Research on integral (3D) image compression technology based on neural network

Three dimensional image systems offer significant benefits over conventional two dimensional systems. Integral imaging is a promising technique for delivering high-quality three-dimensional content. However, a large amount of data produces during acquisition prohibits direct transmission of Integral Image data. In this work we propose volume pixel matching encoder and a neural network color restoration for Integral Image compression. The proposed encoder is based on pixel matching to reconstruct discarded elemental images and use the method of neural network to recovery the color. Comparing with JPEG compression method, the algorithm is more effective as show in PSNR.