Hybrid multi-resolution detection of moving targets in infrared imagery

A hybrid moving target detection approach in multi-resolution framework for thermal infrared imagery is presented. Background subtraction and optical flow methods are widely used to detect moving targets. However, each method has some pros and cons which limits the performance. Conventional background subtraction is affected by dynamic noise and partial extraction of targets. Fast independent component analysis based background subtraction is efficient for target detection in infrared image sequences; however the noise increases for small targets. Well known motion detection method is optical flow. Still the method produces partial detection for low textured images and also computationally expensive due to gradient calculation for each pixel location. The synergistic approach of conventional background subtraction, fast independent component analysis and optical flow methods at different resolutions provide promising detection of targets with reduced time complexity. The dynamic background noise is compensated by the background update. The methodology is validated with benchmark infrared image datasets as well as experimentally generated infrared image sequences of moving targets in the field under various conditions of varying illumination, ambience temperature and the distance of the target from the sensor location. The significant value of F-measure validates the efficiency of the proposed methodology with high confidence of detection and low false alarms.     

时空域双边滤波背景抑制算法研究

红外图像中的强边缘一直是制约红外弱小目标检测概率的重要因素,同时是产生目标虚警的重要来源。采用能够同时利用邻域空间欧式距离和灰度值相似性的双边滤波,并针对天基红外系统中帧间背景缓慢变化的特点,设计一种基于时空域的双边滤波法,实验验证该方法对复杂背景的强边缘具有很好地抑制效果,信杂比增益大于3.6,同时能够很好地保留目标能量,目标能量损失因子D factor小于0.2,有效地提高目标的局域信杂比。     

基于空时联合稀疏重构的红外小弱运动目标检测方法

提出了一种基于空时联合稀疏重构的红外小弱运动目标检测算法。通过学习序列图像内容而构建的空时联合字典能同时刻画目标或背景的形态特征和运动信息;利用多元高斯运动模式从空时联合字典中提取出目标空时字典和背景空时字典,目标空时过完备字典描述移动的目标,背景空时过完备字典表征背景噪声。将连续多帧图像在空时联合字典上进行稀疏分解,然后分别利用目标空时字典和背景空时字典中的最大稀疏系数及其空时原子重构信号,获取重构残余能量差异来区分目标和背景。试验结果表明,由同源的空时字典重构的残余能量小,而由异构的空时字典恢复的残余能量大,该方法不仅能提高序列信号表示的稀疏度,还能有效提高小运动目标的探测能力。     

基于小波变换的红外热波无损检测融合算法

在红外热波无损检测中获取的热像序列存在着背景噪声大、缺陷边缘模糊、对比度低等特点。为了提高由红外热像序列重构的数字图像的缺陷显示能力,以小波变换为热像处理工具,采用基于像素级和特征级的图像融合算法对热像序列进行了处理,并采用基于统计学的图像评估标准对处理效果作了定量评价。通过对铝合金试件的检测实验说明该方法可用于材料内部缺陷的红外热波无损检测。研究结果表明,此种图像融合算法可对不同深度缺陷所对应的两幅最佳热像进行有效地融合,在一幅融合图像中直观地反映出全部缺陷,并能有效地减少加热不均和背景噪声对缺陷识别的不利影响。     

具有紧支撑正交非张量积小波的图像融合

提出了基于一种新的小波——具有紧支撑、正交性、伸缩矩阵为[^2 0 ^0 2]的非张量积小波的图像融合方法。首先根据非张量积小波理论,利用Daubechies构造的单变量滤波器构造出基于四通道的不可分的小波滤波器组,用此滤波器组对参加融合的图像进行分解,然后对低频部分采用取均值、高频部分采用系数绝对值取大的融合算法对分解子图进行融合,最后重构。并采用熵、交叉熵、互信息、均方根误差和峰值信噪比等指标对该方法进行了客观评价。对可见光图像与红外图像、远红外图像与近红外图像、遥感图像、多聚焦图像和其它多类图像的融合实验结果证明本方法有较好的融合效果,其融合性能与采用同样融合算法的张量积db2小波的融合方法的融合性能相当。     

New enhancement of infrared image based on human visual system

Infrared images are firstly analyzed using the multifractal theory so that the singularity of each pixel can be extracted from the images. The multifractal spectrum is then estimated, which can reflect overall characteristic of an infrared image. Thus the edge and texture of an infrared image can be accurately extracted based on the singularity of each pixel and the multifractal spectrum. Finally the edge pixels are classified and enhanced in accordance with the sensitivity of human visual system to the edge profile of an infrared image. The experimental results obtained by this approach are compared with those obtained by other methods. It is found that the proposed approach can be used to highlight the edge area of an infrared image to make an infrared image more suitable for observation by human eyes.     

