Research On The Detection Of Infrared Maritime Targets In Backlighting Environment

In recent years,with the increasing of marine activities,the marine environment is becoming more and more complex,which leads to the increasing probability and complexity of marine emergencies.Marine target detection technology has become a hot issue for many scholars.In this paper,through consulting the literature about maritime emergencies,we know that the backlight environment has always been a low rate of maritime search and rescue.Therefore,this paper mainly analyzes the backlight environment often encountered in detection,and proposes two methods of infrared image target detection.The difficulty of sea target detection in the backlight environment is that the gray level of the target in the backlight image is lower than the background gray level.When the waves are strong,there will be uneven interference bright spots and dark spots on the sea surface.When the waves are weak,there will be connected smooth bright stripes on the sea surface at random.Besides the influence of the backlight environment itself,there may be interference from islands,reefs,etc.Through the analysis of the gray and contrast characteristics of the backlight marine infrared image,it can be seen that the problem of this kind of image target detection is that the gray and contrast information of the target is not significant.To solve this problem,we can adopt the frequency domain processing method to filter out the interference of background clutter.According to the analysis of the gradient direction characteristics of the backlight sea infrared image,the problem of image target detection is that the target in the weak wave backlight image has the strongest gradient correlation in the whole image,but the interference bright spot in the strong wave backlight image has the stronger gradient correlation.Aiming at this problem,the bright spot interference in the strong wave backlight image can be removed so that the target has the whole image The strongest gradient correlation.Through the analysis of the advantages and disadvantages of typical sea target detection methods based on gray,contrast and gradient direction features,the visual attention model method is more suitable for the detection of backlight sea target.Compared with gray and contrast features,it is unnecessary to preprocess the image to extract the target by using gradient correlation,which is beneficial to reduce the time complexity of the algorithm.Therefore,this paper proposes a single feature visual attention model based on the backlight sea infrared image target detection algorithm.Firstly,the algorithm uses standard deviation method to judge the strength of the wave background in the image through comparative analysis.Secondly,different "center periphery" difference operators are selected for the strong and weak wave background to improve the significance of the target.Finally,the fast segmentation method is applied to extract the target.The experiment shows that the algorithm has low time complexity,but there are two problems at the same time.One is that when there is strong wave backlight environment,the real target is not missed but there is a false alarm.The second is that although the low complexity fast segmentation method can guarantee the advantage of processing speed,the area of the target in the final detection result is attenuated,which will inevitably affect the subsequent target location and target tracking.In order to solve the above two problems,a single frame detection algorithm based on two feature visual attention model is proposed.First of all,through the analysis of the gray and contrast characteristics of the target in the sea backlight environment,a Gaussian difference preprocessing method is selected to achieve the effect of enhancing the target significance.Then,the standard deviation is selected as the method to judge the strength of wave background in the image.After that,the "center periphery" difference operator of strong wave backlight image processing is improved to highlight the target significance "center periphery" difference operation,and the weak wave backlight image adopts the single feature"center periphery difference" strategy to achieve the effect of restraining waves and enhancing the target significance due to strong background interference.Finally,the improved binary method is used to segment the target automatically,and the original area of the target is maintained to achieve a more accurate detection result.In order to verify the effectiveness of the algorithm,this paper has carried out many multi scene marine infrared image acquisition experiments for the backlight environment,and used the real infrared image and the infrared image collected on the Internet to verify the reliability of the algorithm.Finally,the experimental results show that the algorithm based on two feature visual attention model has lower false alarm rate and false alarm rate than the algorithm based on single feature visual attention model,wavelet transform and improved visual attention model.In terms of time complexity,the target detection algorithm with single feature is lower than that with two features,while the time complexity of the comparison method is higher than that of the method proposed in this paper.