oa A novel wavelet-based multimodal compression scheme for joint image-signal compression

Background: Nowadays, by considering the important advances in multimedia and networks including telemedicine applications, the amount of information to store and/or transmit has dramatically increased over time. To overcome the limitations of transmission channels or storage systems, data compression is considered a useful tool. In general, data compression addresses the problem of reducing the amount of data required to represent any digital data including images and signals. This can be achieved by removing redundant information where the main challenge is to reduce the bit-rate while preserving a high quality of the information. Objective: This work aims to propose a new multimodal compression methodology allowing compression of jointly various data, not necessary of the same type, using only one codec. Method: The proposed joint signal-image compression methodology consists of inserting the wavelet coefficients of a decomposed signal in the details region of a wavelet transformed image at the finest scale (i.e., the highest frequency sub-bands: horizontal details, vertical details, or diagonal details) according to a spiral way. The mixture is afterwards compressed using the well-known SPIHT algorithm (Set Partitioning In Hierarchical Trees). This process is inverted by decoding the mixture, then separating the signal from the image using a separation function applied to the insertion detail sub-band area. Next, the signal and image are reconstructed using inverse wavelet transformation followed by a dequantization step. Figure 1 illustrates the corresponding compression scheme. Results: The proposed method was evaluated on medical images with biomedical signals. The experimental results obtained show that this method provides better performance compared to the basic version based on inserting the signal samples in the spatial domain of the image to the encoding phase. This is confirmed by an important obtained improvement in terms of PSNR (Peak Signal to Noise Ratio) and PRD% (Mean Square Difference).