A perceptually-lossless wavelet-based image compression algorithm and architecture working directly in the R-G-B color domain is proposed. Even though the highest compression ratios are achievable in the Luminance-Chrominance domain, image sensors generate data commonly in a subsampled R-G-B format. Many real-time applications (e.g., a low cost digital camera) prohibit the on-line conversion of data from R-G-B format to any other format. Therefore, in these applications, it is desirable to apply the compression directly on R-G-B data. Our compression scheme is based on derived perceptually-lossless quantizing thresholds, and can be mapped into an ASIC yielding very high throughput. Examples illustrating the performance of the coder are presented.
ASJC Scopus subject areas
- Theoretical Computer Science
- Computer Science Applications
- Computational Theory and Mathematics
- Artificial Intelligence