A dsp-based video frame sparsity detection system and data processing method

Abstract

The invention discloses a video frame sparsity detection system based on a DSP and a data processing method. A video camera is employed to acquire scene images, image decoding is realized through a decoding chip, video frame data is read through the DSP and is stored, two-dimensional sparse transformation of a video frame is carried out, and a video frame sparse coefficient is calculated through transformation coefficient energy normalization, former address descending sort and energy screening and is further displayed. The system is advantaged in that the DSP hardware is employed to realize quantitative determination of video scene sparse characteristics, the effective and real-time sparsity information can be provided for an adaptive compression imaging system, adaptively selecting a compression sampling rate is facilitated, the video frame is guaranteed to have relatively high reconstruction quality, and reduction of observation value compression and storage space reducing demands for videos can be guaranteed as much as possible.

Description

technical field [0001] The invention relates to the technical field of image and video signal processing, in particular to a compression observation number acquisition method in the hardware realization field of a compression imaging system. Background technique [0002] Compressed sensing (compressive sensing, CS) imaging is a new theory and technology developed in the past ten years. Traditional signal sampling must abide by the Nyquist sampling theorem, and its sampling frequency is at least twice the highest frequency of the signal; traditional image video is based on Nyquist sampling, and then compressed by H.264 and other compression methods, discarding a lot of redundancy information, wasting a lot of storage space and computing resources. Compressed sensing is based on the theory of signal sparse representation, and makes full use of the structural sparsity of the signal itself. By selecting an appropriate measurement matrix, the signal can be compressed and sampled...

AI Technical Summary

Problems solved by technology

In 2006, Baraniuk of Rice University proposed a single-pixel camera using a digital micromirror device (DMD) and a single-pixel detector, but the control is complicated, the cost is high, and the real-time performance is poor.

[0005] In short, most of the current compressed sensing research is non-adaptive compressed sensing imaging, which lacks image sparsity information

Images