e-ISSN 2231-8542
ISSN 1511-3701

Home / Regular Issue / JTAS Vol. 27 (4) Oct. 2019 / JST-1262-2018


Resolution Adaptive Threshold Selection for Gradient Edge Predictor in Lossless Biomedical Image Compression

Urvashi, Meenakshi Sood and Emjee Puthooran

Pertanika Journal of Tropical Agricultural Science, Volume 27, Issue 4, October 2019

Keywords: Lossless image compression, medical imaging, predictive coding, predictor, resolution independent gradient edge detection

Published on: 21 October 2019

The high-resolution digital images generated for medical diagnosis produce the extremely large volume of digital data. This necessitates the use of image compression for medical data to be processed, archived and transmitted through a computer network in an efficient way. Due to the criticality in disease diagnostics and legal reasons, biomedical images require lossless compression to prevent permanent loss of image data. Among various approaches to lossless compression of medical images, predictive coding techniques have high coding efficiency and low complexity. Gradient Edge Predictor (GED) used in predictive coding technique for prediction has higher coding efficiency as compared to Median Edge Detector (MED) used in JPEG-LS. GED has lower computational complexity as compared to Gradient Adaptive Predictor (GAP) used in CALIC. GED is a threshold based predictor, however there is no specific method adopted in literature to decide the threshold value for prediction. This paper presents an efficient prediction solution based on predictive coding technique. The main objective of this research work is to develop a Resolution Independent Gradient Edge Predictor (RIGED) technique for choosing an optimal threshold value for GED predictor which will give minimum entropy value irrespective of the type of modality and resolution of the medical images. The empirical experimentation and analysis gave percentage improvement of the proposed model as 32.4% over MED and percentage difference between high complexity GAP and proposed predictor as 0.68 % in terms of entropy for medical image dataset of different modalities having different resolution.

ISSN 1511-3701

e-ISSN 2231-8542

Article ID


Download Full Article PDF

Share this article

Recent Articles