Research On Ship Identification Technology Based On 3D Modeling

With the continuous improvement of the Navy's role in the modern warfare and the continuous improvement of all kinds of advanced combat ships,ship target recognition based on sea background has become an important research topic in the field of automatic target recognition.At present,there are two ways of radar guidance and image guidance.Compared with radar guidance,the image guidance power is small,the detection distance is long,and the recognition accuracy is high.In this paper,we study the image recognition algorithm of 3D ship target.The main results of this paper are as follows:(1)A target segmentation algorithm based on adaptive threshold for region growing,morphological filtering and geometric processing is proposed.The target segmentation image is obtained by the adaptive selection of the growth threshold,and the subsequent processing is carried out by combining morphological filtering and geometric processing.Experiments show that the method is suitable for most of the images of the target.(2)A 3D object SIFT(Scale Invariant Feature Transform)feature extraction algorithm based on AP clustering feature points is proposed.According to the SIFT,the characteristics of a good translation,rotation and scale invariance,the traditional image library image into SIFT feature vector.The image database is clustered by AP(Affinity Propagation)clustering algorithm,and a large number of duplicate images are removed.The experimental results show that the recognition rate of the target ship type is accelerated in the condition of ensuring accuracy.(3)An improved 3D object recognition algorithm based on ASIFT is proposed.Based on the SIFT feature three invariance,we introduce and improve the ASIFT algorithm by increasing the two parameters which is longitude and the latitude angle of the ship targe to simulate the change of the angle of view,realizing large-scale of the affain invariance.Experimental results show that the SIFT feature vector of the image is increased by increasing the number of points on the target database,which can improve the recognition rate of the target.