ID 原文 译文
48126 但是,针对不同的合成孔径雷达目标图像,预处理算法的自适应性很难得到保证。 However, in view of the different synthetic aperture radar target image, it is hard to guarantee the adaptability of preprocessing algorithm.
48127 将基于核的主成分分析与稀疏表示相结合, Will be based on a combination of principal component analysis and sparse representation of nuclear,
48128 只需很少的观测数据就能得到高识别率的目标识别结果,节省了数据存储量和计算量。 little observation data can get high recognition rate of target recognition results, save the data storage and computation.
48129 首先,阐述了压缩感知的基本理论; First, this paper expounds the basic theory of compression perception;
48130 其次,提出了基于核主成分分析和稀疏表示的合成孔径雷达图像目标识别算法; Secondly, is proposed based on kernel principal component analysis and sparse representation of synthetic aperture radar images target recognition algorithms;
48131 最后,选取MSTAR数据库中的5类目标进行实验。 Finally, select the five types of targets in MSTAR database experiment.
48132 仿真结果表明,在没有方位角预测的情况下,该算法仍能有效地识别目标,与其他识别算法相比,在同等噪声污染的图像下,具有较高的识别率。 The simulation results show that in the absence of azimuth Angle prediction, the algorithm can effectively identify the target, compared with other identification algorithm, under the image of the same noise pollution, has higher recognition rate.
48133 随着通用图形处理器(general-purpose graphics processing unit,GPGPU)的广泛应用,GPGPU成为当前实现计算并行化的主要硬件平台之一。 As general graphics processor (general - purpose graphics processing unit, GPGPU) the wide application of GPGPU become the realization of one of the main hardware platform of parallel calculation.
48134 开放计算语言(open computing language,OpenCL)是一个开放的、面向异构系统平台的并行计算标准,支持在包括图形处理器(graphics processing unit,GPU)在内的多种微处理器架构上开发和运行并行程序。 Open computing language (open computing language, OpenCL) is an open and heterogeneous system platform oriented parallel computing standard, support in the graphics processor (graphics processing unit, the GPU) on the various types of microprocessor architecture development and run parallel programs.
48135 针对OpenCL平台开发了一套较完整的GPGPU微基准测试程序集,全面测试了GPU的单精浮点运算能力、GPU体系结构中各类存储单元的读写带宽及最佳访问模式等。 For OpenCL platform to develop a set of relatively complete GPGPU micro benchmark set, fully test the single precision floating point operation capacity of GPU, the GPU architecture of all kinds of storage cell, speaking, reading and writing the bandwidth and the best access mode, etc.