基于时空双流卷积神经网络的红外行为识别

针对红外视频人体行为识别问题,提出了一种基于时空双流卷积神经网络的红外人体行为识别方法。通过将整个红外视频进行平均分段,然后将每一段视频中随机抽取的红外图像和对应的光流图像输入空间卷积神经网络,空间卷积神经网络通过融合光流信息可以有效地学习到红外图像中真正发生运动的空间信息,再将每一小段的识别结果进行融合得到空间网络结果。同时将每一段视频中随机抽取的光流图像序列输入时间卷积神经网络,融合每一小段的结果后得到时间网络结果。最后再将空间网络结果和时间网络结果进行加权求和,从而得到最终的视频分类结果。实验中,采用此方法对包含23种红外行为动作类别的红外视频数据集上的动作进行识别,正确识别率为92.0%。结果表明,该算法可以有效地对红外视频行为进行准确识别。     

A novel video fusion framework using surfacelet transform

A novel video fusion framework based on the three-dimensional surfacelet transform (3D-ST) is proposed in this paper. Different from the traditional individual-frame based video fusion methods, the proposed framework fused multi-frame images of input videos as a whole rather than frame by frame independently with the 3D-ST. Furthermore, under the proposed framework, two ST-based video fusion algorithms are proposed. In the first algorithm, no special treatment is performed on the temporal motion information in input videos, and only a spatial-temporal region energy-based fusion rule is employed. While in the second algorithm, a modified z-score based motion detection is performed to distinguish the temporal motion information from the spatial geometry information, and then a motion-based fusion rule is present. Experimental results demonstrate that, with the motion selectivity of the 3D-ST, existing static image fusion rules can be extended to video fusion under the proposed framework. Both of the two proposed fusion algorithms significantly outperform some traditional individual-frame based and motion-based methods in spatial-temporal information extraction as well as in temporal stability and consistency. In addition, the second proposed algorithm is with high computation efficiency and can be applied to real-time video fusion.     

Hierarchical CodeBook for background subtraction in MRF

Foreground detection is the key low-level fundamental work in intelligent video surveillance. This paper proposed a hierarchical background subtraction algorithm consisted of block-based stage and pixel-based stage for it. In block-based stage, obvious backgrounds got detected via block-based CodeBook, leaving spatial relations among suspicious foreground pixels undestroyed. Pixel-based stage further eliminated the left background pixels with the introduction of spatial and temporal relations in a MRF-MAP framework. Then comparative experiments were conducted to evaluate the performance of the scheme in three dimensions – detection accuracy, update speed and memory consumption. Proposed approach possesses the highest detection precision and consumes the second least memories. And the update speed is of real-time level.     

Object based dual watermarking for video authentication

Video cameras have been widely installed in public facilities for the surveillance applications. So, video authentication has becoming increasingly attractive. This paper presents a dual watermarking for video authentication based on moving objects. For each frame, the frame index, as a watermark is first embedded into the moving objects of the corresponding frame using a reversible watermarking method, aiming to detect the temporal tampering. Then the principle content and the details of the moving objects combined with the authentication code, as the other watermark, are embedded into the frame for spatial tampering location and recovery. Specially, a synthesized frame method is proposed for lossless recovery of moving objects and effective extraction of frame index. Statistical analysis and experiment results show that the proposed method can locate spatial, temporal and spatio-temporal tampering accurately. The spatial tampered regions can be recovered approximately and the moving objects can be restored completely when the tampered area is limited.     

Compatibility of temporal multiplexed spatial light modulator with optical image processing

The tendency today is to replace high-dynamic light modulators by high-speed binary ones (of which the micromirror is the best example). This kind of spatial light modulators (SLM) fulfils all the present needs in displays. They are used in optical communications as binary systems and also in display applications (video projectors) with temporal multiplexing, in order to generate greyscale or colour images. In optical processing, and in the majority of coherent applications, temporal dithering introduces some distortions. In this paper, this point is studied with simulations. We point out that temporal multiplexing cannot be used in the Fourier plane. In the imaging plane, the distortion is weak if the filter has a positive impulse response.     

Spatial and temporal bilateral filter for infrared small target enhancement

This paper presents a spatial and temporal bilateral filter (BF) to detect target trajectories, by extracting spatial target information using a spatial BF and temporal target information using a temporal BF. Background prediction when it is covered by targets is the key to small target detection. In order to apply the BF to a small target detection field for this purpose, this paper presents a novel spatial and temporal BF with an adaptive standard deviation to predict spatial background and temporal background profiles, based on analysis of the blocks surrounding a spatial and temporal filter window. In order to discriminate between the edge or object regions with a flat background and the target region spatially and temporally, spatial and temporal variances of the blocks surrounding the filter window are calculated in a spatial infrared (IR) image and temporal profile. The spatial and temporal variances adjust standard deviations of the spatial and temporal BF. Through this procedure, spatial background and temporal background profiles are predicted, and then small targets can be detected by subtracting the predicted spatial background (and temporal background profile) from the original IR image (and original temporal profile) and multiplying spatial and temporal target information. To compare existing target detection methods and the proposed method, signal-to-clutter ratio gain (SCRG) and background suppression factor (BSF) are employed for spatial performance comparison and receiver operating characteristics (ROC) is used for detection-performance comparison of the target trajectory. Experimental results show that the proposed method has a superior target detection rate and a lower false-alarm rate.     

