Infrared Target Detection Based On Background Prediction

Infrared thermography technology has found a wide application due to its own advantages. The infrared target detection technology is a key technique in infrared technology application, and it has become an important research subject. At the same time,infrared target detection is also a challenging task, because of strong noise and interference, and lack of effective target shape and texture features.Given the characteristics of the infrared image, infrared target detection based on background prediction has been researched in this thesis. The main content is focused on background prediction based on TDLMS filter, and it has been discussed whether TDLMS filter algorithm could be applied in the infrared target detection. Computer simulations are conducted on infrared test images with various backgrounds, and the TDLMS filter is employed to predict the background in infrared image processing. The experimental result shows that TDLMS filter algorithm have a good performance in the infrared target detection, and the performance indicators illustrate that the TDLMS filter algorithm could effectively suppress the clutter background and enhance the infrared target.In this thesis, some improvements in the structure of the filter have been made,because of some shortcomings of TDLMS filter algorithm in infrared target detection. The input window of TDLMS filter has been modified to make full use of neighborhood information of the predicted pixel, and the background is predicted using nonlinear step size to improve prediction accuracy. At the same time, an output module has been add to the basic TDLMS filter, and two kinds of schemes has been presented in the output module, corresponding to two modified algorithms. The simulation experimental results show that the two modified algorithms have a good performance in infrared target detection. The comparison reveal that the two modified algorithms have better target detection ability and background suppress ability compared with the basic TDLMS filter algorithm, and the two modified TDLMS filter algorithms have their respective advantages in the infrared target detection.