Abstract
Feature extraction of targets radiated-noise and design of targets classifier are key issues of passive sonar target recognition system. A new feature extraction method of continuous spectrum was propose, and then a new adaptive genetic-backpropagation algorithm was proposed for training neural network target classifier. At last, the classification experiment for three different classes of targets was done. The results of experiment show that the passive sonar target recognition system designed in the paper has higher correct classification rate.
| Original language | English |
|---|---|
| Pages (from-to) | 382-386 |
| Number of pages | 5 |
| Journal | Shanghai Jiaotong Daxue Xuebao/Journal of Shanghai Jiaotong University |
| Volume | 36 |
| Issue number | 3 |
| Publication status | Published - Mar 2002 |
| Externally published | Yes |
Keywords
- Adaptive genetic-back propagation algorithm
- Continuous spectrum
- Feature extraction
- Passive sonar target recognition