一种基于编码孔径的双字典视频重构算法

A double dictionary video reconstruction algorithm based on coded aperture

为提高普通相机帧率,通过结合编码孔径成像技术与高/低分辨率双字典学习方法,设计了基于时间维压缩的视频重构算法。利用编码孔径采样获得输入数据,分别训练高/低分辨率双字典,通过限制2套字典对图像的稀疏表达系数相同,既降低了算法复杂度,又实现了图像空间分辨率和时间分辨率的共同提高。仿真结果表明:该算法可有效地从获取的单张编码混叠图像中重构出高质量的多帧连续图像(即视频),同时该算法重建图像的峰值信噪比(PSNR)和结构相似度(SSIM)要优于单字典及GMM算法。

In order to improve the normal camera frame rate,through coded aperture imaging technology and high/low resolution dual dictionary learning methods,a video reconstruction algorithm based on time-dimensional compression was proposed.Use the coded aperture sampling to obtain the input data,training the high/low resolution dictionaries,by restricting the two sets of dictionary have the same sparse expression coefficient of the image,it not only reduced the complexity of the algorithm,but also achieved a common improvement in image spatial resolution and temporal resolution.The simulation proved that the algorithm can effectively reconstruct high-quality multi-frame continuous images(video) from an acquired coded aliasing images,and the peak signal-to-noise ratio(PSNR) and structural similarity(SSIM) of the reconstructed image of the algorithm were better than single dictionary and GMM algorithms.