Kernel based color estimation for night vision imagery

Displaying night vision (NV) imagery with colors can largely improve observer's performance of scene recognition and situational awareness comparing to the conventional monochrome representation. However, estimating colors for single-band NV imagery has two challenges: deriving an appropriate color mapping model and extracting sufficient image features required by the model. To address these, a kernel based regression model and a set of multi-scale image features are used here. The proposed method can automatically render single-band NV imagery with natural colors, even when it has abnormal luminance distribution and lacks identifiable details.     

Third-generation sensors for night vision

Third generation sensors are under development to enhance capabilities for target detection and identification, threat warning, and 3D imaging. Distinct programs for both cooled HgCdTe and uncooled microbolometer devices are part of this thrust. This paper will describe the technology for HgCdle two-colour, high-definition imaging sensors and threat warning devices, avalanche photodiode arrays for 3D imaging, and the supporting technology being developed to enhance the readouts that support these devices. Uncooled detector initiatives will also be described to reduce pixel size in conjunction with the production of 480×640 arrays. Finally, efforts are also beginning to move both photon and thermal detectors closer to radiative-limited performance while simultaneously reducing the cooling requirements for photon detectors. The paper presented there appears in Infrared Photoelectronics, edited by Antoni Rogalski, Eustace L. Dereniak, Fiodor F. Sizov, Proc. SPIE Vol. 5957, 59571Z (2005).     

Review of night vision technology

Night vision based on technology of image intensifier tubes is the oldest electro-optical surveillance technology. However, it receives much less attention from international scientific community than thermal imagers or visible/NIR imagers due to series of reasons. This paper presents a review of a modern night vision technology and can help readers to understand sophisticated situation on the international night vision market.     

Effective contrast enhancement method for color night vision

Image fusion refers to the techniques that integrate complementary information from multiple image sensors’ data in a way that makes the new images more suitable for human visual perception. The paper focuses on the low color contrast problem of linear fusion algorithms with color transfer method. Firstly, the contrast of infrared and visible images is enhanced using local histogram equalization and median filter. Then the two enhanced images are fused into the three components of a Lab image in terms of a simple linear fusion strategy. To enhance the color contrast between the target and the background, the scaling factor is introduced into the transferring equation in the b channel. Experimental results based on three different data sets show that the hot and cold targets are all popped out with intense colors while the background details present natural color appearance. Target detection experiments through target recognition area, detection rate, target-background discrimination also show that the presented method has a better performance than the former methods.     

Transfer color to night vision images

Natural color appearance is the key problem of color night vision field. In this paper, the color mood of daytime color image is transferred to the monochromic night vision image. This method gives the night image a natural color appearance. For each pixel in the night vision image, the best matching pixel in the color image is found based on texture similarity measure. Entropy, energy, contrast, homogeneity, and correlation features based on co-occurrence matrix are combined as texture similarity measure to find the corresponding pixels between the two images. We use a genetic algorithm (GA) to find the optimistic weighting factors assigned to the five different features. GA is also employed in searching the matching pixels to make the color transfer algorithm faster. When the best matching pixel in the color image is found, the chromaticity values are transferred to the corresponding pixel of the night vision image. The experiment results demonstrate the efficiency of this natural color transfer technique.     

Spatial adaptive regularized MAP reconstruction for LD-based night vision

Target recognition plays a vital role in video surveillance, land and ocean exploration, in which imaging detection especially night vision is widely used. It is known that imagery detection is affected not only by absorption and scattering properties of atmosphere or water, but also the optical systems including lenses and sensors. In order to enhance the visual quality of the night vision detecting images to a best possible level, applications such as restoration and super-resolution reconstruction can be applied. The presented effort applies a spatial adaptive regularized MAP reconstruction method to a night vision laser diode (LD) imaging detection system. Blind, objective image quality metrics are adopted for determining the iteration time. Experimental results of the target recognition show adequate improvement in the detecting range as well as the detecting quality provided by the proposed approaches.     

Fast color contrast enhancement method for color night vision

The increasing availability and deployment of imaging sensors operating in multiple spectral bands has led to a large research effort in color image fusion, resulting in a plethora of pixel-level image fusion algorithms. In this study a simple and fast fusion approach for color night vision is presented. The contrast of infrared and visible images is adjusted by local histogram equalization. Then the two enhanced images are fused into the three components of a Lab image in terms of a simple linear fusion strategy. To obtain false color images possessing a natural day-time color appearance, this paper adopts an approach which transfers color from the reference to the fused images in a simplified Lab space. To enhance the contrast between the target and the background, a stretch factor is introduced into the transferring equation in the b channel. Experimental results based on three different data sets show that the hot targets are popped out with intense colors while the background details present natural color appearance. Target detection experiments also show that the presented method has a better performance than the former methods, owing to the target recognition area, detection rate, color distance and running time.     

