The Optimization Method Of Cloud Album Compression Based On Subset Partition And Parallax Compensation

Duo to the popularity of cloud album,more and more people tend to use cloud platform to store massive photos.With the rapid development of image acquisition technology,image resolution continues increasing,and a number of images grow explosively.This feature requires the photos to be compressed to reduce storage space and the volume of data transmission.Traditional image compression methods take use of pixel redundancy,coding redundancy and visual redundancy to compress single image.However in real life,a large number of images are took in same or similar scenarios,there is a certain of set redundancy between them.If the set redundancy can be utilized effectively on the basis of single image compression,will further improve the compression ratio of cloud photo albums.Therefore,this paper studies the optimization method of cloud album compression based on subset partition and parallax compensation.Aiming at the problem of unreasonable group in existing image set compression algorithms,this paper proposes a subset partition algorithm of cloud album based on correlation coefficient averaging.The algorithm devides image set before compressing,and obtains the optimal subset by fitting the quadratic function between the average correlation coefficient and the size of image set.Then compress each subset to ensure the maximum of compression gain.The experimental results show that the optimal subset size calculated by the algorithm proposed in this paper is consistent with by the traversal algorithm,our algorithm reduces the computational complexity greatly under the premision of ensuring maximum compression gain.When the image set has great changes in rotation,scale and illumination,the ompression ratio of existing algorithms appear reduction,.Aiming at this problem,put forwards an optimization method for cloud album subset compression based on parallax compensation.For each subset,organizes the image set into pseudo-video by minimizing the prediction cost in feature domain firstly,then performs hybrid disparity compensation in spatial domain,including global(geometric and photometric changes)compensation and local(inter and Intra frame)compensation.Finally,the redundancy between each compensation signal and the corresponding target image is reduced by the context adaptive entropy coding algorithm in frequency domain.The experimental results show that the method proposed in this paper always remains robust when images existing big changes in rotation,lighting and scale.And at the same quality level,the PSNR gain is more obvious compared with the single image compression algorithm JPEG and JPEG 2000,and the compression performance is better than several existing lossless JPEG compression algorithms.