Resolution And The Classification Of Sar Target Detection Performance Of Correlation Research

The correlations between SAR image resolution and target detection and classification play an important role in target identification system, since the image resolution affects the accuracy of target detection and classification. So, it is meaningful to carry out the impact of SAR system resources on target recognition and to establish system resource constraint models. The main works are summarized as follows:The background and significance of the image resolution to target detection and classification are described at first. Based on the correlation theory, current situation of the research on target detection and classification at home and abroad are summarized.Next, the general theory and concept of SAR target detection and classification is firstly introduced. It described the entire process of SAR target detection and classification from the system point of view, and the commonly used methods of target detection and classification are concluded. Then, the experimental procedure method used by the target detection and classification is proposed, which focus on feature of the testing target. Finally, the assessment method of target detection and classification which is specific to SAR image is presented based on the assessment method of the ATR system.Then, focused on the influence of the image resolution to target detection. First of all, decreasing resolution methods are introduced and the improved CFAR detector based on the different image resolution is proposed. The theoretical analysis and simulation results show that the detection performance differs with the different image resolution and the different features. The resolution little effect on texture and rank filled ratio features. When used these features, the performance did not getting worse for the different image resolution, for example, low false alarm rate and high detection rate. When used size and mass feature, the target detection rate decreased with the image resolution reduction. The reason is that the size features is related to the number of the pixels of the image, however, the texture features is related to non-uniformity of image. The effect of the SAR image resolution to the target classification is discussed at last. The three target of different resolutions, T-72、BMP-2、BTR, did multiple classification. On different resolutions, the accuracy of the classification is improved with the valid features increase for all targets. However, the accuracy of the classification is decreased with the image resolution reduction on the whole fixed features. When the resolution is higher than0.6mx0.6m the classification accuracy is good in T-72and BTR but bad in T-72and BMP-2, no valid features can be used to classify these targets in the dissertation.