Infrared Weak Small Targets Detection Under Complicated Background Technology Research

Infrared dim-small target detection is one of the key technologies in the infrared early warning system, infrared imaging guidance system and wide-field view surveillance system etc. In the complex and low signal-to-noise ratio background environment, the target has only a few pixels. There is no shape and texture information to use. All of the above issues bring great difficulties to infrared dim-small target detection.To solve the above problems, the infrared dim-small target detection in complex background is studied in this thesis. The main research work is focused on target detection based on background suppression and elimination of false targets based on track association. The specific work is as follows:In complex background environment, especially when a large area of background’s brightness is close to the targets’. The existing filter algorithms such as median filtering, high-pass filtering and morphological filtering, usually leak the high frequency noise when inhibiting the background, and increase the false alarm rate. Aiming at this case, this thesis proposed an improved infrared dim-small target detection algorithm based on the mask image. It can better achieve the background suppression and target detection rate. Simulation results show that the proposed method can more accurately detect the infrared dim target, reduce the false alarm rate, as well as reduce the processing time of a single image.The candidate targets detected in the single frame may contain false alarm points, so it’s necessary to use the track association to do the further detection.The improved track association algorithm based on the target’s movement continuity and grayscale continuity in infrared image sequence, used fuzzy theory to calculate the correlative probability between the candidate targets and target trajectory. finally according to track determination criterion, detected the real targets, excluded the false targets. The experimental results with real infrared image sequences verified the validity and accuracy of target detection algorithm proposed in this thesis, which is based on the trajectory associated. The software design of the infrared detection module is realized, which is part of the multi-sensor data fusion system in avionics systems. The follow work is done: definited the module’s external interactive interface, designed the overall framework of the infrared detection module, and completed implementation of each sub-module. The testing results verify anticipated function is achieved.