In the traditional vector quantization (VQ) encoding procedure, the quality of all parts of the image is evenly distributed. The code words selected for encoding depend on the complexity of the image and quality requirements. However, VQ cannot be used to enhance the quality of only the regions of interest (ROI). Therefore, an improved weighted cell-split algorithm based on weighted ROI blocks is proposed to train a code book to generate improved code words with emphasis on the ROI blocks. Furthermore, code words can be manipulated to allow the user to define the percentage of ROI emphasis. Experimental results show that the image quality of ROIs is significantly improved by more than 1 dB peak signal-to-noise ratio. Moreover, the improved cell-split algorithm can be easily integrated into the traditional VQ coding procedure since only the contents of the code book are modified. The algorithm can be applied to medical images where the integrity of the ROI blocks is important.
We introduce a new secret sharing scheme, where the novelty lies in the use of a fixed angle segmentation technique to create circular shadow images called shares. In this way, the shares can be stacked in different angles to reveal different secret messages. The participants must be in possession of both the shares and the stacking angles, which creates the additional degree of protection.