Study of Digital Image Compression Techniques and Framework for Suitability Selection based on Similarity Measuring Metrics

In recent years, capturing and transmitting color images through the network to convey visual information in a short time has become a crucial task. Choosing the appropriate compression technique is a challenging endeavor due to technological advancements with inccrement of the memory usage. Various lossless and lossy image compression techniques have been developed over the years. However, the studies pertains to a lack of extensive assessment across various compression algorithms and metrics, necessitating a practical comparative analysis of existing methodologies for color image optimization in graphical representation and storage. Hence, a comprehensive comparative study is conducted on six types of lossless compression techniques (Huffman, RLE, LZW encoding, JPEG2KLeGall53, PNG, WebP) and six types of lossy compression techniques (JPEG-DCT, SVD, K-means, JPEG2K-CDF97, PNG, WebP), evaluating their performance based on input parameters for compresion such as quality factor, selection of singular values, and number of clusters. The outputs are analzed followed by the extracted parameters such as compression ratio (CR), time, RMSE, PSNR, SSIM, FSIM, and bpp to determine the best case of compression. Furthermore, a weighted Friedman ranking test is conducted to determine the most effective compression technique. It is concluded that JPEG2K-CDF97 is optimal for lossy compression, while WebP is considerable for lossless compression.