【作者】 王瑶；

【导师】 熊木地；

【作者基本信息】 大连海事大学 ， 信号与信息处理， 2008， 硕士

【摘要】 及时有效的海上搜救工作可以最大限度地减少事故损失,保障人民生命财产安全。为保证搜救工作的顺利进行,首要条件是快速准确地发现目标。红外热成像系统由于其独特的优点,在目标自动检测、识别和跟踪方面有了越来越多的应用,因此也成为今后海上目标搜救的主要发展方向。但是,现有的红外成像系统不具备目标自动识别功能,仅依靠人眼对红外视频图像进行目标检测,在船体颠簸、红外图像对比度低的情况下,极易造成视觉疲劳,丢失目标。因此,设计出一套适用于海上搜救的红外目标识别系统具有巨大的意义。本文分析了热红外图像的产生机理、系统特性以及不同波段红外辐射的成像特点。通过编程仿真对多种图像预处理算法进行对比,得出了不同算法的优缺点,并据此选择了自适应门限滤波法作为海上搜救系统的图像预处理算法。针对海上搜救中小目标红外图像信噪比低,且易受到背景杂波干扰等特点,采用了一种基于海上小目标的检测方法,该方法的特点是:首先使用自适应门限滤波来平滑背景,以减少背景噪声的干扰并对目标增强;然后采用图像局部熵函数检测出目标所在位置,这其中使用了遗传算法对函数进行优化,提高了检测效率。最后设计了基于DSP的图像处理硬件电路,使用以上算法实现海上红外小目标的实时检测。通过对标准中波段红外图像的仿真实验及数据分析,结果表明,该方法能够准确识别出目标,达到了预期效果,具有较高的实用性。更多还原

【Abstract】 Fast and effectual salvage can reduce loss in accident; ensure people’s lives and belongings safety when shipwreck occurs. It is very important that discover targets betimes and exactly to insure the salvage going on wheels. Because of its unique advantages, infrared imaging systems have been more and more applied on targets automatic detecting, identifying and tracking, it will become main developing direction on salvage. However, existing infrared imaging system does not have the function of automatic recognition, just recognize targets by human eyes. In the condition of ship bumping and low contrast infrared images, this will result in visual tired and lost targets easily. Therefore, it is significant that design an infrared targets recognition system suitable for salvage.This paper analysis principle, nature and imaging character of different wave band infrared images. It contrasts several pre-treatment algorithms by simulating programming, obtains advantages and disadvantages of different algorithms. Against salvage characters of low image SNR and being easy interfered by background clutter, this paper uses a recognition method based on small targets: first smoothes background by adaptive threshold filtering for reducing disturbing of noise, then recognizes targets by image local entropy function. In the processes of recognition, uses genetic algorithms optimizing local entropy function to enhance searching efficiency. At last, designs hardware circuit for images processing based on DSP, and realizes real-time recognition on small infrared targets on the sea with above methods.According simulating experiment and data analysis on standard media-wave infrared images, obtains the results that this method can recognize targets effective, achieve the desired objectives, have higher utility.更多还原