单目图像中姿态角估计的坦克目标测距方法

为了克服经典成像测距对相机姿态和视差信息依赖的问题,以观瞄发控装置的电荷耦合器件获取的单目图像为研究对象,提出了基于姿态角估计的坦克目标测距方法.该方法首先利用多级尺度不变特征变换匹配算法实现目标姿态角的快速估计,获取与当前目标姿态偏差最小的模板;然后利用该模板和目标图像模拟立体视觉,建立不依赖于相机姿态和视差的单目测距模型.该模型在远距离测距和小角度偏差的前提下,将模板中的匹配特征点投影到等效平面世界坐标系中,建立被测坦克图像点及其空间点的映射关系,将测距问题转化为对相机外参量的标定,从而求解出相机中心到坦克中心的距离.采用200m到2 500m范围内的坦克验证测距方法的性能,对测量误差随距离的变化规律进行分析,并将其与不同射程时导弹末制导交班的最大容许误差进行对比.结果表明,该方法适用于观瞄发控装置获取的任意姿态图像,实现了野战条件下单兵反坦克武器系统的远距离、快速测距;测量结果中相对误差小于4.9%,绝对误差小于87.7m,满足导弹各个射程的最大容许误差的要求.     

一种基于特征点间线段倾角的姿态测量方法

基于目标特征点间线段倾角信息,提出了一种适合于目标远距离成像和相机内参未知条件下解算目标姿态的目标3维姿态测量方法.采用仿真图像对该方法的正确性进行了验证.实验结果:姿态测量误差绝对值均值小于0.6°,且目标成像尺寸为350pixel时,姿态测量误差绝对值小于0.5°.实验表明该算法具有较高解算准确度和较强的收敛性.     

基于语义目标匹配的三维跟踪注册方法

提出了一种基于语义目标匹配的三维跟踪注册方法。通过改进的单发多框检测(SSD)深度卷积神经网络对图像进行语义分割,获取场景中不同目标的像素级语义分割结果。在求取相机姿态的目标函数时,融合了图像的灰度约束与几何约束对相机的姿态进行估计。所提方法减小了特征点的缺乏或误匹配问题对三维跟踪注册算法性能的影响,且能够适应不同结构的场景。研究结果表明,该方法的误差不超过2.2pixel,基本满足了实时性的要求。     

基于方向性靶标和多约束优化的双目相机标定

提出了一种基于方向性靶标和多约束优化的双目相机标定方法。新型方向性平面靶标能够判断靶标的旋转方向,并对每个标定角点进行编码,以保证双目相机在拍摄到局部靶标的情况下依然能够完成同名点匹配,进而完成双目相机标定。根据方向性靶标建立双目相机标定模型,引入用于描述平面靶标姿态的天顶角和方位角,通过天顶角和方位角筛选出在不同位置上的靶标姿态均具有明显差异的图像作为标定图像,这提高了双目标定结果的稳定性;结合靶标的三维几何信息,建立了多维度约束的双目参数优化模型,提高了双目标定结果的精度。实验结果表明,与传统的张氏标定方法相比,所提出的标定方法能够有效提高所获得的标定结果的稳定性和精度;通过对标准量块进行多次测量,进一步验证了所提方法的有效性。     

移轴组合相机拼接模型及自检校技术研究

研究了非量测型移轴组合相机的图像拼接技术,以双拼移轴相机为研究对象,探讨了其虚拟像空间坐标系建立,并以此为基础建立子影像至虚拟影像的变换模型。提出一种基于初值约束的投影矩阵自检校方法,该方法利用移轴相机检校参数,结合子影像变换模型,推导和计算自检校方程,精确解求虚拟影像拼接参数。设计了室内外地面实验以及航飞实验,对提出的方法进行了验证。实验结果表明,利用该方法进行移轴组合相机虚拟影像拼接,精度优于0.4像素。     

基于消失点检测算法的束靶红外图像畸变校正

传统的霍夫变换、Cannylines直线检测算法、霍夫概率变换方法在图像上的直线检测效果不佳,存在检测线段不连续不正确的问题,因而,利用Sobel滤波对红外图像横轴和纵轴两个方向分别进行锐化,通过线段检测(LSD)算法实现线段特征检测,进而经线段聚类拟合获得图像中完整的直线,通过对直线交点计算获得消失点,最后依据透视关系计算得到校正图像。实验结果表明,该方法可以实现对中性束红外图像的自动有效校正。     

基于垂线包络和平行线对的城市道路图像消失点检测算法

从单幅图像中识别道路消失点,是无人驾驶、智能导航等领域的关键技术之一。针对城市典型道路和街道图像,提出了一种新的基于垂线包络和平行线对的城市道路图像消失点检测算法。提取单幅图像中的所有平行线对,利用图像中存在的竖直线,提出了竖直区域的包络线估计方法,并实现了对道路区域的分割,有效提取了道路区域内的平行线对。使用方向分组策略将道路区域平行线分成两类,并对近似对称的直线对求交点。利用Cmeans聚类和统计相结合的方法估计了城市道路场景中的消失点。实验结果表明,该算法能够准确有效地估计城市道路场景中的消失点。     

