Typical Target Detection And Recognition Based On Optical And SAR Remote Sensing Image Fusion

With the continuous development of sensor technology,the utilization of multi-source remote sensing information has become a hot spot in the field of remote sensing technology.Traditional multi-source remote sensing interpretation studies pay more attention to regional scenes because of the limitation of resolution and the frequent change of location of some targets(such as airplane,ships,etc.)over time.In recent years,with the enrichment of remote sensing satellite sensor types,the improvement of detection resolution and the increase of data volume,it is possible to acquire high resolution multi-source remote sensing images of the same region at the same time.Multi-source remote sensing image interpretation has also moved from the regional level to the target level.For this reason,this paper uses optical and SAR high-resolution remote sensing images to carry out the research of multi-source remote sensing fusion of target detection and recognition,based on the classical multi-source feature extraction and fusion method and aiming at the two typical targets of aircraft and ship.For aircraft targets,this paper uses simultaneous optical and SAR data to carry out airport detection,aircraft detection and aircraft recognition respectively,in order to improve the accuracy and speed of aircraft detection and recognition.Aiming at the airport false detection caused by poor imaging quality and cloud occlusion of optical images,this paper proposes a saliency detection model based on multi-source and multi-feature fusion for airport rough detection,and uses the method of pixel classification based on fusion SAR image to realize airport fine extraction,which greatly reduces the range of aircraft targets to be detected and improves the speed of subsequent detection and recognition.In the airport,aiming at the obscured airplane target with weak optical image,this paper uses edge convolution method to locate the suspected airplane target quickly,and further classifies the suspected airplane target by multi-feature classifier to realize the detection of airplane target.Then,for the detected aircraft target,the main direction angle of the aircraft is estimated by the angle of the contour point,and the four types of aircraft are recognized by the method of aircraft target type recognition based on the contour hit rate and uniformity.The experiment proves that the method has a high recognition rate.For ship targets,in order to improve the accuracy and speed of ship detection and recognition,this paper uses simultaneous optical and SAR data to study port detection,ship detection and ship recognition respectively.In order to solve the problems of obscure regional features and more candidate slices of optical image port,this paper adopts a port slice classification method based on frequency-tuned saliency of fusion SAR image to screen out port slices containing ship targets from a large number of port slices.In the slice of port containing ship targets,aiming at the ship misdetection caused by the complex environment of optical image port and the serious shadow interference of ship targets,this paper proposes a joint shape analysis method to obtain suspected ship slices,and further classifies the suspected targets by multi-feature classifier to achieve ship target detection.Then,seven groups of ship position information are detected by using ship position detection method based on feature point matching,contour extraction and brightness saliency.According to the voting result of position information,three types of ship types are recognized.The experiment proves that this method has high recognition rate.