图像弯曲的自适应校正  

An Adaptive Method for Correcting Image Curve

作　　者：徐刚锋[1] 马琳茹[1] 沈振康[1] 

机构地区：[1]国防科学技术大学ATR国防重点实验室,长沙410073

出　　处：《信号处理》2007年第4期626-629,共4页Journal of Signal Processing

摘　　要：由于采取线扫的数据读出方式,红外非制冷探测器在旋转平台上获取的图像存在弯曲现象。本文提出了一种基于图像的自适应弯曲校正方法,根据图像弯曲的机理,建立转速和直线参数的非线性估计模型。为降低计算的复杂性,将非线性模型简化为线性估计模型,最后分析了模型的线性化误差及其测量误差。仿真和真实图像数据的实验验证了模型及算法的正确性。At present, the working of un-cooled infrared sensors is line-scanned for reading charge out. Because of this mechanism,it is occurred to images captured by the un-cooled IR sensor that straight objects become curvy, if the un-cooled IR sensor is installed in rotating system. This paper presents an adaptive method for correcting curve based on images without rotation speed provided by the top equipment. In this paper,we analyze the mechanism of image curve and found non-linear model for estimating rotation speed and parameters of lines. The non-linear model is linearized to reduce computation. At last, model linearization error and measure error are analyzed. The experiments for simulated data and infrared images validate correctness of the model.

关 键 词：图像弯曲校正 非线性参数估计 参数估计误差 

分 类 号：TP278[自动化与计算机技术—检测技术与自动化装置]