Effective Video Encoding in Lossless and Near-lossless Modes

Abstract

The main goal of the research is to develop an algorithm appropriate for hardware realization of a lossless and near-lossless video sequences. The importance of compression is unquestionable in contemporary digital video signal processing systems, with an extensive requirement on a high storage and bandwidth imposed on data storage and transmission. For example, a 60 minutes long uncompressed movie with the TV PAL quality (i.e., 25 frames per second, 720x576 pixels) requires 104.28 GB hard disk space and 9.89 Mpixels/s of bandwidth in the 4:4:4 profile. One of the major requirements for the system is its work in the real-time. This is caused by one of the primary applications of the lossless video compression, i.e., edition of television programs, movies etc. (Andriani et al., 2004). In this situation it is not recommended to use lossy compression methods, such as MPEG2, MPEG4 etc. Other important applications of the lossless video sequence compression is a storage of medical 2D and 3D images (in general, multi-planar 3D objects) (Xie et al., 2007), as well as astronomical images, and satellite photos compression (photographs of the Earth, Mars and other celestial bodies made by satellites and spaceships should be sent and stored in a lossless form) (Chen et al., 2004; CCSDS, 2007). A novel kind of application is the slow motion video technique for recording from scientific experiments up to dozen of thousands frames per second (fps). For example, recording of one second long video sequence (lumination signal, 8 bit/pixel) using 1000 fps and the SD (720576 pixel) resolution needs 3955 MB of memory. Nowadays video cameras in the slow motion mode stores the images without any compression and the time of the recording is limited with the volume of the storage card in the device. Recording and archiving files of a larger size becomes an important issue that may be solved by introducing fast, hardware-based, parallel compression realization, e.g., in the near-lossless mode (see Section 3). In combination with the matrix of the effective SSD discs, it may lead to the increase of the functionality of the scientific site using the slow-motion camera.