High-speed temporal characterization and visualization of spatial light modulators and flat-panel displays

An apparatus that can characterize and visualize temporal dynamics of spatial light modulators (SLM's) and flat-panel displays is constructed and evaluated by use of a commercially available SLM. The apparatus is based on the stroboscopic video sampling method and has a temporal resolution of the order of microseconds, permitting measurement of a long event (>100ms) with a high signal-to-noise ratio. Experimental results demonstrate the visualization of the temporal image sequencing and addressing in an SLM.     

Experimental analysis and visualization of spatiotemporal patterns in spouted fluidized beds

A numerical characterization based on experimental data of the spouting regime in a two-dimensional fluidized bed is presented. The aspect ratio of the bed allowed for good visualization of the spouting and solids circulation as the spouting jet gas velocity was varied to highlight the visited bifurcation sequence. Digital video sequences were recorded and then preprocessed for numerical analysis. In this paper, the proper orthogonal decomposition (POD) was applied to these data sets in order to identify and separate the dominant spatial features from the temporal evolution of the spouting dynamics. The results indicate that the overall spatiotemporal dynamics can be captured by a few POD eigenfunctions, and that the POD amplitudes can be used to distinguish between varying degrees of spouting.     

光场时空对称性及其在光学技术中的应用

本文论述光波场的时问、空间对称性,包括拍频导致的“空间放大”与“时间放大”;多束光干涉提高条纹细锐度“空间压缩”、锁模技术中脉冲的“时间压缩”;空间调制与时间调制;非线性效应中的时间倍频与空间倍频;空间相干性与时间相干性;时域测不准与空域测不准。     

基于STA-CRNN模型的语声情感识别*

语声情感识别对人机交互和情感计算研究领域具有重要作用,各类研究方法层出不穷。近期研究学者应用卷积神经网络和长短期记忆网络方法提取对数Mel谱图空间特征和时间特征,取得了一定的成果。然而不论是卷积神经网络还是长短期记忆网络提取特征时,都会产生特征冗余,导致语声情感识别效果下降。针对这一问题,该文提出了一种基于时空注意力机制的卷积-递归神经网络模型,采用对数Mel谱图和其一阶差分、二阶差分作为特征输入,在使用卷积神经网络提取空间特征和长短期记忆网络提取时间特征时,加入空间注意力和时间注意力机制,从而使上述网络能够更好地提取到对数Mel谱图中有效表征情感的空间特征和时间特征。该模型在Emo-DB和IEMOCAP语声数据集上的加权准确率分别达到86.8%、69.4%,未加权准确率分别达到84.7%、65.5%,优于当前大多数先进方法。     

Simultaneous Spatial and Temporal Focusing in Nonlinear Microscopy

Simultaneous spatial and temporal focusing (SSTF), when combined with nonlinear microscopy, can improve the axial excitation confinement of wide-field and line-scanning imaging. Because two-photon excited fluorescence depends inversely on the pulse width of the excitation beam, SSTF decreases the background excitation of the sample outside of the focal volume by broadening the pulse width everywhere but at the geometric focus of the objective lens. This review theoretically describes the beam propagation within the sample using Fresnel diffraction in the frequency domain, deriving an analytical expression for the pulse evolution. SSTF can scan the temporal focal plane axially by adjusting the GVD in the excitation beam path. We theoretically define the axial confinement for line-scanning SSTF imaging using a time-domain understanding and conclude that line-scanning SSTF is similar to the temporally-decorrelated multifocal multiphoton imaging technique. Recent experiments on the temporal focusing effect and its axial confinement, as well as the axial scanning of the temporal focus by tuning the GVD, are presented. We further discuss this technique for axial-scanning multiphoton fluorescence fiber probes without any moving parts at the distal end. The temporal focusing effect in SSTF essentially replaces the focusing of one spatial dimension in conventional wide-field and line-scanning imaging. Although the best axial confinement achieved by SSTF cannot surpass that of a regular point-scanning system, this trade-off between spatial and temporal focusing can provide significant advantages in applications such as high-speed imaging and remote axial scanning in an endoscopic fiber probe.     

Effect of geometrical stress concentrators on the band formation and the serrated deformation in aluminum–magnesium alloys

The effect of holes on the band formation and the serrated deformation in planar specimens of aluminum–magnesium alloys AlMg5 and AlMg6 is studied by high-speed video filming of moving deformation bands. It is found that the concentration of an elastic field near a hole causes early nucleation of macrolocalized deformation bands and decreases the critical deformation of the first stress drop. Differences between the spatial–temporal patterns of deformation bands near holes under various deformation conditions are revealed.     