基于数字高程模型的航空相机图像清晰度检测方法

为解决弱特征区域中航空相机成像清晰度检测的问题,根据前后两次图像间具有重叠区域的特点,提出了一种基于数字高程模型（DEM）的航空相机图像清晰度检测方法。该方法在引入高精度DEM数据后,以重投影误差最小化原则修正航空成像模型,应用尺度不变特征转换（SIFT）算法进行特征匹配,并利用特征点偏移位置计算主距变化量,最终将主距变化量作为清晰度检测的标准以实现弱特征区域航空图像的清晰度检测。实验表明,所提算法对不同清晰度的弱特征区域航空图像均能进行清晰度检测,其检测均方根误差为16.275μm,小于光学系统的半焦深（19.2μm）,能够满足航空相机的实际工程精度要求。     

基于空间相机时间延迟积分传感器拼接区图像的像移测量

为了对由飞船姿态不稳或隔振装置的振动抑制残余量等因素产生的宽频带像移进行测量,提出了基于时间延迟积分(TDI)传感器拼接区所成图像的像移测量方法。定义了偏移系数,用以衡量成像期间像移速度与TDI传感器积分时间之间的失配度,以调制传递函数(MTF)为相机成像质量评价指标,分析了失配度与成像质量之间的关系,确定了满足成像质量的像移测量方法的精度范围;利用两片TDI传感器拼接区所拍图像存在相同内容、拍摄时刻不同的特点,阐述了基于TDI传感器拼接区图像的像移测量原理。实验结果表明测量精度达到0.2377pixel,测量频带达到228Hz,证明该方法能够对宽频带、高精度的像移进行有效测量。     

低曝光条件下遥感相机微振动量检测

针对搭载平台颤振对遥感相机成像的影响,利用辅助高速面阵CCD相机拍摄高帧频图像序列,通过像元合并来补偿曝光时间不足,增加图像的亮度、对比度和信噪比,同时结合一种区域选择算法来选择用于计算的图像区域,最后使用灰度投影算法对振动位移量进行估计.实验数据表明,提出的改进算法误差为一个像元,在准确性和稳定性方面均明显优于原始算法.     

关于"太阳在银河系中所受引力与到银心距离r的关系"的讨论

讨论了银道面内的引力场强分布和太阳在银河系中所受引力与到银心距离r的关系，指出引力场强g并不是与r的平方成反比，这是由于银河系的大小和形状不能忽略造成的．强调了万有引力定律的适用条件．     

On the predictability of the positions of the convergence zones in the ocean

The disagreement between the experimental and calculated positions of the first convergence zone are known from many publications. The most probable cause for such a disagreement, namely, the incorrect specification of the input data for the calculations, is considered. The lack of simultaneity between the hydrological surveys of the region and the acoustic experiments is emphasized. The experimental data obtained by the author in five ocean regions are presented. These data characterize the diurnal variability of the distance from the source to the nearest boundary of the convergence zone. The relations proposed by different researchers for calculating the sound speed from the temperature, salinity, and hydrostatic pressure are analyzed. It is shown that these relations lead to a substantial difference in the estimated depth dependence of the hydrostatic gradient of the sound speed. The position of the first convergence zone is calculated for the propagation conditions determined by vertical temperature and salinity profiles with the subsequent recalculation of these profiles into sound speed profiles by using eight different formulas known from the literature. It is shown that different formulas lead to different values of the distance to the first zone; this difference is substantially greater than that between the calculations and experiment. The necessity of improving the recalculation relations in view of the experimental data on sound propagation in natural oceanic waveguides, including the data on the actual positions of the convergence zones, is emphasized.     

On the calculation of the bremsstrahlung in the region of the plasma frequency

On the basis of the classical theory of radiation produced in incomplete Coulomb interaction, a formula is derived for the coefficient of the bremsstrahlung of fully ionized plasma. In the region of rectilinear approximation a comparison is made with the results of the quantum formula derived from the Born approximation. Furthermore, the influence of dispersion upon the emission coefficient is taken into account; some relations for the region of dispersion are given.     

Dependence of the Intensity of the Spectral Lines and the Composition of the Sample and the Plasma Parameters of the Unipolar Arc

The investigation was made of the dependence of the intensity of Tl, Ga, Mo, Mg, Mn, Sn, Bi, Ni, Zn, Pt and Au spectral lines and the plasma parameters (temperature T, electron concentration n_e, degree of 6 ionnization α) from concentration of lithium additive.     

Extracting the wavefunction of the LSP at the LHC

We consider associated production of squarks and gluinos with the lightest supersymmetric particle (LSP), or states nearly degenerate in mass with it. Though sub-dominant to pair production of color SU(3)-charged superpartners, these processes are directly sensitive to the wavefunction composition of the lightest neutralinos. Exploiting event-shape variables - including some introduced here for the first time - we are able to identify the composition of the LSP by selecting events involving a single high-pT jet recoiling against missing transverse energy. We illustrate the proposed technique on a set of benchmark cases and propose methods for applying these results in more realistic experimental environments.     

Set the controls for the heart of the Sun

In this paper we describe experiments conducted with high-power lasers that are attempting to replicate, for a very short time and in miniature, conditions found in the Sun. Experiments to date have reached conditions in the outer part of the Sun. To reach the Sun's centre requires compression of material to very much greater than solid density and heating to over ten million degrees. To achieve this, a new class of experiments and a new generation of high-power lasers are required.     

On the Question of the Source of the Apokamp

Technical Physics - The object of this work is the apokamp—a new type of plasma jet, which is formed from a bright offshoot emerging at the bending point of a channel of a high-voltage...