Continuous Patch Stitching for Block-wise Image Compression
🤖 AI Summary
To address the trade-off between high computational cost of full-image processing and blocking artifacts induced by conventional block-based compression in high-resolution image coding, this paper proposes the Continuous Patch Stitching (CPS) framework. CPS employs a padding-free, parallel overlapping stitching strategy to theoretically eliminate blocking artifacts; it further introduces a synergistic architecture comprising even-kernel functional residual blocks and bottleneck residual blocks to jointly optimize efficient downsampling/upsampling and deep feature expansion. Leveraging padding-free convolutions, end-to-end learning, and seamless patch-level reconstruction, CPS achieves superior rate-distortion performance while maintaining low computational complexity: it consumes less than half the computational resources of mainstream models and attains state-of-the-art compression performance across multiple benchmark datasets.
Application Category
📝 Abstract
Most recently, learned image compression methods have outpaced traditional hand-crafted standard codecs. However, their inference typically requires to input the whole image at the cost of heavy computing resources, especially for high-resolution image compression; otherwise, the block artefact can exist when compressed by blocks within existing learned image compression methods. To address this issue, we propose a novel continuous patch stitching (CPS) framework for block-wise image compression that is able to achieve seamlessly patch stitching and mathematically eliminate block artefact, thus capable of significantly reducing the required computing resources when compressing images. More specifically, the proposed CPS framework is achieved by padding-free operations throughout, with a newly established parallel overlapping stitching strategy to provide a general upper bound for ensuring the continuity. Upon this, we further propose functional residual blocks with even-sized kernels to achieve down-sampling and up-sampling, together with bottleneck residual blocks retaining feature size to increase network depth. Experimental results demonstrate that our CPS framework achieves the state-of-the-art performance against existing baselines, whilst requiring less than half of computing resources of existing models. Our code shall be released upon acceptance.
Problem
Research questions and friction points this paper is trying to address.
Eliminates block artefacts in image compression.
Reduces computing resources for high-resolution images.
Achieves seamless patch stitching without padding.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Continuous patch stitching framework
Padding-free operations strategy
Functional residual blocks design
🔎 Similar Papers
No similar papers found.
