Search papers, labs, and topics across Lattice.
This paper introduces Cluster-Guided Vector Quantization (CGVQ), a novel method for compressing Gaussian-based image representations by partitioning Gaussian parameters into homogeneous groups before quantization. This approach significantly enhances rate-distortion efficiency, achieving a 20% reduction in bits per pixel (bpp) compared to baseline methods while preserving visual fidelity. The findings underscore the potential of structured quantization techniques in optimizing image compression without sacrificing quality.
CGVQ achieves a remarkable 20% reduction in bits per pixel while maintaining visual quality, revolutionizing Gaussian-based image compression.
Gaussian-based image representations effectively model image content using compact parametric primitives while preserving high visual fidelity, yet storing a large number of floating-point parameters per primitive degrades rate-distortion efficiency at higher fidelity targets. To improve the rate-distortion performance in Gaussian representation, we present our Cluster-Guided Vector Quantization (CGVQ), a Gaussian primitive based image compression method. Our key idea is to partition Gaussian parameters further into homogeneous groups prior to quantization, enabling higher compression efficiency and accurate parameter reconstruction. In practice, our extensive experiments show that CGVQ decreases the bpp by 20% with respect to our baseline, while maintaining on-par visual quality
NVIDIA