Wavelet-modified binary phase-only morphological correlation for color pattern recognition

This paper reports a morphological phase-only correlation technique based on bit-map representation for recognition of color as well as grey images in a hybrid digital-optical correlation architecture. The color image is decomposed into its R, G and B components, and each component is further decomposed into eight disjoint elementary images depending upon the bit-map representation of the color value at each pixel. Bit-map representation of the pixel values of an image reduces the required computational time. A set of twenty-four disjoint wavelet-modified binary phase-only filters (WBPOFs) are generated from these bit-map decomposed images. The target image is similarly decomposed into eight disjoint images each of R, G and B and their digital Fourier transforms multiplied with the corresponding WBPOFs. The product functions thus obtained are added up to form a single resultant product function, whose optical Fourier transformation gives the correlation peaks for the presence of R, G and B components in the image. The single product function overcomes the necessity of obtaining the final optical Fourier transformation of the R, G and B components separately. The novelty of this approach lies in the fact that the WBPOFs synthesized by this procedure are thus able to identify both colored as well as gray images and can tolerate salt-and-pepper noise to a considerable extent.     

Stripe noise removal for infrared image by minimizing difference between columns

This paper introduces a novel approach to eliminate stripe noise in infrared images. The differences between bias voltages in column readout circuit of uncooled infrared sensors result in strong stripe noise which changes slowly in time. The problem can be solved by estimating the bias of each column of infrared images and correcting infrared images with the estimated biases. The bias estimation is translated into an energy optimization problem in the paper. The optimization aims to minimize difference between neighboring columns of images. Our approach can be processed on a single image, or in a recursive way in order to significantly reduce the computation in one frame time. Our approach is compared to the state-of-the-art the stripe noise removal method using realistic infrared images, and the experimental results show the effectiveness and efficiencies of our proposed approach.     

Lateralization of dichotic noise bursts: effect of onset and offset disparity

Sound image position associated with the interaural onset or offset disparity of a signal was quantified by a scaling procedure in three experiments. Lateralization cues derived from the steady-state portion of the broadband noise signal that would support a specific image position were minimized by the use of independent noise sources for each ear. Onset disparities produced lateralization toward the ear at which the sound was presented first, while offset disparity produced lateralization toward the ear at which the sound remained on longer. Disparity was systematically varied between 0 and 10 ms and for a given disparity, a greater shift in the sound image position was obtained when the disparity was at the onset rather than the offset. The duration of the shorter signal ranged from 2.5-100 ms and for either onset or offset disparity, the image of stimuli of long duration tended to remain near the center of the head, while those of shorter duration could be moved to more extreme positions. In an attempt to rule out dichotic loudness cues as a basis for the lateralization associated with offset disparity, stimuli were presented with equal energy at each ear. Image position for equal energy was virtually identical to that for equal sound pressure, suggesting that loudness differences are not mediating lateralization associated with offset disparity.     

大功率光纤激光焊焊缝跟踪偏差红外检测方法

精确控制激光束使其始终对准并跟踪焊缝是保证激光焊接质量的前提.针对大功率(激光功率10 kW)光纤激光焊接304型不锈钢紧密对接焊缝(间隙为0-0.1 mm),研究一种基于红外热像的焊缝跟踪偏差检测新方法. 采用红外传感高速摄像机摄取焊接区域熔池红外动态热像,分析激光束对准和偏离焊缝中心时的熔池温度分布和红外辐射特性,以熔池匙孔形变参数和热堆积效应参数作为激光束与焊缝中心偏差检测特征值,通过图像识别技术研究和分析特征值与焊缝偏差之间的关系. 激光焊接试验结果表明,熔池匙孔形变参数和热堆积效应参数与焊缝偏差 关键词： 大功率光纤激光焊 焊缝跟踪偏差 红外热像 检测     

The effects of the addition of low-level, low-noise noise on the intelligibility of sentences processed to remove temporal envelope information

The intelligibility of sentences processed to remove temporal envelope information, as far as possible, was assessed. Sentences were filtered into N analysis channels, and each channel signal was divided by its Hilbert envelope to remove envelope information but leave temporal fine structure (TFS) intact. Channel signals were combined to give TFS speech. The effect of adding low-level low-noise noise (LNN) to each channel signal before processing was assessed. The addition of LNN reduced the amplification of low-level signal portions that contained large excursions in instantaneous frequency, and improved the intelligibility of simple TFS speech sentences, but not more complex sentences. It also reduced the time needed to reach a stable level of performance. The recovery of envelope cues by peripheral auditory filtering was investigated by measuring the intelligibility of 'recovered-envelope speech', formed by filtering TFS speech with an array of simulated auditory filters, and using the envelopes at the output of these filters to modulate sinusoids with frequencies equal to the filter center frequencies (i.e., tone vocoding). The intelligibility of TFS speech and recovered-envelope speech fell as N increased, although TFS speech was still highly intelligible for values of N for which the intelligibility of recovered-envelope speech was low.     

