A New Image and Video Compression Technique in an Asynchronous Transfer Mode Network Environment

Abstract

A new motion video compression technique that combines the popular block matching motion compensation, vector quantization (VQ), and discrete cosine transform (DCT) coding is introduced in this paper. This technique is designed for an environment where the probability of partial transmission loss is high. This technique can be used in its entirety for moving digital video compression or a single stage can be extracted for still image coding. The proposed technique efficiently reduces the transmission bit rate, requires only a small priority layer (approximately 4% of channel space), is robust to packet loss, and is simple to implement due to its regular structure. The technique relies on creatively clustering 1D DCT frequency coefficients for VQ. This approach, as compared with other compression schemes, reduces the subjective impact of partial image loss on decoded image quality. The effects of packet loss on decoded images are studied, and the experimental results are shown.