A COMPRESSION SCHEME FOR DIGITAL CINEMA APPLICATION
Original Publication Date: 2002-Jan-07
Included in the Prior Art Database: 2003-Jun-20
COMPRESSION SCHEME FOR DIGITAL CINEMA APPLICATION Ligang Lu Jack Kouloheris 1. INTRODUCTION Compressed digital video is being used in more and more applications. Recently the multi-industry effort to develop a Digital Cinema Standard has received wide attention. This effort is to develop and standardize the techniques so that the films will be replaced by digital medium and movies will be produced, stored, and distributed in compressed digital format. Several major standardization groups and industry associations are involved in the standardization effort, such as SMPTE (Society of Motion Picture and Television Engineers), MPEG (Motion Picture Expert Group), MPA (Motion Picture Association), and NIST (National Institute of Standards and Technology). Currently the focus is on developing and determining the best enabling technologies [1,2]. The Digital Cinema application requires visually lossless video compression at a bit rate range from 35 to 90 Mbits/s for storage and playback. The original movies for Digital Cinema application are 422 sources running at 24 frames/s with 1920x1080 resolution and a bit depth of 10-bits per component sample. This demands approximately a 30:1 compression ratio at the high end. The compression techniques currently under consideration are wavelet based coding, MPEG-2 , fractal coding, etc. . MPEG-2 is a motion picture coding standard widely used in digital TV broadcasting and DVD applications. However it was developed and optimized mainly for applications with bit rates below 35 Mbits/s. Also it does not support the 10-bit sample depth. The fractal encoding process is very computational intensive and thus very costly for real time implementation. Recently the joint photo expert group (JPEG) has finalized a new still image compression standard called JPEG-2000 . JPEG-2000 is largely based on a state-of-the art digital wavelet based coding technique called embedded block coding with optimal truncation (EBCOT). JPEG-2000 provides superior performance over existing still image compression standard on rate-distortion and subjective image quality. In addition, it includes many modern features, such as fine spatial and resolution scalabilities, strong error resilience, etc., which are very useful and important to many high-end and emerging applications. Therefore it is interesting and meaningful to investigate the potential of applying JPEG-2000 to the Digital Cinema application. However, JPEG-2000 is a still image compression technique and due to the content variations and camera operations, movies are statistically nonstationary video sources. Although, in compressing an image, JPEG-2000 has the ability to achieve a given target number of bits very accurately, the target number of bits must be first determined smartly. Compressing the movies at a fixed bit rate per picture will inevitably result in undesirable and often dramatic visual quality swing over the reconstructed picture sequence. Hence a rate allocation scheme is essential to use JPEG-2000 to compress movies. This rate allocation scheme should wisely determine the bits spent on each picture to achieve smooth visual quality over the picture sequence while maintaining the required total bit rate.