Adaptive Pixel-Based Technique for Grayscale Image Compression

Abstract

Grayscale images are extensively used due to their simplicity and cheapness in storage and transmission compared to color one of RGB base. It can also be considered a solution for color-blind people to ensure a better view of things and reading. However, unfortunately, it is still quite overburdened with redundancy(s) in which the data compression exploits them efficiently depending on the type and way of redundancy removal. This work introduces a hybrid compression system to compress grayscale images using the adaptive pixel-based technique (PBT) of optimized modeling base, with incorporated Minimize Matrix Size Algorithm (MMSA) of three digits values (C321) to encode the residual compactly along the need to overhead information (index, mean). Adapting traditional PBT led to overcoming the problems in the conventional PBT system in terms of large size. It achieved an acceptable reduction in bytes for the deterministic part (M, Indx) of over 400 bytes and the deterministic part (Res), where the size was reduced to more than 5000 bytes on average. The size has been reduced by nearly 50% compared to traditional PBT. The tested results indicate higher quality compared to the standard JPEG and traditional PBT in terms of performance. This includes a Compression Ratio (CR) of 13 and a PSNR of 48 dB.