Hybrid Image Compression Using DCT and Autoencoder
Dibyalekha Nayak, Tejaswini Kar, Kananbala Ray, J. V. R. Ravindra, Sachi Nandan Mohanty · 2024
Medical image processing is an emerging area that has an effect on recognition, analysis and treatment techniques of the diseases. In medical image processing compression is required to reduce the bandwidth and storage related issues. In this method DCT and Autoencoder based image compression technique is used for the better compression of the medical brain images. To get the separate Region of Interest(ROI) and non-ROI based region Otsu method is considered. DCT and Autoencoder based hybrid compression is achieved on ROI and non-ROI regions of an image respectively. The evaluation parameter of this method are Peak Signal to Noise Ratio (PSNR), Structure Similarity Index matrix (SSIM), Feature Similarity Index Matrix (FSIM), Compression Ratio (CR) and Space saving(SS) percentage. For the medical brain images proposed method has PSNR of 44.45db, SSIM of 0.97, FSIM of 0.99, CR 9.26 and SS of 89.9(%).