Spatial frequency channels for stereoscopic depth perception

We investigated spatial frequency tuning of the mechanisms for stereoscopic depth perception, using a masking technique, and compared these tunings with those for luminance pattern detection. Observers discriminated depth direction in random-dot stereograms with various contrasts of a masking pattern, which was dichoptic stimulus of uncorrelated random dots. The strength of masking effect as a function of spatial frequency (i.e., masking function) was measured: masking function approximates the spatial frequency tuning of the mechanism that detects the depth in the test pattern. The masking functions for stereoscopic depth were found to be similar to either of the three of the six spatial frequency channels proposed by Wilson and Gelb (1984) for the detection of luminance patterns.     

微光夜视仪分辨力客观测量方法研究

利用灰度 梯度共生矩阵模型对微光夜视仪分辨力测量图像的纹理特征进行了分析 ,通过对大量图像的 5个纹理特征参数的计算 ,结果表明 ,这些参数能对微光夜视仪分辨力测量图像的纹理特征作出有效的客观评价 ,此计算结果与人眼的观察结果具有较好的一致性。     

直视型微光夜视系统最小可分辨对比度模型

从信号检测和人眼视觉角度出发,研究了直视微光夜视系统综合性能评估理论,推导出直视型微光成像系统的MRC（最小可分辨对比度）模型。该模型将光学传递特性与系统信噪比结合,建立了系统直视微光夜视MRC与目标特性、直视微光夜视仪以及人眼各部分参数之间的数学关系。验证结果表明,由该模型得到的系统极限分辨率相较于系统实际测出的分辨率相对误差为3.08%。     

折反式微光夜视物镜的光学设计

介绍一种长焦距、大相对孔径微光夜视物镜的光学设计。设计参数为焦距100 mm,相对孔径1/1.4,视场10。为增大视距,减小物镜尺寸和质量,并且满足在宽光谱范围消色差的要求,选择均为球面的折反式物镜结构,在相同焦距和相对孔径的条件下折反式系统比折射系统尺寸更小,质量更轻。在设计过程中引入曼金反射镜,增加设计自由度。经优化设计达到较好的成像质量,空间频率在50 lp/mm时,轴上传递函数大于0.4,轴外传递函数大于0.2,与像增强器极限分辨率相匹配,全视场畸变小于2 %,物镜总长达到67 mm。     

三波段微光彩色夜视方法研究

为克服夜视设备产生单色图像的缺陷,得到更符合人眼视觉的彩色夜视图像。根据RGB理论模型以及滤光片的分谱作用原理,基于第二代像增强器设计出一种彩色夜视方案,通过测量滤光片的光谱透过率,得出滤光片对像增强器信噪比的影响;提出利用全波图像的四帧融合解决像增强器信噪比低的问题,并进行实验验证。实验结果显示室内和室外情况下四帧融合彩色图像熵和方差分别达到6.8、35.39及7.0、45.07,均优于三帧融合时对应值,说明提出的四帧融合方法能够克服信噪比降低的缺陷,得到清晰且信息量丰富的夜视图像。     

夜视领域几个热点技术的进展及分析

作为夜视技术的主要技术途径,微光夜视和红外热成像技术在现代战争中具有重要的地位,掌握先进的夜视技术对于控制战场形势至关重要。随着各种现代夜视技术在世界各国军队中的装备中应用,"拥有黑夜"已逐步成为"共享黑夜"。为继续保持夜视技术的优势,美国及其盟友不断地发展更为先进的夜视技术和装备,并取得重要进展。介绍了夜视技术的几个重要的技术进展,分析了其对整个夜视技术发展以及应用的影响,希望对加快发展我国的夜视技术,促进夜视装备的有效应用起到促进作用。     

Recognizing pedestrian’s unsafe behaviors in far-infrared imagery at night

Pedestrian behavior recognition is important work for early accident prevention in advanced driver assistance system (ADAS). In particular, because most pedestrian-vehicle crashes are occurred from late of night to early of dawn, our study focus on recognizing unsafe behavior of pedestrians using thermal image captured from moving vehicle at night. For recognizing unsafe behavior, this study uses convolutional neural network (CNN) which shows high quality of recognition performance. However, because traditional CNN requires the very expensive training time and memory, we design the light CNN consisted of two convolutional layers and two subsampling layers for real-time processing of vehicle applications. In addition, we combine light CNN with boosted random forest (Boosted RF) classifier so that the output of CNN is not fully connected with the classifier but randomly connected with Boosted random forest. We named this CNN as randomly connected CNN (RC-CNN). The proposed method was successfully applied to the pedestrian unsafe behavior (PUB) dataset captured from far-infrared camera at night and its behavior recognition accuracy is confirmed to be higher than that of some algorithms related to CNNs, with a shorter processing time.     

