Distributed coding with sorting and assignment

Distributed coding with sorting and assignment

연구 분야: Databases

학회: Signal, Image and Video Processing

We propose a new way to perform distributed source coding by using sorting in coding and assignment for decoding. Specifically, sorting the symbols will separate information into values and positions. The decoder will use an assignment algorithm that exploits side information to try to recover the original positions of the symbols. For lossless decoding, we will show that the encoder only needs to send the minimum of position information to help modeling the assignment cost matrix. In addition, there are many ways to send this information, allowing new possibilities for implementing practical distributed coding. This technique also simplifies the modeling of probabilities and reduces the number of coding operations. Thus, we used it within the Stanford distributed video coding architecture and tested its practicality and efficiency. The video coding tests and results show that this idea can work, and its performance is similar to the existing distributed video coding techniques.