Bayesian analysis of in vivo dynamic 13C-edited 1H images

We propose an application of a Bayesian methodology to dynamic MR images of protons J-coupled to 13C nuclei for monitoring the in vivo 13C-glucose uptake of mouse brain. The very low population of these protons and the random noise make the analysis of these images extremely difficult. The proposed method restores the images and provides an "activation" map of the mouse brain by means of a hypothesis testing procedure. The restoration step is performed in the Bayesian framework so that among the other advantages of a stochastic approach, it is possible to model spatial and temporal information about neighboring pixels. This leads to a restoration procedure able to reduce the noise level while preserving the information about the edges of signal areas. Based on the restored images, the testing procedure provides us with a reliable map of pixels characterized by the 13C-glucose uptake.     

Multi-sampling and filtering technology of IRFPA

In order to overcome the drawbacks of low SNR of IR (infrared) images and resolve the problem that widely used 2D-TDI (two dimensional time delay integration) algorithm could only be applied to static or slow-moving targets, the application of multi-sampling and filtering technology is discussed in this paper. The technology adopted the method of sampling the same pixel repeatedly in one image which ensured that it could be applied to fast-moving targets and lowered random noise of IR images in the meantime. Both the images effects and the experimental data show that multi-sampling and filtering technology of infrared focal plane array (IRFPA) have the same ability of improving SNR (signal noise ratio) of IR images as the 2D-TDI algorithm which can raise the SNR of IR images to times of the original ones (m equals to the sampling times at the same pixel), and the experimental result also proves that the technology can be applied to all moving targets.     

Statistical treatment applied to infrared thermoelastic analysis of applied and residual mechanical stresses

Very small temperature variations are quantified by a statistical treatment of standard infrared equipment images. This procedure determines the signal amplitude value, comparing the noise and the noisy signal dispersions characterized by their variances. This robust and simple method has the advantage of needing no link to the reference signal and of treating any kind of signal shape. It is applied here to thermoelastic analysis of applied and residual stresses.     

Wavelet domain de-noising of time-courses in MR image sequences

Magnetic resonance images acquired with high temporal resolution often exhibit large noise artifacts, which arise from physiological sources as well as from the acquisition hardware. These artifacts can be detrimental to the quality and interpretation of the time-course data in functional MRI studies. A class of wavelet-domain de-noising algorithms estimates the underlying, noise-free signal by thresholding (or 'shrinking') the wavelet coefficients, assuming the underlying temporal noise of each pixel is uncorrelated and Gaussian. A Wiener-type shrinkage algorithm is developed in this paper, for de-noising either complex- or magnitude-valued image data sequences. Using the de-correlation properties of the wavelet transform, as elucidated by Johnstone and Silverman, the assumption of i.i.d. Gaussian noise can be abandoned, opening up the possibility of removing colored noise. Both wavelet- and wavelet-packet based algorithms are developed, and the Wiener method is compared to the traditional Hard and Soft wavelet thresholding methods of Donoho and Johnstone. The methods are applied to two types of data sets. In the first, an artificial set of complex-valued images was constructed, in which each pixel has a simulated bimodal time-course. Gaussian noise was added to each of the real and imaginary channels, and the noise removed from the complex image sequence as well as the magnitude image sequence (where the noise is Rician). The bias and variance between the original and restored paradigms was estimated for each method. It was found that the Wiener method gives better balance in bias and variance than either Hard or Soft methods. Furthermore, de-noising magnitude data provides comparable accuracy of the restored images to that obtained from de-noising complex data. In the second data set, an actual in vivo complex image sequence containing unknown physiological and instrumental noise was used. The same bimodal paradigm as in the first data set was added to pixels in a small localized region of interest. For the paradigm investigated here, the smooth Daubechies wavelets provide better de-noising characteristics than the discontinuous Haar wavelets. Also, it was found that wavelet packet de-noising offers no significant improvement over the computationally more efficient wavelet de-noising methods. For the in vivo data, it is desirable that the groups of "activated" time-courses are homogeneous. It was found that the internal homogeneity of the group of time-courses increases when de-noising is applied. This suggests using de-noising as a pre-processing tool for both exploratory and inferential data analysis methods in fMRI.     

