ID 原文 译文
49106 首先,利用非局部去噪的思想将结构相似的图像块聚类,每一类图像块单独进行字典学习,增强了字典的自适应性;其次,利用稀疏K-SVD替代传统的K-SVD进行类内字典学习,改善了字典的结构性;最后,引入稀疏系数误差正则项来修正稀疏系数以进一步改善图像的重构效果。 First of all, using the ideas of non-local denoising image block clustering structure similar, each image block independently dictionary learning, enhance the adaptability of the dictionary;Secondly, the use of thin K - K - SVD of SVD to replace the traditional dictionary within class learning, improve the structural the dictionary;Finally, the sparse regularization coefficient error are introduced to correct the sparse coefficient in order to further improve the image reconstruction effect.
49107 实验结果表明,与传统的K-SVD算法相比,该算法能够有效地保持图像的结构信息,并且提升了去噪效果,同时,在不降低图像结构相似度的基础上,峰值信噪比很接近甚至部分好于目前先进的去噪算法。 The experimental results show that compared with the traditional K - SVD algorithm, this algorithm can effectively maintain the structure of the image information, and enhance the denoising effect, at the same time, without reducing the image structure on the basis of the similarity of the peak signal-to-noise ratio part is very close to or even better than the current advanced denoising algorithm.
49108 为了满足车载武器的捷联惯导系统(strapdown inertial navigation system,SINS)动基座初始对准中对辅助信息自主性的要求,提出了基于里程计辅助的SINS动基座初始对准; In order to satisfy the vehicle-mounted weapons of the strapdown inertial navigation system (strapdown inertial navigation system of SINS) in the moving base initial alignment of the requirements of the auxiliary information autonomy, is proposed based on odometer aided SINS moving base initial alignment;
49109 针对车载SINS在行驶过程中,地面的不平坦、阵风以及人为操作等因素引起的载体姿态变化,文中利用姿态更新来实时记录载体在各种因素干扰下的姿态变化,结合初始姿态的最优估计值得到载体导航前的姿态,真正地实现了导航前一刻的初始对准。 For vehicular SINS in the process of driving, the ground is not flat, the wind and artificial factors such as the operation of carrier attitude change, in this paper, we use gesture updates to real time record carrier attitude changes under various interference factors, combined with the attitude of the initial optimal estimate to get in front of the vehicle navigation attitude, really realize the moment of initial alignment before navigation.
49110 仿真结果表明,基于里程计辅助的SINS动基座初始对准方法对准精度高,收敛速度快,能够实现车载武器系统的动基座对准要求。 Simulation results show that based on odometer aided SINS on moving base initial alignment method of high precision, fast convergence speed, can realize vehicle weapons systems moving base alignment requirements.
49111 针对无源观测平台量测数据高度的非线性特征和各种随机噪声对滤波器估计精度所造成的严重影响,提出了一种对量测数据进行预处理的算法。 Passive observation platform for measurement data is highly nonlinear characteristics and all kinds of random noise estimation precision caused by serious impact to filter, this paper proposes a algorithm for the measurement data preprocessing.
49112 首先,利用量测数据的置信度对量测数据进行滑窗式检测,初步判别无效数据,然后对支持向量机(support vector machine,SVM)的核函数进行改进,利用数据自身的特征信息对核函数进行加权处理,再用改进的SVM对数据进行再分类处理,从而最终确定无效数据。 First, use the test data of confidence to sliding window detection of measured data, the preliminary judging invalid data, then the SVM (support vector machine, SVM) of the kernel function was improved, using the data of characteristic information to the weighted kernel function, and with improved SVM to classify the data again and thus ultimately determine the invalid data.
49113 利用所提算法对实测数据进行预处理,结果说明所提算方法能有效剔除各类噪声对测量数据的影响,具有工程应用价值。 Proposed algorithm is used to analyse the measured data pretreatment, the results show that the proposed method can effectively eliminate various types of noise on the measurement data, the influence of engineering application value.
49114 综合应用冲激滤波器、改进的前向后向(forward and backward,FAB)扩散滤波器和全变差(total-variation,TV)扩散算法,提出一种双正交映射约束的混合偏微分方程扩散图像放大算法。 Before the integrated application of impulse filter, an improved back to (forward and backward, FAB) diffusion filter and total variation (total variation, TV) diffusion algorithm, and proposes a biorthogonal mapping constraints of mixed partial differential equation of diffusion algorithm of image magnification.
49115 改进的FAB滤波器能够很好地增强图像的小边缘,参数约束的冲激滤波器可增强图像的强边缘,基于水平集方法实现的TV扩散可以消除边缘的锯齿波,使边缘光滑,所提算法综合以上优点。 Improvement of the overall filter can effectively enhance image edges of small, impulse filter parameter constraints can enhance the image edge, the TV diffusion based on level set method can eliminate the edge of the sawtooth wave, make the edge smooth, the proposed algorithm integrated the above advantages.