摘要
针对电力监控系统中输电线路易晃动造成覆冰图像模糊,提出基于光流法运动配准的覆冰图像超分辨率重建算法。该算法首先利用基于光流法的图像配准算法估计图像间的亚像素级运动矢量,得到前向和后向配准图像;然后,利用迭代反投影(IBP)算法对估计出的图像分别进行超分辨率重建;最后,根据配准误差加权得到最终的输出图像。实验结果表明,相对于传统算法,该算法的重建结果无论是在峰值信噪比和标准差还是基于对比敏感度的无参考图像清晰度上,均有明显提高,具有较好的客观指标和视觉效果。
Concerning the penumbra of icing image caused by transmission line shaking in power monitoring system, we propose a super-resolution image reconstruction algorithm based on optical flow motion registration. The algorithm firstly estimates the sub-pixel level motion vector between images based on optical flow method, and the forward and backward registration images are obtained. Secondly, the iterative back-projection （IBP） algorithm is used to reconstruct the estimated image respectively. Finally, the output image is obtained by weighting the registration error. Experimental results show that compared with the traditional algorithms, the reconstructed results of this algorithm have obvious improvement in both the PSNR and standard deviation or the no reference image sharpness assessment based on contrastsensitivity （NRSACS） , and have better objective and visual
出处
《计算机应用与软件》
2017年第6期80-84,155,共6页
Computer Applications and Software
基金
重庆市教委科学技术研究项目(KJ1501022)
重庆高校创新团队建设计划资助项目(CXTDX201601034)
关键词
超分辨率重建
光流法
IBP算法
运动配准
覆冰图像
Super-resolution reconstruction Optical fleffects.method IBP algorithm Motion registration Icing image
