高斯模糊下舰船图像的矩特征  

Moment Features of Ship Image under Gaussian Blur

作　　者：张海[1] 于吉红[1] 

机构地区：[1]海军航空工程学院指挥系,山东烟台264001

出　　处：《海军航空工程学院学报》2015年第5期479-483,共5页Journal of Naval Aeronautical and Astronautical University

摘　　要：针对舰船目标图像受到自身运动、环境和图像预处理的影响出现模糊,丢失轮廓信息,造成像素点阵模糊和矩特征变化的问题,在高斯模糊低阶矩特征分析的基础上,提出并证明了Hu矩、仿射不变矩特征在高斯模糊作用下的变化定理;通过2个仿真实验验证了定理的正确性;为了保证舰船目标的识别效率,提出了建立多尺度特征库的建议。Ship target images appear blurring; lose contour information for the reasons of movement, environment and image preprocessing. All of these result in blurring pixel and moment features changing. Based on the Gaussian blur feature analysis of the lower order moments, Hu moment and affine invariant moment features change theorems were proposed. Through both theoretical analysis and simulation experiments the correctness of these theorems was verified. Finally, the suggestion to establish a multi-scale features database for the recognition rate of ship targets was proposed.

关 键 词：目标识别 矩特征 高斯模糊 多尺度 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]