海上搜救视觉增强系统-视觉增强系统在海上SAR系统的应用.docx

海上搜救视觉增强系统施朝健 许开宇 肖宝家 莫剑英（上海海事大学）摘要：由于人眼的生理特性造成海上复杂环境中的搜索效果不佳，研究应用于海上SAR系统的视觉增强系统是十分必要的。本文针对搜救环境中人眼观测的弱点，提出了基于可见光成像系统和红外成像系统的视觉增强系统进行海上搜救的方案，给出了系统框图。该系统辅助搜救人员进行搜救，将提高海上目标尤其是弱小目标探测的灵敏度和发现率，能够有效地挽救人民的生命和财产，具有巨大的经济效益和社会作用。关键词：视觉增强系统；搜寻救助；人眼视觉特性；目标检测Vision enhancement system for maritime search and rescueShi Chaojian, Xu Kaiyu, Xiao Baojia and Mo JianyingShanghai Maritime UniversityAbstract Because of the limitation of physiological characteristics of human eyes, it is necessary to develop vision enhancement system for maritime search and rescue (SAR) to improve the poor searching performance of human vision at sea. And a system with its framework based on the visible light imaging system and infrared imaging system is proposed in the paper to compensate the weakness of the human observation in the SAR environment. The system will promote the sensitivity and the detection rate of targets, especially small ones, at sea, ensure the safety of life at sea, and reduce economical losses in marine accidents.Keywords vision enhancement system; search and rescue; human vision characteristics; target detection1 引言随着航运业的发展,我国水上交通安全保障和应急搜救任务日益繁重。2006年，我国海上搜救中心和海事机构共组织协调了海上搜救行动1620次，组织、协调各类船舶5322艘次。共发生遇险人员17498人。中国政府和海事部门非常重视研究和发展海上搜救技术。国家中长期科学和技术发展规划纲要指出：“重点开发交通事故预防预警、应急处理技术，开发运输工具主动与被动安全技术，交通运输事故再现技术，交通应急反应系统和快速搜救等技术。”交通部公路水路交通“十一五”发展规划重点还明确提出“十一五”期间做到“现场救助能力明显提高，人命救助有效率由目前的87%提高到90%左右。”为了达到规划要求，对海上救助搜寻视觉增强系统的关键技术进行研究是十分必要的。在过去几年，中国的海上搜救组织发生了很大的变化，体现在参与了国际上搜救的合作和采用了先进的技术。然而，现有的海上救助搜寻方法，尤其是搜索人，主要借助人眼进行搜寻。而人眼视觉的生理特性决定了在复杂环境和疲惫状态下，搜索效果不佳。研究视觉增强系统（Vision Enhancement System, VES）辅助人眼进行海上搜救的关键技术是非常必要的。视觉增强系统应用机器视觉设备和技术增强和扩展了人眼的视觉。它是智能交通和交通安全领域的重要前沿基础技术和研究热点1-3，在汽车驾驶、飞机导航等方面得到较为广泛的应用4-6，尤其在不利天气或夜间驾驶和飞行时有效增强了人的视觉，在保障交通安全方面起到良好的作用7-9。同时，视觉增强技术也被应用于辅助老年人、视力障碍人员改善视力和特殊环境中增强视力等10-12。这些视觉增强系统的研究成果对本课题的研究有重要的借鉴价值。2 人眼的视觉特性现有的海上救助搜寻方法主要借助望远镜或者红外望远镜依靠人眼进行搜寻，因而人眼的结构特征和视觉特性是值得关注的。2.1 人眼结构图1 人眼结构图人眼结构图如图1所示。是由水晶体、虹膜和视网膜等组成的。正常视力情况下，眼睛可以聚焦到物体并能在视网膜的中央凹（fovea）成像。视网膜中央凹处是视网膜的中心和最敏感之处，产生精确的视觉。在0º位置附近，为了看清小的细节部分视网膜的一小部分会起作用。在大的区域内，2º - 4º，是用于看近处物体的。当视觉使用所有的视网膜定向的看目标时，没有能力分辨细节。光是从物体表面反射到视网膜的感光细胞，然后根据神经的冲击进行响应。这里有两种类型细胞，锥状和杆状细胞。锥状细胞是一种明视觉器官，只有在明亮条件下它才起作用，但却能够分辨颜色和细节。杆状细胞是一种暗视觉器官，它对入射光的强度很敏感，它能分辨亮度的差别，而对颜色的分辨本领极差，不能分辨颜色与细节。最好的视力是在最中部的凹处,称为微孔（foveola）. 中央凹处大小是1.5毫米直径( 5.2 º )，微孔占有0.3毫米直径( 1 º)。其中有50000个锥状细胞,但没有杆状细胞，这就是为什么我们在弱光下无法看到任何东西。2图2 视网膜上的锥状细胞和杆状细胞的分布图2所示为眼的锥状和杆状细胞分布的密度图。在缺少细胞分布的地方就是“盲点”（Blind spot）。除此区域外，分布情况基本上是关于中央凹对称的。值得注意的是，锥状细胞在视网膜中央部分是密度最高的，而杆状细胞中心偏移了约20度后开始增加密度,然后下降直到视网膜的边缘. 2.2 视觉外界光辐射到人眼中刺激视网膜所引起的知觉叫做视觉。视觉是一个复杂的生理过程。外界光线通过角膜、虹膜、晶状体聚焦在视网膜上, 刺激感光细胞发出神经脉冲, 经视神经传递到大脑产生视觉。人眼有视觉暂留现象，当两种不同的色光间隔时间很短，先后对视网膜刺激，视网膜就分不出刺激的先后，只能产生一个总体的刺激知觉，这便是视觉的时间混色效应。此外，人眼还具有一定限度的分辨本领，当两束不同的色光同时对视网膜极小的范围刺激后，视网膜在某一极小范围就无法分辨两种刺激，只能产生总体的刺激知觉，这便是视觉的空间混色效应。结合人眼的视觉特性，以及对人眼视觉特性的实验和研究，可以得出几种典型的人眼视觉模型14-18。即，指数型、高斯型、Barten型和复合模型，分别如图3 所示，是几种类型的M T F 曲线。由此可以看出：a. 高斯型和指数型是低通滤波器；Barten 和复合模型是带通滤波器，复合模型较Barten 模型向低频移动；b. Barten模型和复合模型在大于15cyc/ deg的区域大体处于0.20.3 mrad 的高斯型曲线之间，而指数型则偏高；图3 几种人眼视觉的M T F 曲线c. 由于Barten 模型和复合模型的复杂性, 以往分析大都采用指数型或高斯型MTF模型；d.从以上视觉模型中可以看出人眼的视觉特性的调制传递函数响应是非均匀的，这在搜救的复杂环境中，容易将弱小目标抑制掉，不易搜索到小目标。3 视觉增强系统尽管SOLAS公约规定船舶装有GMDSS设备等对于搜救环节已经有所改善，但是对许多非SOLAS公约船如渔船、小艇筏的检测效果一直是不理想的，加之海上环境非常复杂，对于落水人员的搜救更是束手无策。此外，目前海上搜救大都还停留在使用多个常规搜寻设备联合搜寻，依靠人工视觉进行判断的阶段。在搜救的复杂环境中，根据人眼的生理特征以及人眼的视觉模型，造成了搜救人员在逆光观测、夜间观测、小目标检测、海浪环境和长期观测中，很难发现被搜寻的小目标，延误救助时间。视觉增强系统，可以克服上述的缺陷和不足，并应用于对水上船筏和遇险人员的救助搜寻，可以提高救助搜寻的速度和准确性，对提高救助的成功率和效率有重要作用。应用于海上救助搜寻的视觉增强系统，通过可见光、红外成像设备对目标进行搜寻，辅助人眼的视觉系统，从而获得遇险落水人员、救生艇筏等目标的位置及动态。该视觉增强系统可以很好的辅助救援人员完成搜救工作，尤其是对救生艇筏和遇险落水人员等海上弱小目标，具有很好的检测和识别效果。4 海上搜救视觉增强系统根据视觉增强系统各个部分在海上SAR系统中所起的作用，建立了视觉增强系统，其结构框图如图4所示。