RGB Medical Video Compression Using Geometric Wavelet

Abstract

The video compression is used in a wide of applications from medical domain especially in telemedicine. Compared to the classical transforms, wavelet transform has significantly better performance in horizontal, vertical and diagonal directions. Therefore, this transform introduces high discontinuities in complex geometrics. However, to detect complex geometrics is one key challenge for the high efficient compression. In order to capture anisotropic regularity along various curves a new efficient and precise transform termed by bandelet basis, based on DWT, quadtree decomposition and optical flow is proposed in this paper. To encode significant coefficients we use efficient coder SPIHT. The experimental results show that the proposed algorithm DBT-SPIHT for low bit rate (0.3Mbps) is able to reduce up to 37.19% and 28.20% of the complex geometrics detection compared to the DWT-SPIHT and DCuT-SPIHT algorithm.