Scalable compressed video collection and reconstruction system based on structured sparsity

Abstract

The invention provides a scalable compressed video collection and reconstruction system based on structured sparsity. The system comprises a structured sparse learning module, a signal decomposition module, a telescopic sensing module and a telescopic reconstruction module, and the structured sparse learning module uses data to drive a subspace joint model and layered subspace learning at a codingend and a decoding end to generate a sparse basis matrix. The scalable compressed sampling provided by the invention conforms to a distributed progressive structure of a video sampling process; the progressive construction of the structured sparse basis matrix also improves the accuracy and efficiency of reconstruction, improves the scalable sampling efficiency of video signals, obtains reconstruction gain under different sampling compression ratios compared with other methods, and also has good expandability.

Description

technical field [0001] The invention relates to the technical field of multimedia signal processing, in particular to a scalable and compressed video acquisition and reconstruction system based on structured sparse. Background technique [0002] In the era of intelligent information, various devices are connected through heterogeneous networks for wired or wireless transmission of information. In a heterogeneous network, the fluctuation of network bandwidth conditions, as well as the different information processing capabilities and application scenarios of various hardware devices, due to the fixed video stream of high-dimensional multimedia signal codec technology (especially video coding technology) , so there is not enough flexibility for different needs. [0003] In the past 20 years, various Scalable Video Coding (Scalable Video Coding, SVC) technologies have been proposed and standardized to reduce the coding complexity, fill the vacancy of scalable implementation so...

AI Technical Summary

Problems solved by technology

After searching the literature of the prior art, it was found that Stankovic et al. proposed a scalable compression video sampling method in the article "Scalable compressive Video" published at the "IEEE IEEE Int.Conf.ImageProcess." (IEEE ICIP) conference in 2011 , this method provides a quality scalable coding scheme for compressed video samples, but the sparse base of this method is a fixed base matrix, which cannot provide more effective sparsity and adaptability

Subsequently, B.Bojana et al. proposed to use KSVD to adaptively learn sparse patterns in the paper "Learning scalable dictionaries with application to scalable compressive sensing" published at the "IEEE Proc.European Signal Process.Conf" (IEEE EUSIPCO) conference in 2012. To support the compressed video sampling scheme with scalable quality, however, because it is based on a general sparse representation, it cannot effectively describe the non-stationary statistical properties of high-dimensional signals

Images