Locality-Aware Density Control for Efficient Gaussian-based Image Representation
2026-07-20 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors study a way to represent images using 2D Gaussians, which are simple spot-like shapes that can be drawn quickly. They notice current methods don’t use these Gaussians efficiently, either missing details or keeping too many similar spots. To fix this, they create a system called LocoADC that smartly adds Gaussians where more detail is needed and merges similar ones to avoid overlap. Their approach uses ideas about how errors and Gaussians are grouped locally to improve image quality without using more Gaussians. Tests show LocoADC improves image reconstruction quality on several datasets.
2D Gaussian Splattingimage representationrasterizationGaussian allocationlocalityreconstruction errorregion-wise densificationGaussian mergingPSNRCLIC dataset
Authors
Jiacong Chen, Qingyu Mao, Xiandong Meng, Shuai Liu, Chao Li, Fanyang Meng, Youneng Bao, Yongsheng Liang
Abstract
2D Gaussian Splatting is an attractive direction for image representation due to its explicit formulation, fast rasterization, and favorable decoding efficiency. The representation quality of this paradigm depends on the proper allocation of Gaussian capacity to the demanding regions. However, existing methods fail to allocate Gaussian capacity efficiently during optimization: under-reconstructed content is often refined in a fragmented pixel-wise manner, while neighboring optimized Gaussians with similar attributes are redundantly retained. This inefficiency motivates the need for a density control framework that jointly addresses insufficient allocation in under-reconstructed regions and redundant allocation in over-reconstructed regions. Our key insight is that this framework should exploit two complementary forms of locality: the local continuity of reconstruction errors in image space for improved Gaussian allocation, and the local similarity of neighboring Gaussians in Gaussian space for redundant elimination. Based on this insight, we propose Locality-Aware Density Control (LocoADC), a plug-and-play framework that improves Gaussian capacity utilization through Region-wise Gaussian Densification (RGD) and Similarity-Driven Gaussian Merging (SDGM) strategies, together with a local color consistency constraint for more reliable merging. Extensive experiments on diverse datasets show that LocoADC consistently improves multiple baselines by enabling more effective local Gaussian allocation, including a 2.93 dB PSNR gain over GI on the CLIC dataset under the same 30k Gaussian budget. Code is available at: \textit{https://github.com/ChenJiaCong-1005/LocoADC}.