微光彩色夜视光学系统设计与像质评价

为了充分利用可见光和长波红外所蕴含的不同信息进行数据融合、彩色显示,分析了3种不同类型双波段光学系统及其成像性能,设计了全折射光学系统、折反射光学系统和离轴三反射双谱光学系统。采用不同的分光形式,反射可见光波段,透射长波红外波段,使3种不同类型光学系统的全视场角和相对孔径分别为5°和1∶1.7;光学系统在可见光波段成像质量对应的空间频率50 lp/mm处的MTF=0.60,在长波红外波段成像质量对应的空间频率15 lp/mm处的MTF=0.30。对不同结构类型的系统特点进行分析和讨论,给出了各种像差曲线和光学传递函数曲线,为多波段彩色夜视设计了不同类型的光学系统。     

基于深度学习的无人车夜视图像语义分割

为了增强无人车对夜视图像的场景理解,在夜间模式下更快更精确地探测和识别周围环境,将深度学习应用于夜视图像的场景语义分割,提出了一种基于卷积-反卷积神经网络的无人车夜视图像语义分割方法。在传统的卷积神经网络中加入反卷积网络,构建卷积-反卷积神经网络,无需手工选取特征。通过像素到像素的学习和训练,得到图像语义分割模型,可直接用该模型预测夜视图像中每个像素所属的场景语义类别,实现无人车夜间行驶时的环境感知。实验结果表明,该方法具有较好的准确性和实时性,平均IU达到68.47。     

Apparent distance theory revision for low-light-level night vision system based on noise factor

This paper introduces basic concept of the detection range of a low-light-level (LLL) imaging system, and on the basis of previous derivation of visual distance detection equation for the LLL imaging system, the influence of noise factor on visual distance of LLL night vision system is discussed emphatically. Image intensifier is the most important part of LLL night vision device and SNR is one of the important parameters of micro-light image intensifier, whose value determines the detection distance and image sharpness of LLL imaging system in low-light conditions. In this paper, using system noise factor relation, by modifying the noise factor, a more practical and perfect horizon detecting equation is established, and the horizon estimation is carried out with the experiment, and the practicability of the corrected Horizon formula is verified. It has a certain guiding significance for improvement of the LLL imaging system and the development of night vision technology.     

微光夜视系统光晕效应定量表征与建模方法北大核心CSCD

研究了微光夜视系统输出图像中光晕尺寸和灰度分布与系统参数之间的关系,建立了系统光晕效应的定量化表征模型。首先分析了系统输出图像中光晕的产生机理;其次依据微光夜视系统成像原理,通过分析光电阴极产生的光电子数及初角度分布、光电子在光电阴极与微通道板之间的运动、光电子与微通道板非开口壁碰撞后的运动规律及微光成像系统各环节对能量的逐级传递,研究了光晕的定量化表征方法;结合上述理论研究,建立了光晕的数字仿真模型。结果表明:所提出的光晕效应表征模型能够与实验结果中光晕的灰度分布及尺寸较好地吻合;随着强光源能量增大,系统光晕效应越明显,光晕效应对系统成像质量影响越大。     

微光夜视器件划代方法初探

随着微光夜视技术不断发展,微光夜视器件呈现出了品种多、性能差异大、应用领域广的特点,如何从众多的微光夜视器件中梳理出其发展主线,已成为微光夜视器件从业者的研究热点。回顾了零代、一代、二代、三代、超二代微光夜视器件的发展历程,总结归纳了行业公认的零代、一代、二代、超二代、三代微光夜视器件的划代方法,并提出了“四代”微光夜视器件的概念,同时对非典型的几种微光夜视器件在微光夜视器件技术领域中所处的位置进行了说明,绘制了“树状图”,以便读者能够直观、准确、全面地了解每种微光夜视器件的技术特点以及它们在微光夜视器件技术领域所处的位置,对从事微光夜视器件技术研究者具有一定参考作用。