MODIFIED FRACTAL ANIMAL IMAGE COMPRESSION USING ARTIFICIAL NEURAL NETWORKS FOR IMPROVED ENCODING AND DECODING EFFICIENCY

Abstract

The research paper, titled “Modified Fractal Animal Image Compression Using Artificial Neural Networks for Improved Encoding and Decoding Efficiency” describes a new method of fractal image compression with the usage of ANNs. Fractal image compression mechanism heavily relies in the use of transforms whereby data within an image is compacted by partitioning an image into many small blocks and can represent a block using information from other similar blocks. The proposed method improves this technique by using ANNs for prediction of the matching blocks and transformation, which normally are computationally intensive in fractal coding. A more efficient approach for partitioning range and domain blocks is used and BPTT is applied to make updates to the encoding and decoding. The ANNs are designed to reduce the time for search and transformation prediction to reduce the complexity of the algorithm while increasing the quality of images. In this method, large improvements in compression rate and other quantifiable values such as PSNR in comparison with standard fractal compression algorithms are realized. Moreover, in terms of computational complexity and the proposed algorithm, this approach is highly recommended for applications entailing a swift results encoding and decoding process, including multimedia transmission, and storage enhancement. The results of the experiment prove the idea of this combined method, as there is relatively low file size and almost no loss in image quality. Through the application of fractal compression with the integration of ANNs to predict distant pixels and revise incorrect details, this approach responds to the problems that have a long time been associated with image compression such as time consumption and resource consumption. Highlights of the study relate to the centrality of artificial intelligence in the evolution of compression techniques required for enhanced scale and scope of digital image facilitation.