Abstract Second generation image coding techniques, which use information about the human visual system to reach high compression ratios, have proven very successful when applied to single images. These methods can also be applied to image sequences. A directional decomposition based sequence coding technique is presented, in which spatial lowpass and highpass components are analyzed and coded separately. A simple law for sharing the available bits between these components is stated and analytically proved by using a minimum cost/resolution optimality criterion. The detection of directional elements is carried out by using both linear and nonlinear (median) filtering. The coding is based on near optimal estimators which retain only the innovation part of information, and is well suited for differential pulse code modulation. The results of applying this method to a typical sequence are shown. The estimated compression ratio is approximately 320 : 1 (0.025 bits per pixel), allowing a transmission rate of about 41 kbit/second. The resulting image quality is reasonably good.
[1]
W. F. Schreiber,et al.
Synthetic Highs — An Experimental TV Bandwidth Reduction System
,
1959
.
[2]
Ernest L. Hall,et al.
Computer Image Processing and Recognition
,
1980,
IEEE Transactions on Pattern Analysis and Machine Intelligence.
[3]
Murat Kunt,et al.
Very Low Bit Rate Image Sequence Coding Using Object Based Approach
,
1988,
Other Conferences.
[4]
Murat Kunt,et al.
Recent results in high-compression image coding (Invited Papaer)
,
1987
.
[5]
Patrick Bouthemy,et al.
A Maximum Likelihood Framework for Determining Moving Edges
,
1989,
IEEE Trans. Pattern Anal. Mach. Intell..