节点文献
基于改进SURF的遥感图像目标识别
Remote sensing image target recognition based on improved SURF
【摘要】 针对目前遥感图像背景复杂信息量大,导致目标识别过程中特征检测准确率低,特征匹配识别时间长等问题,提出一种基于改进SURF(speeded-up robust features)算法的目标识别方法.该方法采用Grab Cut算法对目标模板进行分割,通过高斯混合模型(Gaussian mixture model,GMM)进行初始化以达到能量最小化分割,并通过快速Hessian矩阵进行特征检测,采用欧式距离完成匹配识别.实验结果表明:使用改进算法进行遥感图像目标识别,能有效去除冗余特征点,提高算法的识别精度和运算速度.
【Abstract】 Aiming at the problems that remote sensing image contains complex background and large amount of information which leads to low feature detection accuracy and long feature matching recognition time in target recognition,a new remote sensing image target recognition method is proposed based on improved SURF(speeded-up robust features)algorithm.Firstly,the Grab Cut algorithm is used to segment the target template,which makes the target template be initialized by Gaussian mixture model(GMM)and the energy minimization segmentation is achieved.Then,in the feature detection phase the fast Hessian matrix is implemented to describe the feature points and the Euclidean distance is used for feature matching.The experimental results show that,the redundant feature points of remote sensing image can be removed validly,the accuracy of recognition and running speed can be improved by using advanced algorithm.
【Key words】 image processing; remote sensing; target recognition; SURF; Grab Cut;
- 【文献出处】 扬州大学学报(自然科学版) ,Journal of Yangzhou University(Natural Science Edition) , 编辑部邮箱 ,2018年03期
- 【分类号】TP751
- 【被引频次】7
- 【下载频次】241