Satellite Image Compression, Dimensionality Reduction and Quality Metrics

Dr. Qian has been granted a series of patents worldwide on onboard near-lossless data compression techniques that tackle the huge data volume generated by satellites and the bottleneck of transmitting the data from space to ground. Modern satellites produce scientific data with large dimensionality, which presents a challenge to traditional satellite data processing techniques. Many conventional data processing methods may not be applied without dimensionality reduction as a preprocessing step. He developed novel nonlinear dimensionality reduction techniques as well as techniques to simultaneously reduce dimensionality and noise of multidimensional satellite data. To fill void in the assessment of satellite image quality after postprocessing or enhancement, Dr. Qian proposed four reduced reference quality metrics in the absence of a full reference. These reduced reference metrics have been well used to assess the satellite image quality. In this lecture, Dr. Qian will briefly introduce a number of innovative satellite signal processing technologies developed at the Canadian Space Agency for overcoming technical challenges in the development of Canadian space missions. He will also show the experimental results of the technologies and demonstrate how the innovative technologies benefit scientific community and better serve the Earth observation applications.