JPEG Quantization Table Optimization by Guided Fireworks Algorithm

• Published in: Combinatorial Image Analysis Vol. 10256; pp. 294 - 307
• Main Authors: Tuba, Eva, Tuba, Milan, Simian, Dana, Jovanovic, Raka
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2017; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: Fireworks algorithm; Image processing; JPEG algorithm; Quantization tables; Swarm intelligence
• ISBN: 9783319591070; 331959107X
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-59108-7_23

Digital images are very useful and ubiquitous, however there is a problem with their storage because of their large size and memory requirement. JPEG lossy compression algorithm is prevailing standard that solves that problem. It facilitates different levels of compression (and the corresponding quality) by using recommended quantization tables. It is possible to optimize these tables for better image quality at the same level of compression. This presents a hard combinatorial optimization problem for which stochastic metaheuristics proved to be efficient. In this paper we propose an adjustment of the recent guided fireworks algorithm from the class of swarm intelligence algorithms for quantization table optimization. We tested the proposed approach on standard benchmark images and compared results with other approaches from literature. By using various image similarity metrics our approach proved to be more successful.