

[0003] Traditional detection algorithms include frequency-domain high-pass filtering algorithm and frequency-domain low-pass filtering algorithm. These two algorithms are better for gentle sky detection, but they are not good for clutter backgrounds with drastic gray changes (such as cloud clutter). It is difficult to guarantee the filtering effect; other novel methods such as two-dimensional least mean square (TDLMS) filtering, adaptive autoregressive background estimation method, and random field model method can handle the cloud background better, but the algorithm complexity is higher

[0004] The single above-mentioned filtering algorithm or a combination of some algorithms adopted by the traditional infrared early warning system device is only suitable for some specific occasions with complex backgrounds; in addition, the traditional infrared early warning system uses more linear arrays on the basis of rotating platforms. Scanning detection method, there will be blind spots between frames during the scanning process, which reduces the signal-to-noise ratio of the image

Images