LucidSee is a research software on image and video data compression. It makes significant improvement in high quality image coding technology. It pursues and excels at enhancing the real (visual) image quality regardless of numerical metrics such as PSNR or SSIM. It especially excels at high performance of "visually lossless" data compression in high definition (HD).
Performance LucidSee can offer the visual quality of raw image in only data size of JPEG. With the same data size, it may improve image visual quality by 30 - 60% over JPEG XR and may outperform JPEG2000 by 20 - 50%. A demo version of LucidSee may download here to verify its high performance. (To better observe the difference in image quality, the best display resolution is 1024*768. The best brightness and contrast levels are both at 60%.)
Core algorithm of LucidSee is independently invented by Dr. Xiteng Liu, PhD in Mathematics, University of South Carolina, USA. Functionalities and techniques are related to JPEG2000, JPEG XR (Microsoft HD Photo), H.264, MPEG4 and JPEG. LucidSee has evolved through a number of versions which are based on diverse mathematical transforms and different quantization and entropy coding mechanisms. For details of LucidSee coding algorithms, click here.
Comments After carefully reviewed the performance of LucidSee, managing director of Microsoft Research, Dr. Henrique Malvar said "It's very interesting," in IEEE & Computer Society Data Compression Conference 2007. On the same occasion, Dr. William Pearlman, professor of Rensselaer Polytechnic Institute, also commented that "It's amazing."
This website is designed to demonstrate LucidSee, including its technical performance, algorithm and testing results. One important fact is that all testing results reported here are verifiable, including the comparison results of LucidSee vs JPEG, vs JPEG XR, and vs JPEG2000.
