Novel 2-D Histogram-Based Soft Thresholding for Brain Tumor Detection and Image Compression

Chiranjeevi Karri, G. Babu, P. Prasad, M. Naidu

OAI: oai:igi-global.com:292497 • DOI: 10.4018/IJAMC.292497

The objective of image compression is to extract meaningful clusters in a given image. Significant groups are possible with absolute threshold values. 1-D histogram-based multilevel thresholding is computationally complex and reconstructed image visual quality comparatively low because of equal distribution of energy over the entire histogram plan. So, 2-D histogram-based multilevel thresholding is proposed in this paper by maximizing the Renyi entropy with a novel hybrid Genetic Algorithm, Particle Swarm Optimization and Symbiotic Organisms Search (hGAPSO-SOS), and the obtained results are compared with state of the art optimization techniques. Recent study reveals that PSNR fails in measuring the visual quality because of mismatch with the objective mean opinion scores (MOS). So, we incorporate a weighted PSNR (WPSNR) and visual PSNR (VPSNR). Experimental results examined on Magnetic Resonance images of brain, and results with 2-D histogram reveal that hGAPSO-SOS method can be efficiently and accurately used in multilevel thresholding problem.

Algorithms Computer Science and Information Technology Artificial Intelligence