人眼观测可见光成像系统红外成像系统目标检测目标检测图像融合场景标注搜救人员决策海面环境图4海上搜救视觉增强系统结构框图在搜救过程中，由可见光成像和红外成像系统组成机器视觉系统，搜寻目标时辅助人眼进行观测并克服了视觉系统的生理缺陷。高分辨率和高速成像系统可以用于捕获可见光信息。同时红外成像系统也可以进行目标检测，这在夜间搜寻和白天克服波浪影响时都能够起到作用。利用图像融合技术，改善了单个视觉系统的观测敏感度，同时在场景中标注出可能的被搜救对象，有利于搜救人员做出决策。5结论由于人眼视觉的限制，研究先进的技术提高海上搜救性能是必要的。在海上SAR系统中应用视觉增强系统，可以提高搜救人员海上搜寻的速度和准确性，对提高救助成功率和效率有重要作用，可以提高我国水上救助能力，追赶上发达国家的遇险救助能力,为我国的航运事业做好保障工作。它将有效的挽救人民生命安全和财产，有较大的经济效益和社会作用。参考文献1 S. Boverie, A new class of intelligent sensors for the inner space monitoring of the vehicle of the future, Control Engineering Practice 10 (2001), no. 2002, 11691178.2 Druid, "Vision enhancement system -does display position matter?," Department of Computer and Information Science, Linköping University, Vårgårda, Sweden, 2002.3 K. Rumar, "Night vision enhancement systems: What should they do and what more do we need to know?" Transportation Research Institute, The University of Michigan., Ann Arbor, Michigan 48109-2150, 2002.4 Schenkman, B. & Brunnström, K., Aspects of Vision Enhancement Systems in cars based on infrared imaging. Acreo-report, No acr003758. 2000.5 E. Hollnagel, Effects of a night vision enhancement system (NVES) on driving: Results from a simulator study, The Second International Driving Symposium on Human Factors in Driver Assessment, Training and Vehicle Design, 2003. 6 F. Sadjadi, M. Helgeson, J. Radke and G. Stein, Enhanced vision for adverse weather aircraft landing, International Journal of Infrared and Millimeter Waves 17 (1996), no. 1, 1-50.7 Nilsson, L. & Alm, H. Effects of a Vision Enhancement System on Drivers Ability to Drive Safely in Fog. VTI särtryck, No 264. Linköping, Sweden, 19968 Karlsson, J. (2002). Vision Enhancement Systems - Support for Night-time Driving? Master Thesis, Linköping University.9 Gish, K. (2001). Driving Behaviour and Performance using an infrared Night Vision Enhancement System. Final report. Transportation Safety Division, PA: USA.10J. K. Caird, W. J. Horrey and C. J. Edwards, Effects of conformal and nonconformal vision enhancement systems on older-driver performance, vol. 1759 2001.11R. C. Peterson, J. S. Wolffsohn, M. Rubinstein and J. Lowe, Benefits of electronic vision enhancement systems (EVES) for the visually impaired, American Journal of Ophthalmology 136 (2003), no. 6, 1129-1135.12Blair R.Dobbie,Steven A.Brillhart,Charies D.Willey,Mounting/adjusting mechanism for vision enhancement system,US Patent 7193783.March 20,2007.13周燕，金伟其. 人眼视觉的传递特性及模型.光学技术. 2002. Vol128. No11:57-59,6214Lloyd J M. Thermal imaging system. 尹白云等译,北京: 国防工业出版社, 1981) .15Ratches J A l. Night vision laboratory static performance model for thermal viewing systems R . U. S. Army Electronics Command , Report No1 ECOM 7043 , 197516 Barten P GJ. The SQRI method : a new method for the evaluation of visible resolution on a display J , Proc1 Soc1 Inf1 Dispel , 1987 ,28 :253 26217Jacobson RE. An evaluation of image quality metrics, The journal of photographic science , 1995 ,43 : 7 16Vision Enhancement System for Maritime Search and RescueSHI Chaojian, XU Kaiyu, XIAO Baojia and MO JianyingShanghai Maritime UniversityAbstract- Because of the limitation of physiological characteristics of human eyes, it is necessary to develop vision enhancement system for maritime search and rescue (SAR-VES),which improve the poor searching performance of human vision at sea. A system with its framework based on the visible light imaging system and infrared imaging system is proposed in the paper to compensate the weakness of the human observation in the maritime search and rescue environment. The system will promote the sensitivity and the detection rate of targets, especially small ones, at sea, ensure the safety of life and reduce economical losses