Multi-scale Sky Background Infrared Target Detection

With the development of infrared imaging technology and digital image processing technology, infrared target detection technology is widely used in missile guidance, target tracking, automatic control, artificial intelligence and many other areas. However, due to the complexity of the scenes and the lack of information, the technology of infrared target detection is a complex process.In this paper, we introduced a new algorithm based on morphological filtering pretreatment and multi-frame pipeline accumulation for weak infrared target detection, and a new algorithm based on edge detection and double window segmentation for close infrared target detection. And then, on this basis, we proposed a new algorithm for multi-scale infrared target detection.There are two key objectives for infrared target detection. The first is the target enhancing and the background suppression, the second is the segmentation by threshold. In response to the first key, we analyzed the mean filtering, one-dimensional and two-dimensional morphological filtering, and concluded that the two-dimensional morphology filter can be a better background suppression and target enhancer. For the second key, we analyzed the classical segmentation method on the Gaussian distribution, and finally made a appropriate method of determining the threshold.According to the situation that the close targets have high gray values and strong edges, we proposed a new algorithm for target detection, which is based on two-step segmentation. Tests show that the algorithm can accurately detect the close infrared targets.