CPE COIN++: Towards Optimized Implicit Neural Representation Compression Via Chebyshev Positional Encoding

Haocheng Chu, Shaohui Dai, Wenqi Ding, Xin Shi, Tianshuo Xu, Pingyang Dai, Shengchuan Zhang, Yan Zhang, Xiang Chang, Chih-Min Lin, Fei Chao, Changjiang Shang, Qiang Shen

Abstract: COIN++ is a special variant of Implicit Neural Representation (INR), which encodes signals as modulations applied to the base INR network. It is becoming a promising method for applications in image compression. However, INR’s effectiveness is hindered by its inability to capture high-frequency details in the image representation. We propose a novel COIN++ framework using Chebyshev approximation to enhance high-frequency signal learning and image compression. In addition, we design an adaptable image partitioning technology and an integrated quantization method to further the image compression performance of COIN++ in the framework. Experiments demonstrate our framework significantly enhances both representational capacity and compression rate compared to the COIN++ baseline, with notable PSNR improvements.