in marine accidents.Keywords-vision enhancement system; search and rescue; human vision characteristics; target detection1 IntroductionMaritime search and rescue (SAR) became an enormous task with the vast growth of marine transportation and other marine activities. In the year of 2006, the maritime SAR centers and maritime authorities in China organized and coordinated 1620 SAR operations, which involved 5322 vessels and 17498 human lives. Chinese government and maritime authorities have underscored the importance of research and development of SAR technology. In the national program for medium- to long-term scientific and technological development, transportation safety technologies, such as traffic accident prevention, warning and emergency processing techniques, active and passive safety techniques of transportation vehicles, traffic emergency response system and quick search techniques, are listed on the top agenda. It is also specified explicitly in the emphasis of 11th five-year development plan of highway and waterway transportation, proposed by the Ministry of Communications, that, in the 11th five-year period, the maritime SAR capability should be improved obviously, and the rate of life saving should be increased from 87% at present to about 90%. The past few years have witnessed tremendous changes in the organization of marine rescues. A large part of this evolution stems from the involvement on an international scale and the contribution of the advanced technology. However, current maritime search operation, especially searching people over board, depends mostly on human eyes. Because of the limitation of human eyes, it is necessary to study the key technique of vision enhancement system for maritime search and rescue. Vision enhancement technique is a potential approach to overcome the limitation of human eyes in maritime SAR, and thereby may hopefully improve the searching performance in complex environment or in a fatigued state of human being.Vision enhancement system applies machine vision equipment and technique to enlarge the vision extent of human eyes. It is an advanced technique and an important research hotspot in the field of intelligent transportation and traffic safety1-3. It has a wide spread application in the fields of road vehicle driving and airplane navigation4-6. It enhances the human vision effectively in bad weather or night driving and flying, and is of advantage in ensuring the traffic safety7-9. At the same time, the vision enhancement technique has also been applied to aid the older-driver or the visually impaired to improve their performance10-12. 2 Characteristics of human eyesAt present, most information used in maritime search and rescue is human vision. The structure of human eyes and its vision characteristics are worth noticing.2.1 Structure of human eyes Fig. 1. Structure of a human eye The structure of a human eye is shown in Fig. 1. A condition for normal vision is that the eye can focus on an object and keep the image of the object steady projected on fovea. Fovea is the most central and sensitive part of retina and provides the most accurate precise vision. Only a very small part of the retina is used in order to see small details around a 0º position. A larger area, 2º - 4º from the centre, is used when looking at very near objects. Vision used for orientation uses all of retina without the ability to discern small details outside the centre. Light that is reflected from a surface hits the photoreceptors on the retina, which then translates the light to nerve impulses. There are two types or receptors, cones and rods. These are differently distributed over