查询结果:   沈学利,陈鑫彤.三元组描述符的特征匹配算法[J].计算机应用与软件,2019,36(2):205 - 210.
中文标题
三元组描述符的特征匹配算法
发表栏目
人工智能与识别
摘要点击数
44
英文标题
FEATURE MATCHING ALGORITHM FOR LEARNED ARRANGEMENTS OF THREE PATCH CODES
作 者
沈学利 陈鑫彤 Shen Xueli Chen Xintong
作者单位
辽宁工程技术大学电子与信息工程学院 辽宁 葫芦岛 125105 辽宁工程技术大学研究生院 辽宁 葫芦岛 125105    
英文单位
School of Electronic and Information Engineering, Liaoning Technical University, Huludao 125105, Liaoning, China Graduate School, Liaoning Technical University, Huludao 125105, Liaoning,China    
关键词
AKAZE/KAZE算法 三元组二进制描述符 非线性金字塔 双线性差值算法
Keywords
AKAZE/KAZE algorithm Learned arrangements of three patch codes Nonlinear pyramid Bilinearity interpolation algorithm
基金项目
国家自然科学基金项目(61772249)
作者资料
沈学利,教授,主研领域:信息安全,图像处理。陈鑫彤,硕士生。 。
文章摘要
针对“风”(KAZE)/加速“风”(Accelerated KAZE)算法鲁棒性差、速度慢等问题,提出三元组描述符(Learned Arrangements of Three Patch Codes)与KAZE/AKAZE算法结合的方法,称作KAZE/AKAZE-LATCH算法。利用AOS算法或者FED算法解非线性方程搭建金字塔;利用海森矩阵在非线性金字塔上寻找特征点;以特征点为圆心按照尺度大小选择相应的采样窗口以建立描述符;利用三元组算法建立二进制描述符。将该算法与KAZE算法和AKAZE算法在公开数据集上对具有模糊变换、光照变换、视角变换和JPEG变换的图像进行匹配时间和匹配正确率的对比实验。经实验表明:该算法的匹配正确率得到巨大提升,匹配速度增加。与现有算法相比,该算法的鲁棒性和实时性更好,可用于对匹配速度和精度要求较高的场景。]
Abstract
Aiming at the problems of poor robustness and slow speed of KAZE/Accelerated KAZE algorithm, a method of combining learned arrangements of three patch codes with KAZE/AKAZE algorithm was proposed, which was called KAZE/AKAZE-LATCH algorithm. We used AOS algorithm or FED algorithm to solve nonlinear equations to build pyramid. The Hessian matrix was used to find the feature points on the nonlinear pyramid, and the corresponding sampling window was selected according to the size of the feature points as the center of the circle to establish the descriptor. A binary descriptor was created by using the three tuple algorithm. The proposed algorithm was compared with KAZE algorithm and AKAZE algorithm in the open dataset for matching time and matching accuracy of images with fuzzy transformation, illumination transformation, view transformation and JPEG transformation. Experiments show that the matching accuracy of the proposed algorithm is greatly improved and the matching speed is increased. Compared with existing algorithms, the proposed algorithm has better robustness and real-time performance, and can be used in scenarios with higher matching speed and accuracy requirements.
下载PDF全文   

根据该篇关键词查找到本刊已发表相关论文供参考
序号
文  章  标  题
作者1
发表栏目
页码
摘要
1
三元组描述符的特征匹配算法
沈学利
人工智能与识别
2019
2
205
[摘要]