Detection of point targets becomes increasingly more difficult as targets become weak and engagement takes
place in highly dense, varying and complex background like clouds. To detect weak point targets in this
scenario, detection threshold should be sufficiently low. And this leads to high false alarm rate. In order to
make detection system robust to dense clutter (we mean 'clouds') and noise, post processing algorithms are
required. Almost all detection/tracking systems use post processing techniques, but less has been reported in
this area. In this paper, we propose a simple and computationally efficient post processing algorithm to
encounter false alarms due to dense and varying clouds. Models for target and cloud edges are presented.
Results demonstrate that proposed algorithm is able to reduce false alarms to a large extent.
KEYWORDS: Target detection, Signal to noise ratio, Infrared imaging, Detection and tracking algorithms, Image segmentation, Digital filtering, Computer simulations, Image filtering, Electronic filtering, Algorithm development
An integrated algorithm for the detection of dim point/extended size, slow/fast-moving targets has been presented in this
paper. In the proposed algorithm, essentially an innovation over an existing algorithm reported by Nengli Dong et al ,
morphological operations are carried out on the incoming IR data to improve signal to noise ratio (SNR). Methods of
entropy thresholding and conjunction functions are integrated together. Conjunction function based algorithm has been
significantly modified to take care of fast moving targets, a limitation of the method proposed by Nengli Dong et al. Our
proposed algorithm is able to detect point as well extended size targets with low contrast and having frame to frame
movements varying from sub-pixel to tens of pixels.