retina. The receptors measure light and have different functions. The cones are specialized on distinguishing different colors and discern small details in good light. The rods are specialized on registration of very low levels of light. If the light is too weak, colors cannot be perceived because the cones need a light level of at least 10 cd/m2, which is the lowest level of photopic vision. The best acuity is achieved in the most central part of fovea, called foveola. The size of fovea is 1.5 mm (5.2º) in diameter while foveola occupies 0.3 mm (1º). There are 50 000 cones but no rods at all in fovea. That is why it is not possible to see anything with fovea in very weak light conditions.2 Fig. 2 shows the density of rods and cones for a cross section of the right eye passing through the region of emergence of the optic nerve from the eye.The absence of receptors in this aera result in the so called “blind spot”. Except for this region, the distribution of receptors is radially symmetric about the fovea. Receptor density is measured in degrees from the fovea. It can be noticed that the cones are mostly dense in the center of retina (in the center area of the fovea), and the rods increase in density from the center out to approximately 20 º off axis and then decrease in density out to the extreme periphery of the retina.Fig. 2 Distribution of rods and cones in the retina2.2 Human visionThe perception caused by the external light radiating into the human eyes to stimulate the retina is called vision. Vision is a complex physiological process. External light is focused on the retina through cornea, iris and lens, and stimulates photo-sensory cell to generate nerve pulse, which is transmitted to the brain through vision nerve to produce vision. Because of the visual temporary retaining phenomenon of human eyes, when the interval of two different lights is very short, the retina cant distinguish the order of the stimulation by the lights. It only produces a total stimulation perception, causing the visions time blending effect. Besides, human eyes also have a limited distinguishing ability, when the minimal region of the retina is stimulated by two different lights at the same time, the retina cant distinguish two stimulations in the minimal region, only produce total stimulation perception, which causes visions space blending effect.Fig. 3. A few MTF curves of the human visionSeveral typical frequency response models have been introduced to represent the characteristics of human visual systems 13-17, i.e., exponent model, Gauss model, Barten model and compound model, which are illustrated in Fig. 3 by curves of modulation transfer function (MTF). From theses curves, it can be noted that, Gauss model and exponent model are low-pass filters, while Barten model and compound model are band-pass filter. Compared to Barten model, compound model is inclined to low frequency. When the region is greater than 15 cyc/deg, Barten model and compound model are generally located between 0.2 mrad gauss model curve and 0.3 mrad one, but the exponent one is much higher. Because of the complexity of Barten model and compound model, exponent MTF model or gauss MTF model is mostly used in the previous analysis. Modulation transfer function response of human visual characteristic is non-uniform in the above visual models. This can easily inhibit the small target and cant search the small one.3 Vision enhancement systemSOLAS convention prescribes that ships must be equipped with GMDSS equipments, which have improved the search and rescue. However, for many non SOLAS convention ships, such as fishing boats and small crafts, the detecti