基于改进的空域相关的多聚焦图像融合

提出了一种简单有效的像素级多聚焦图像融合方法。针对正交小波变换缺乏平移不变性而产生视觉失真的缺陷,采用Atrous算法将原图像分解在不同频率域上。Atrous算法先将滤波器h0(n),h1(n)各点间插入适当的零值后再与低频信号做卷积,故又称为"多孔算法"。将具有抑制噪声性能的空阈相关法作为高频子图像的融合规则,选取相关性强边缘特征显著的点作为最终融合子图像的像素点。实验表明,由此融合的图像能完好的保留边缘纹理信息。融合后的图像在客观评价和主管视觉效果上均有提高。     

New Phase-Shifting Method Using Two Images

We propose a continuous two-step method (CTSM) to obtain an image from two interference images for the measurement time reduction of optical coherence tomography. Image processing is performed for each pixel in CTSM. The calculated signal intensity has an average and standard deviation after processing since the heterodyne beat signal contains the noise in the measurements. We assume that the heterodyne beat signal contains the random noise in simulations and calculate the influence of the normalized standard deviation (NSD = standard deviation/average) on the random noise intensity for both CTSM and the conventional three-step method (TSM). We found that the NSD of the CTSM is only 1.4 times greater than that of the TSM despite processing only two images.     

Performance analysis of dispersion compensating optical FIR filters designed using a quasi-analytic synthesis approach

In this paper we investigate the performance of a quasi-analytic synthesis approach for dispersion compensating optical finite impulse response filters. It is shown by eye opening penalty and optical signal to noise ratio penalty calculations that with this method the filters dispersion can be adjusted reliably and with low effort to compensate for a specific amount of residual chromatic dispersion in optical transmission systems. The maximum attainable dispersion is then only limited by the filter order and the free spectral range rather then by optimization problems within the filter design.     

The effect of temporal placement on gap detectability

The detectability of a masked sinusoid increases as its onset approaches the temporal center of a masker. This study was designed to determine whether a similar change in detectability would occur for a silent gap as it was parametrically displaced from the onset of a noise burst. Gap thresholds were obtained for 13 subjects who completed five replications of each condition in 3 to 13 days. Six subjects were inexperienced listeners who ranged in age from 18 to 25 years; seven subjects were highly experienced and ranged in age from 20 to 78 years. The gaps were placed in 150-ms, 6-kHz, low-passed noise bursts presented at an overall level of 75 dB SPL; the bursts were digitally shaped at onset and offset with 10-ms cosine-squared rise-fall envelopes. The gated noise bursts were presented in a continuous, unfiltered, white noise floor attenuated to an overall level of 45 dB SPL. Gap onsets were parametrically delayed from the onset of the noise burst (defined as the first nonzero point on the waveform envelope) by 10, 11, 13, 15, 20, 40, 60, 110, 120, and 130 ms. Results of ANOVAs indicated that the mean gap thresholds were longer when the gaps were proximal to signal onset or offset and shorter when the gaps approached the temporal center of the noise burst. Also, the thresholds of the younger, highly experienced subjects were significantly shorter than those of the younger, inexperienced subjects, especially at placements close to signal onset or offset. The effect of replication (short-term practice) was not significant nor was the interaction between gap placement and replication. Post hoc comparisons indicated that the effect of gap placement resulted from significant decreases in gap detectability when the gap was placed close to stimulus onset and offset.     

一种基于时域滤波的红外序列图像去噪算法

红外图像是现代光学设备常用的图像源,图像显示效果直接影响设备的用户体验,而红外序列图像中的噪声会导致显示效果的下降。为了减轻噪声对红外序列图像显示效果的影响,通过历史多帧灰度值的加权和当前帧无噪声图像的估计,提出了一种基于时域高斯滤波的去噪方法。参考空域双边滤波的权值分配方法,引入了灰度值的影响对时域高斯滤波的权值进行修正,解决时域滤波导致的序列图像中运动目标拖尾和模糊。实验结果表明,时域滤波方法能够有效平滑帧间噪声,减轻噪声导致的红外序列图像显示效果的恶化,引入灰度值的影响进行滤波权值修正之后,能够解决时域滤波导致的运动目标拖尾和模糊问题。