Infrared Small Target Detection Under Long Distance Sea-Sky Background

The long-distance infrared small target detection under complex sea-sky conditions has always been a hot and difficult point in the field of image processing.Practical applications often require the infrared detection system to capture the target as far as possible and get the target information.Thus,infrared small target in sea-sky background is generally small,lack of texture information and brightness characteristics.In addition,due to cloud and Sea clutter interference,the SCR(signal to clutter ratio)of infrared image is so low that the target is easily to be submerged in the background noise and clutter.Therefore,small target detection in complex sea-sky background remains rather difficult and challenging.In order to realize adaptive detection of different size targets in different sea-sky backgrounds,and reduce the influence of sea clutter and noise,this thesis do deeply research into two new single-frame infrared small target detection algorithms and do some optimization on the performance of target enhancement,background suppression,high true positive rate,low false positive rate,realtime,etc.The specific research work contains two parts,listed as follows:For the purpose of adaptive detection of different-size targets and better background suppression effects,which eliminate background clutter and noise in complex sea-sky background,this thesis study and improve the latest line-based image reconstruction and information entropy enhancement algorithm.It is found that the algorithm is difficult to distinguish targets and backgrounds in complex sea-sky background,because the use of entropy-induced suppression enhances the targets as well as the noise,resulting in the disappearance of target.And meanwhile this algorithm cannot be self-adaptive for targets with different sizes.Therefore,this thesis proposes an infrared small target detection algorithm via line-based reconstruction with adaptive radius.It eliminates the entropy-induced suppression from the original algorithm,and adopts the adaptive target-radius calculation,so that it can adaptively detect different size targets.The simulation experiments are processed on the new algorithm and five other algorithms,such as Top-Hat,wavelet,etc.The experimental results indicate that the new algorithm can achieve adaptive and stable detection for different size targets in different sea-sky backgrounds,ensuring higher true positive rate,lower false positive rate and superior background-suppression performance.The new algorithm is suitable for infrared small target detection in complex sea-sky background with very few requirement for realtime.In order to get better realtime and target enhancement performance as well as suppress sea clutter of image edge,the latest algorithm of multiscale patch-based contrast measure is studied and improved.The original algorithm can adaptively detect targets with different sizes,but it is not suitable for complex sea-sky background.Considering the obvious directivity of the sea clutter,this thesis integrates this algorithm with multi-directional composite window to simultaneously utilize gray level and contrast ratio to distinguish targets.So that it adapts to infrared small target detection in complex sea-sky background.At the same time,for fear of the low adaptability of image-segmentation threshold,which remains constant in different cases,the adaptive control algorithm is employed to adjust the threshold to different situations.Thus,a new method is proposed to combine the patch-based contrast measure,multi-directional composite window and adaptive control algorithm in infrared small target detection.The new algorithm is compared with six other algorithms,such as Top-Hat transform,wavelet transform,etc.The experimental results show that compared with other algorithms,the new algorithm can achieve stable and adaptive detection for different size targets in different sea-sky backgrounds,and meanwhile it can ensure high true positive rate,low false positive rate,better target enhancement and real-time performance.The algorithm is suitable for infrared small target detection with high requirement for realtime,and it can detect target with low-contrast and energy.