自适应H.264可变块编码模式选择算法

为了减少H.264/AVC帧间编码模式选择的计算复杂度,利用编码模式之间的相关性以及视频序列时空域的相关性,提出了有选择性的小块搜索技术和有选择性的帧内编码模式搜索技术。模拟结果表明,该算法在保持率失真性能的前提下可以大幅度减少模式选择的计算复杂度,有利于H.264的实时应用。     

Studies on Temporal Resolution Characteristics of Flame Temperature for Flame Atomic Absorption Spectroscopy

In this paper we report a method and experiment system developed by us to useful for temporal and spatial temperature distribution of the flame for atomic obsorption spectroscopy in detail. The method and system principle is based on the modified sodium line reversal method. The studies on temporal (20–50 μsec range) and spatial (π 1mm) resolution of the flame temperature aim at establishing optimum analysis conditions and improving analysis characteristics of flame atomic absorption/emission spectroscopy.     

Excitation of standing gravity-capillary waves at an interface between two immiscible liquids by a periodic sequence of ultrasound pulses

The experimental characterization of gravity-capillary waves excited at an interface between two immiscible liquids by a periodic sequence of focused ultrasound pulses propagating perpendicular to the interface is presented. The experiments have been performed in a glass cylinder filled with two liquids: Fluorinert FC70 and silicone oil. The spatial and temporal evolution of the interface deformation is recorded by a high-speed video camera. The effect of the duration and amplitude of ultrasound pulses on the amplitude and shape of interfacial oscillations is analyzed. Prospects of the proposed approach and possible applications of the observed phenomena are discussed.     

Multiwavelength pulse generator using time-lens compression

We demonstrate a novel method of generating a multiwavelength pulse train by use of time-lens compression. In addition to pulse compression, this time lens simultaneously displaces the pulses according to their center wavelengths, resulting in a temporally evenly spaced multiwavelength pulse train. We further demonstrate a new aberration-correction technique based on the temporal analog of a spatial correction lens to improve the quality of the compressed pulses. Through the use of cw distributed-feedback lasers and electro-optic phase modulators, the all-fiber system allows complete tunability of temporal spacing, spectral profile, and repetition rate.     

基于生物视觉特征和视觉心理学的视频显著性检测算法

提出了一种空域和时域相结合的视频显著性检测算法.对单帧图像,受视觉皮层层次化感知特性和Gestalt视觉心理学的启发,提出了一种层次化的静态显著图检测方法.在底层,通过符合生物视觉特性的特征图像（双对立颜色特征及亮度特征图像）的非线性简化模型来合成特征图像,形成多个候选显著区域;在中层,根据矩阵的最小Frobenius-范数（F-范数）性质选取竞争力最强的候选显著区域作为局部显著区域;在高层,利用Gestalt视觉心理学的核心理论,对在中层得到的局部显著区域进行整合,得到具有整体感知的空域显著图.对序列帧图像,基于运动目标在位置、运动幅度和运动方向一致性的假设,对Lucas-Kanade算法检测出的光流点进行二分类,排除噪声点的干扰,并利用光流点的运动幅度来衡量运动目标运动显著性.最后,基于人类视觉对动态信息与静态信息敏感度的差异提出了一种空域和时域显著图融合的通用模型.实验结果表明,该方法能够抑制视频背景中的噪声并且解决了运动目标稀疏等问题,能够较好地从复杂场景中检测出视频中的显著区域.     

Optical terminal analysis of a multigrating tiled compressor in a PW-class CPA-laser

In the highest-power chirped-pulse amplification lasers, the pulse must be stretched in time, amplified, compressed in a grating compressor and subsequently focused by off-axis parabola to obtain a high peak power. In the optical terminal, the temporal and spatial effects of mismatched multigrating tiled compressor on the far-field pulse are critical factors to be analysed. In this paper, a k-space raytracing model is proposed for the temporal and spatial analyses of possible errors in a four-grating single-pass tiled compressor. The results show that the last grating affects mainly the partial focal spot, while the middle two gratings affect the temporal waveform, and the partial focal spot needs much higher error control than that in the temporal domain in a picosecond pulse compression.     

Dynamic photoelastic analysis by optical heterodyne polarimetry

This paper describes the mapping of the spatiotemporal principal stress distribution evolved with time in an epoxy photoelastic sample. In the optical heterodyne polarimeter exploited, the signal beam of light transmitted by the sample under continuously loaded condition is photomixed with the local oscillator beam of light made up of orthogonal linearly polarized two-frequency components. Every pixel of a MOS video camera used generates a beat photocurrent that possesses the two orthogonal field components of the elliptically polarized signal beam. The spatiotemporal principal stress distributions can be uniquely determined simultaneously and independently from these two orthogonal field components, and are successfully mapped in a time-sequential form. The spatial and temporal resolutions in the maps are 0.18 mm and 2.9 ms, respectively.