Author Affiliations
School of Information Engineering (College of Big Data), Xuzhou University of Technology, Xuzhou, Jiangsu 221018, Chinashow less
Fig. 1. Flow chart of multi-resolution representation algorithm
Fig. 2. SAR image targets under different resolutions. (a) Original image; (b) 0.4m; (c) 0.5m; (d) 0.6m
Fig. 3. Flow chart of obtaining label distribution
Fig. 4. Convergence curve of network training process
Fig. 5. Feature maps of proposed network output. (a) Input image; (b) first convolutional layer
Fig. 6. Flow chart of SAR image target recognition process combining multi-resolution representation and complex domain CNN
Fig. 7. Schematic of target to be identified. (a) BMP2; (b) BRT70; (c) T72; (d) T62; (e) BRDM2; (f) BTR60; (g) ZSU23/4; (h) D7; (i) ZIL131; (j) 2S1
Fig. 8. Identification results of 10 categories of targets under standard operating conditions
Fig. 9. Comparison curves of different methods under random noise identification problem
Class | Training set(depression angle 17°) | Test set(depression angle 15°) |
---|
BMP2 | 233 | 195 | BTR70 | 233 | 196 | T72 | 232 | 196 | T62 | 299 | 273 | BRDM2 | 298 | 274 | BTR60 | 256 | 195 | ZSU23/4 | 299 | 274 | D7 | 299 | 274 | ZIL131 | 299 | 274 | 2S1 | 299 | 274 |
|
Table 1. Number of images for training and test samples under standard operating conditions
Method | Average recognition rate/% |
---|
Proposed method | 99.42 | MR 1 | 98.78 | MR 2 | 99.02 | CNN | 99.08 | CCNN | 99.16 |
|
Table 2. Average recognition rates of different methods under standard operating conditions
Class | Training set(depression angle 17°) | Test set(depression angle 15°) |
---|
BMP2 | 233 (Sn_9563) | 196 (Sn_9566)196 (Sn_c21) | BTR70 | 233 (Sn_c71) | 196 (Sn_c71) | T72 | 232 (Sn_132) | 195 (Sn_812)191 (Sn_s7) |
|
Table 3. Number of images for training and test samples in model identification problems
Method | Average recognition rate /% |
---|
Proposed method | 98.92 | MR 1 | 97.64 | MR 2 | 98.08 | CNN | 97.26 | CCNN | 98.23 |
|
Table 4. Average recognition rates of different methods under model recognition problem
Sample | Depressionangle | 2S1 | BDRM2 | ZSU23/4 |
---|
Training set | 17° | 299 | 298 | 299 | Test set | 30° | 288 | 287 | 288 | 45° | 303 | 303 | 303 |
|
Table 5. Number of images for training and test samples in pitch angle recognition problem
Method | Average recognition rate/% |
---|
30° | 45° |
---|
Proposed method | 98.56 | 73.62 | MR 1 | 97.54 | 69.56 | MR 2 | 97.82 | 71.08 | CNN | 97.43 | 67.92 | CCNN | 98.02 | 72.02 |
|
Table 6. Average recognition rates of different methods under pitch angle recognition problem