基于局部卷积的视频图像像素补偿方法  

Design of Pixel Compensation Algorithm for Video Image Based on Local Convolution

作　　者：查怀志 ZHA Huaizhi(School of Education,Anqing Normal University,Anqing 246133,Anhui China)

机构地区：[1]安庆师范大学教师教育学院,安徽安庆246133

出　　处：《吉首大学学报（自然科学版）》2023年第6期1-8,共8页Journal of Jishou University(Natural Sciences Edition)

基　　金：安徽省高校人文社会科学研究项目(SK2018A0557)。

摘　　要：为了修复高速运动目标视频图像的缺失像素,提出了一种基于局部卷积的视频图像像素补偿方法(局部卷积补偿方法).首先引入局部卷积方法,分割神经网络模型,重建视频图像帧;再激活源视频图像及对应像素空间;最后采用线性均值补偿方法补偿视频图像内插帧空洞,并判断视频图像匹配程度,完成视频图像像素补偿.对比实验结果表明,局部卷积补偿方法的平均梯度为11.239,明显高于遥感图像分类方法、单幅图像超分辨率重建方法和多尺度递归图像超分辨率重建方法的平均梯度,说明经过局部卷积补偿方法补偿后的视频图像具有较高的清晰度.A video image pixel compensation method based on local convolution(local convolution compensation)is proposed to repair missing pixels in high-speed moving targets.The method introduces local convolution to segment the neural network model and reconstruct video image frames.It activates the source video image and its corresponding pixel space,applies linear mean compensation to fill the interpolated frame holes in the video image,and evaluates the matching degree of the video image to complete the pixel compensation.Comparative experimental results show that the average gradient of the local convolution compensation method is 11.239,significantly higher than the other three comparative methods,which indicates that the video image after compensation using the local convolution method has higher clarity.

关 键 词：局部卷积 视频图像 补偿方法 视频帧 图像块 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]