基于小波变换和深层稀疏编码的SAR目标识别

发布时间：2018-01-15 06:25

本文关键词：基于小波变换和深层稀疏编码的SAR目标识别　出处：《电视技术》2014年13期 　论文类型：期刊论文

【摘要】：针对SAR图像预处理算法自适应能力差、带标签图像不足、目标特征提取困难等问题,提出了一种基于小波变换和深层稀疏编码的SAR图像目标自动识别算法。首先利用灰度值和尺度缩放获得大量的无标签SAR目标,并采用离散小波变换对图像进行高效的降维,再结合深层稀疏编码提取目标的深层抽象特征并完成识别任务。采用MSTAR数据库中3类军事目标进行算法仿真与验证。实验结果表明,在没有预处理的情况下,该算法能够有效地完成多目标SAR图像分类,且具有较高的识别率和鲁棒性。
[Abstract]:In order to solve the problems such as poor adaptive ability of SAR image preprocessing algorithm, insufficient tagged image and difficult target feature extraction, etc. An automatic target recognition algorithm for SAR images based on wavelet transform and deep sparse coding is proposed. Firstly, a large number of untagged SAR targets are obtained by using gray value and scale scaling. And the discrete wavelet transform is used to reduce the dimension of the image efficiently. Combined with deep sparse coding to extract the deep abstract features of the target and complete the task of recognition. The algorithm is simulated and verified by using three kinds of military targets in MSTAR database. The experimental results show that. Without preprocessing, the algorithm can effectively achieve multi-target SAR image classification, and has high recognition rate and robustness.
【作者单位】：空军工程大学航空航天工程学院;甘肃省张掖市山丹县95876部队;
【基金】：国家自然科学基金项目(61372167;61379104) 航空科学基金项目(20115896022)
【分类号】：TN957.52
【正文快照】： 合成孔径雷达(Synthetic Aperture Radar,SAR)由于其全天候作战的特点而迅速发展起来。SAR图像作为一种新型的遥感图像具有特殊的乘性斑点噪声及丰富的纹理信息等特点。通常情况下,SAR图像目标自动识别算法在识别前必须要进行图像预处理,要求在有效抑制相干斑噪声的同时较好地

【相似文献】