Xidian Meeting Xidian Guide About Help Search Home Login Control Panel AddBookMark Rui Song's MessageBoard

Quality Assessment for 3D Image.
In the past 20 years, visual quality assessment for 2D images has attracted lots of attention. Thousands of researchers devoted themselves to design a robust and universal objective metric which is expected to be closest to the subjective scores. And the target is nearly being accomplished in the coming years.
But for 3D images, which is a totally different game, the design of the corresponding objective metric is still in its infancy. For 3D TV, the acknowledged MVD scheme which is first proposed by Fehn is expected to be most prevalent. However, there are still some unanswered problems. DIBR bring us new challenges. We wish we can do better rendering, but quality metrics for 2D images such as PSNR or SSIM failed to characterize the trait, and no effective objective quality metric support our results. Therefore, the quality assessment is of great interests.


Image Super-resolution.
Limited by the image acquisition equipment or environmental condition, e.g., earth observation by satellite, the image resolution can hardly meet the application requirements. Super resolution is an effective way to get a higher resolution image under current hardware configuration. Given several images of one scene, super resolution technique could reconstruct a high resolution image by exploring the displacement between different images and fuse these complemental information.


Video Coding Algorithms.
Algorithm research for video coding is applicaiton oriented. The newly released HEVC standard is characterized by more complex transform, prediction, and motion estimation methods, which brings us more challenge for realtime encoding and decoding. Fast algorithms for prediction, motion estimation and other algorithms are requested by the community.


The most efficient viedo compression system is the one based on elaborately designed VLSI architecture. The hardware architecture, pipeline, and dataflow control are important issues for hardware implementation of video coded. What we concerned are the performance and balance among logic resource, memory, power, maximum frequency and size.


For implementation of custom-built realtime video processing systems, we always use em[ant]beded subsystems. The most popular combination is FPGA+DSP.