Layer Selection in Progressive Transmission of Motion-Compensated JPEG2000 Video

MCJ2K (Motion-Compensated JPEG2000) is a video codec based on MCTF (Motion- Compensated Temporal Filtering) and J2K (JPEG2000). MCTF analyzes a sequence of images, generating a collection of temporal sub-bands, which are compressed with J2K. The R/D (Rate-Distortion) performance in MCJ2K is better than the MJ2K (Motion JPEG2000) extension, especially if there is a high level of temporal redundancy. MCJ2K codestreams can be served by standard JPIP (J2K Interactive Protocol) servers, thanks to the use of only J2K standard file formats. In bandwidth-constrained scenarios, an important issue in MCJ2K is determining the amount of data of each temporal sub-band that must be transmitted to maximize the quality of the reconstructions at the client side. To solve this problem, we have proposed two rate-allocation algorithms which provide reconstructions that are progressive in quality. The first, OSLA (Optimized Sub-band Layers Allocation), determines the best progression of quality layers, but is computationally expensive. The second, ESLA (Estimated-Slope sub-band Layers Allocation), is sub-optimal in most cases, but much faster and more convenient for real-time streaming scenarios. An experimental comparison shows that even when a straightforward motion compensation scheme is used, the R/D performance of MCJ2K competitive is compared not only to MJ2K, but also with respect to other standard scalable video codecs.

A large-scale solar dynamics observatory image dataset for computer vision applications

The National Aeronautics Space Agency (NASA) Solar Dynamics Observatory (SDO) mission has given us unprecedented insight into the Sun’s activity. By capturing approximately 70,000 images a day, this mission has created one of the richest and biggest repositories of solar image data available to mankind. With such massive amounts of information, researchers have been able to produce great advances in detecting solar events. In this resource, we compile SDO solar data into a single repository in order to provide the computer vision community with a standardized and curated large-scale dataset of several hundred thousand solar events found on high resolution solar images. This publicly available resource, along with the generation source code, will accelerate computer vision research on NASA’s solar image data by reducing the amount of time spent performing data acquisition and curation from the multiple sources we have compiled. By improving the quality of the data with thorough curation, we anticipate a wider adoption and interest from the computer vision to the solar physics community.