IMAGE SEQUENCE COMPRESSION: A HYBRID TECHNIQUE USING MOTION COMPENSATION

Abstract

With the evolution of digital techniques for information processing, storage and transmission the need arises for digital representation of visual information. In digital form computer processing and digital techniques can be used to manipulate and interpret visual information. Efficient storage and economical transmission of Imagery calls for coding algorithms, which produce as little digital data as possible ,while at the same time satisfying objective and subjective criteria for reproduced image quality. In this work,both temporal and spatial redundancy present in an image sequence are studied to achieve the desired compression. Discrete cosine Transform is applied to bring down interpixel redundacies in the images. Temporal redundacies present between consecutive frames is reduced by motion compensation, wherein different block matching algorithms for motion estimation are made use of, and the concept of predictive coding is applied. A comparative performance analysis of the above said algorithms, when they are used in the hybrid coder is done.The need for quantization and utility of variable length coding in further reducing the bit rate are emphasised in such a scheme.