Distributed video compressed sensing system and method based on non-feedback bit rate control

Abstract

The invention discloses a distributed video compressed sensing system and method based on non-feedback bit rate control. The system is composed of a coder, a non-feedback bit rate controller and a decoder. Compressed sensing frames in the coder provide block measurement value residual error information for the non-feedback bit rate controller. The non-feedback bit rate controller carries out blockbit rate allocation on first compressed sensing frames (CS frames) through combination of a measurement rate-quantization parameter distortion model according to the block measurement value residualerror information and a target bit rate, provides a measurement rate and a quantization parameter of a current coding block and trains a BP neural network of a 2*3*2 structure through utilization of an allocation result. The image block CS measurement rates and quantization parameters of the CS frames are predicted through adoption of the trained BP neural network. The decoder decodes the measurement rates and the quantization parameters of received bit streams and then carries out joint decoding. According to the system and the method, the deficiency that in the prior art, only the measurement rates are allocated when the compressed sensing frames are coded is overcome, and the reconstruction effect of the compressed sensing frames is good.

Description

technical field [0001] The invention relates to a distributed video compression sensing system and method based on no-feedback code rate control, and belongs to the technical field of video coding. Background technique [0002] In wireless video communication, a low-complexity, high-rate-distortion video coding system needs to be considered. Compressed sensing technology acquires video signals by continuously using single-pixel cameras, and completes signal compression during sampling, which greatly reduces the storage space at the acquisition end and the algorithm complexity at the encoding end, which is very suitable for the requirements of the wireless environment. Distributed video compression sensing is a novel video coding scheme. This technology combines the characteristics of compressed sensing and distributed video coding, and transfers a large number of calculations from the video coding end to the decoding end, reducing the energy consumption of the video acquisit...

AI Technical Summary

Problems solved by technology

For distributed video compression sensing systems, there are relatively few studies on bit rate estimation. The existing bit rate control methods mainly focus on the research on how to reasonably allocate the number of image block measurement values ​​according to the temporal correlation and texture complexity of image blocks in video sequences. Feedback-free bit rate control with comprehensive consideration of quantization parameters is rarely reported in the literature

Images