Small Targets Detection In Infrared Image With Complex Background

The detection problem of small point targets in IR images is a hot spot nowadays and many scholars have researched into this field and put forward valuable methods. The infrared image this paper involved is with sky-sea background conditions, and the targets are isolated bright spot with lack of structure and texture information, and the targets move slowly between the sequence images. In allusion to the above trait, we design a method to detect small target in low SNR image. The method bases on the adaptive filter with Gauss function and then we can detect the target in the filtered image with higher SNR. Applied this method to the sub-image and use the coefficient by conversed wavelet transform to construct image, then, detect the target in the constructed image. Target segmentation is that we found the threshold adaptively to segment the target in the Infrared image according to the SNR value. Simulation results show that these two methods are better than other methods, such as equal-value filtering method and median-value filtering method, in the aspect of SNR in the disposed image. Moreover, it is effective and timesaving in engineering and these methods are efficient and the capabilities of the methods are better.