An Optimized Framework of Video Compression Using Deep Convolutional Neural Networks (DCNN)

Video streaming demand has risen significantly in the modern world and now accounts for a significant portion of internet traffic, making it a difficult job for service providers to stream videos at high speeds while using fewer storage spaces. The existing video compression prototypes necessitate non learning based designs in order to follow inefficient analytical coding design. As a result, we propose a DCNN technique for obtaining optimal set of frames by relating each frame pixel with preceding and subsequent frames, then identifying related blocks and reducing unnecessary pixels by incorporates OFE-Net, MVE-Net, MVD-Net, MC-Net, RE-Net, and RD-Net. The proposed DCNN technique generates high video quality at low bit rates with respect to MSSIM and PSNR.