基于像素空间相关度的微光双谱图像实时配准  

Dual-band Low Level Light Image Real-time Registration Based on Pixel Spatial Correlation Degree

作　　者：张毅[1] 柏连发[1] 陈钱[1] 顾国华[1] 

机构地区：[1]南京理工大学近程高速目标探测技术国防重点学科实验室,江苏南京210094

出　　处：《南京理工大学学报》2009年第4期506-510,共5页Journal of Nanjing University of Science and Technology

基　　金：国家自然科学基金(60572107)

摘　　要：针对目前不同光谱波段图像配准算法复杂、精度低和实时性差的问题,该文提出了基于像素空间相关度的双通道图像融合效果评价方法,将该指标用作实时图像配准的控制参数,在基于DM642的视频图像实时处理仿真平台上,对微光双谱图像进行了实时配准。实验结果表明:与常用的基于控制点、边缘等图像配准算法相比,该文提出的实时配准方法准确、有效,且不受图像灰度变化、噪声及不同微光成像波段成像器的光谱响应特征的影响。In view of the problem of the complexity of different spectrum image registration arithmetics, low registration precision and bad real-time property, an evaluation method of the dual-channel image fusion effect based on pixel spatial correlation degree is proposed. This index is used as the control parameters of real-time image registration. Dual band real-time image registration is carried out based on real-time video image processing emulational platform based on DM642. The experimental results show that, compared with the commonly used algorithm based on the control point or the edge, the image registration algorithm proposed here is more accurate and effective, and it is not influenced by image gray scale change, noise and spectral response characteristics of different band low level light imaging device.

关 键 词：图像配准 彩色夜视 像素空间相关性 灰度共生矩阵 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]