Lossy Image Compression -- A Frequent Sequence Mining perspective employing efficient Clustering

TL;DR

This work tackles lossy image compression by replacing transform coding with a pipeline that combines closed frequent sequence mining (CFSM) with $k$-means clustering across image blocks. The compressor operates on RGB components in parallel, clustering blocks into $k$ clusters, mining closed frequent sequences with a modified support $\psi_{mod}$, and coding with Huffman codes to form the binary stream $C_I$. Empirical results on standard images show higher compression ratios than JPEG and GIF with competitive PSNR/SSIM, aided by parallel block clustering that reduces encoding time. The authors suggest future directions including alternative clustering methods and exploiting neighborhood information to further boost compression and speed.

Abstract

This work explores the scope of Frequent Sequence Mining in the domain of Lossy Image Compression. The proposed work is based on the idea of clustering pixels and using the cluster identifiers in the compression. The DCT phase in JPEG is replaced with a combination of closed frequent sequence mining and k-means clustering to handle the redundant data effectively. This method focuses mainly on applying k-means clustering in parallel to all blocks of each component of the image to reduce the compression time. Conventional GSP algorithm is refined to optimize the cardinality of patterns through a novel pruning strategy, thus achieving a good reduction in the code table size. Simulations of the proposed algorithm indicate significant gains in compression ratio and quality in relation to the existing alternatives.

Figures (5)

• Figure 1: $\psi$ vs $C_r$
• Figure 2: $k$ vs $C_r$
• Figure 3: $k$ vs $C_t$
• Figure 4: $k$ vs PSNR
• Figure 5: Lena compressed Images at 224kB