基于非线性相关信息熵的SAR图像多分辨率选择及目标识别 下载: 835次
Multi-Resolution Selection of SAR Images and Target Recognition Based on Nonlinear Correlation Information Entropy
图 & 表
图 1. 多分辨率构造示意图。 (a) 0.3m; (b) 0.4m; (c) 0.5m; (d) 0.6m
Fig. 1. Diagram of multi-resolution representations. (a) 0.3m; (b) 0.4m; (c) 0.5m; (d) 0.6m
下载图片 查看原文
图 2. 基于非线性相关信息熵的SAR图像多分辨率表示选择及目标识别方法
Fig. 2. Selection of multi-resolution representations and target recognition method based on nonlinear correlation information entropy for SAR images
下载图片 查看原文
图 3. 基于PCA的数据降维流程
Fig. 3. Procedure of data reduction based on PCA
下载图片 查看原文
图 4. MSTAR目标的光学图像和SAR图像
Fig. 4. Optical image and SAR image of MSTAR targets
下载图片 查看原文
图 5. 10类目标分类的结果
Fig. 5. Classification result of 10-class targets
下载图片 查看原文
图 6. 噪声干扰下的平均识别率
Fig. 6. Average recognition rates under noise corruption interference
下载图片 查看原文
表 1不同分辨率组合下的NCIE值
Table1. NCIE value under different subsets of resolution
Subset of resolution | NCIE |
---|
(0.3, 0.35, 0.45, 0.5, 0.55) | 0.88 | (0.3, 0.35, 0.4, 0.5) | 0.82 | (0.3, 0.35, 0.45, 0.55) | 0.79 | (0.35, 0.4, 0.45) | 0.71 | (0.3, 0.35, 0.4) | 0.68 |
|
查看原文
表 2标准操作条件下的训练集和测试集
Table2. Training set and test set under standard operating condition
Class | Training set(17°) | Test set(15°) |
---|
Configuration(SN) | Number of samples | | Configuration(SN) | Number of samples |
---|
BMP2 | 9563 | 233 | 95639566C21 | 195196196 | BTR70 | — | 233 | — | 196 | T72 | 132 | 232 | 132812S7 | 196195191 | T62 | — | 299 | — | 273 | BRDM2 | — | 298 | — | 274 | BTR60 | — | 256 | — | 195 | ZSU23/4 | — | 299 | — | 274 | D7 | — | 299 | — | 274 | ZIL131 | — | 299 | — | 274 | 2S1 | — | 299 | — | 274 |
|
查看原文
表 3标准操作条件下的平均识别率
Table3. Average recognition rate under standard operating condition
Method | Proposed method | Multi-resolution 1 | Multi-resolution 2 | Monogenic | CNN |
---|
Average recognition rate/% | 98.92 | 98.17 | 98.43 | 97.81 | 97.68 |
|
查看原文
表 4俯仰角差异下的训练集与测试集
Table4. Training set and test set under different pitch angles
Class | Training set | Test set |
---|
Pitch angle | Number of samples | | Pitch angle | Number of samples |
---|
2S1 | 17° | 299 | 30°45° | 288303 | BRDM2 | 17° | 298 | 30°45° | 287303 | ZSU23/4 | 17° | 299 | 30°45° | 288303 |
|
查看原文
表 5不同俯仰角下的平均识别率
Table5. Average recognition rate under different pitch angles
Method | Average recognition rate/% |
---|
30° | 45° |
---|
Proposed method | 97.86 | 74.17 | Multi-resolution 1 | 96.52 | 71.26 | Multi-resolution 2 | 97.07 | 71.82 | Monogenic | 96.48 | 70.26 | CNN | 95.68 | 65.56 |
|
查看原文
何洁, 陈欣. 基于非线性相关信息熵的SAR图像多分辨率选择及目标识别[J]. 激光与光电子学进展, 2020, 57(22): 221020. Jie He, Xin Chen. Multi-Resolution Selection of SAR Images and Target Recognition Based on Nonlinear Correlation Information Entropy[J]. Laser & Optoelectronics Progress, 2020, 57(22): 221020.