Low-complexity video coding based on two-dimensional singular value decomposition.

Area of Science:

• Digital Signal Processing
• Image and Video Compression

Background:

• Traditional video coding relies on motion estimation (ME), increasing computational complexity.
• Frame decoding dependency in hybrid codecs limits random access capabilities.

Purpose of the Study:

• To propose a low-complexity video coding scheme leveraging 2-D singular value decomposition (2-D SVD).
• To achieve high coding efficiency by exploiting the energy compaction property of 2-D SVD coefficients.
• To offer a superior compromise between computational complexity and temporal redundancy reduction compared to existing methods.

Main Methods:

• Utilizing 2-D singular value decomposition (2-D SVD) to exploit temporal correlations in visual signals.
• Applying the energy compaction property of 2-D SVD coefficient matrices for efficient coding.
• Developing a non-motion-estimation-based video coding approach.

Main Results:

• The proposed 2-D SVD scheme achieves high coding efficiency.
• It offers a better trade-off between computational complexity and temporal redundancy reduction.
• The method avoids issues like unavailability of random access inherent in frame-dependent coding.

Conclusions:

• The 2-D SVD video coding scheme presents a viable low-complexity alternative.
• It demonstrates advantages over existing non-ME-based low-complexity codecs.
• The approach shows significant potential for various video coding applications.