首页|基于稀疏编码的数字视频图像压缩方法研究

基于稀疏编码的数字视频图像压缩方法研究

扫码查看
针对数字视频图像采集过程中受外部环境噪声干扰及原始图像分辨率低的影响,在压缩过程中可能出现很多的失真和伪影,并且每次压缩和解压缩都会引入一定的误差,误差逐渐积累,导致最终的压缩效果较差的问题,提出基于稀疏编码的数字视频图像压缩方法研究。利用多阈值迭代方法对数字视频图像中的噪声实施去除,利于后续的图像压缩处理;通过稀疏编码方法获取去噪后的数字视频图像的正交基系数,对该系数进行冗余字典稀疏编码和压缩传输,建立多帧去压缩伪影网络,利用网络中的运动补偿模块对数字视频图像实施运动偏移估计以及像素补偿;将运动补偿帧输入去压缩伪影模块中完成压缩伪影的消除,实现数字视频图像压缩。实验结果验证该方法能有效去除压缩数字视频图像中的伪影,具有较高的压缩效率和信噪比。
Image Compression Method of Digital Video Based on Sparse Encoding
During the process of digital video image acquisition,due to external environmental noise interference and low resolution of the original image,there may be significant distortion and artifacts during the compression process.Each compression and decompression introduces a certain amount of error,which gradually accumulates,resulting in poor compression performance.A research on digital video image compression method based on sparse encoding is proposed.Using multi threshold iterative methods to remove noise from digital video images is beneficial for subsequent image compression processing.The orthogonal basis coefficients of the denoised digital video image are obtained through sparse encoding method,redundant dictionary sparse encoding and compression transmission are performed on this coefficient,a multi frame decompressing artifact network is established,and the motion compensation module is used in the network to perform motion offset estimation and pixel compensation on the digital video image.The motion compensated frames are inputted into the decompressing artifact module to eliminate compressed artifacts and achieve digital video image compression.The experimental results verify that this method can effectively remove artifacts in compressed digital video images,and has high compression efficiency and signal-to-noise ratio.

sparse encodingdigital video image compressionimage denoisingwavelet transformdecompression of artifacts

张舒野

展开 >

西安美术学院影视动画系,西安 710065

稀疏编码 数字视频图像压缩 图像去噪 小波变换 去压缩伪影

2024

吉林大学学报(信息科学版)
吉林大学

吉林大学学报(信息科学版)

CSTPCD
影响因子:0.607
ISSN:1671-5896
年,卷(期):2024.42(6)