| SAR(Synthetic Aperture Radar) has been used in many fields, owing to its all-day work ability, all-weather work ability and a certain amount of penetration. And these advantages also make SAR more and more important in field of national security and defense. Along with the maturity of the SAR imaging technology, research of the airborne SAR and the arising of the high-resolution SAR, the research of SAR interpretation technique become intensified. And mature imaging technology makes the target classification and recognition in SAR images possible. As a kind of important target, detecting and recognizing the aircraft targets from SAR images have the actual demand. This paper will research the technologies of aircraft detection and recognition in high-resolution airborne SAR images by referencing relevant researches:(1) Speckle noise suppression algorithms. The special imaging procedure of SAR generates the speckle noises, which arethe interferences in procedures of SAR target detection and recognition. So this paper studies the relevant speckle noise suppression algorithms firstly, and then combined with SAR image features, this paper proposes a adaptive integrated filtering algorithm based on wavelet transform, and the comprehensively objective and subjective evaluation experiments are selected to confirm the effectiveness of this algorithm.(2) Apron and runway automatic detection algorithm. This paper’s purpose is to extract the aircrafts which are parking on the apron and runway, so the apron and runway become very important contextual information in detection procedure. By considering the premise of that the aircrafts always park on the apron and runway of the airport, this paper improves a simple algorithm based on labling the lowest reflection connected region, by analyzing the different scattering features of the aircraft and apron and runway. And the experimental results demonstrate this algorithm.(3) Aircraft detection technology in high-resolution SAR images. In order to solve the problems of that current detection algorithms cannot obtain intactly connected targetswhile remaining well structural features, this paper proposes a new detection algorithm based on gradient textural saliency by analyzing SAR images’ characteristic. This algorithm can meet real-time requirement of the project implementing because of its simple computation. And the designed experiments demonstrate this proposed algorithm.(4) Aircraft target discrimination algorithm in high-resolution SAR images. The detection results will still remain many false alarms while ensuring high detection rate, so the detected suspicious targets need further identification. After analyzing relevant manmade target discrimination technologies in SAR images, this paper proposes a new discrimination method based on global area feature and local extended fractal structural features by combining aircraft features in SAR images. The experimental results demonstrate the well discrimination performance of this proposed algorithm. |
