基于梯度天空背景的非均匀性校正和点目标探测

发布时间：2018-04-11 18:51

本文选题：信息处理技术 + 梯度天空背景　；参考：《吉林大学学报(工学版)》2017年05期

【摘要】：针对黑体两点校正方法不适用于校正梯度天空背景的问题,提出一种新的基于梯度天空背景的两点校正方法。首先,采用邻域像素替代法剔除位置固定不变的盲元;然后,利用红外焦平面探测器拍得的两幅不同俯仰角度的纯净天空图像作为两个温度点,进行两点校正获得增益系数和偏置系数;最后,通过修正这两个系数剔除少量随机噪点。实验结果验证了该方法的有效性。本文在校正基础上还提出了一种小窗口双边滤波和梯度模板联合检测算法,用于探测暗弱点目标。首先采用小窗口的双边滤波去除高斯噪点,然后与梯度模板做卷积,自适应阈值分割检测出暗弱点目标。实验结果表明,提出的模板算法对复杂背景的抑制作用较强,可有效提高目标的信噪比,而且算法复杂度不高,易于实时实现。
[Abstract]:Aiming at the problem that blackbody two-point correction method is not suitable for correcting gradient sky background, a new two-point correction method based on gradient sky background is proposed.Firstly, neighborhood pixel substitution method is used to eliminate blind elements with fixed position, and then, two pure sky images with different pitch angles are used as two temperature points.Gain coefficient and bias coefficient are obtained by two-point correction, and a small amount of random noise is eliminated by modifying these two coefficients.The experimental results show that the method is effective.On the basis of correction, a small window bilateral filter and gradient template detection algorithm are proposed to detect dark vulnerable targets.Firstly, Gao Si noise is removed by two-sided filtering of small windows, then convolution with gradient template is performed, and adaptive threshold segmentation is used to detect dark weakness targets.The experimental results show that the proposed template algorithm can suppress the complex background effectively and improve the SNR of the target effectively. Moreover, the complexity of the proposed algorithm is not high and the algorithm is easy to realize in real time.
【作者单位】：中国科学院长春光学精密机械与物理研究所;中国科学院航空光学成像与测量重点实验室;中国科学院大学;吉林大学仪器科学与电气工程学院;
【基金】：国家自然科学基金项目(61308099) 吉林省重大科技攻关专项项目(11ZDGG001) 国家重点研发计划(2016YFB0501003) 上海市科委重点项目(16DZ1120400)
【分类号】：TN215;TP391.41

【相似文献】