Multiwavelet Multi-Resolution Texture Analysis Based Small Detection In Infrared Image

Multiwavelet is based on wavelet analysis in recent years. Multiwavelet can possess symmetry, orthogonality, short support for a given approximation order.Since infrared imaging guidance is a very important guidance system with strong capability to resist perturbations, it has become a main trend of development. The detection of targets in long distance with small spatial extent and low contrast in infrared clutter background is a problem of crucial importance.In this paper, the application of multiwavelet analysis in small infrared target detection is studied. The detection of dim targets in infrared images is difficult. Because the dim targets have a small area, dim targets often submerge in complex clutter background which are difficult to detect. So the detection of dim targets are a bottleneck in infrared images. The detail of the background is weaken in some grade. The background can show some texture characteristics. Through the background texture analysis, this paper picks up the part texture characteristic from the maps after multi-scale multiwavelet. And calculates the distance between characteristic vector and center characteristic. This can gain a correlative multi-scale distance map. The histogram of distance map is employed for the automatic selection of threshold value. Experiments show that this approach can achieve quite satisfactory-results. And this experiment spends 20 seconds in the computer and the computer's CPU is 1.7G and EMS memory is 256M.Designs a new method that is robust to noise to enhance the contrast between target and background, and a novel technique is presented to extract the spot target in image, based on the normal distributing of the local energy of the target. Experiment results with a lot of real IR image show its effectiveness in extraction spot target in a single frame.