Compressed Sensing Image Processing Method Based on Optimized Layered Discrete Cosine Transform

Abstract

Compared with the layered DCT coded image, the compressed sensing image processing method based on the optimized layered discrete cosine transform of the present invention still has multi-resolution characteristics, the amount of data collected by the image and the computational complexity of encoding and decoding are all reduced, and Image compression has been significantly improved. In the present invention, the top-level data of the layered DCT is no longer subjected to DCT transformation coding, but the top-level data are directly used as low-frequency sub-bands, and the high-frequency sub-bands are randomly measured by compressed sensing, the coding steps are simplified, and the coding efficiency is further improved. It is beneficial to further optimize the network transmission of images / videos; in the decoding process, the top-level data IDCT process is correspondingly omitted. The invention makes the coding and decoding complexity of the layered DCT simpler than that of the original layered DCT, and the coding effect is more remarkable. By comparing the standard test images, the experimental results show that the quality of the restored images is relatively good, and the peak signal-to-noise ratio of the standard 512×512 test image Lena compression can be increased by 2-4dB when the compression rate is high.

Description

technical field [0001] The invention relates to digital image and digital video compression coding methods in digital signal transmission, in particular to a discrete cosine transform coding compression sensing image processing method with multi-resolution characteristics suitable for "multi-channel environment" or progressive transmission. Background technique [0002] In the transmission of images, videos or other network signals, the conversion of signals from analog to digital has always strictly followed the Nyquist sampling theorem, that is, the sampling rate must be more than twice the signal bandwidth to accurately reconstruct the signal. With the sharp increase in people's demand for information and the continuous enhancement of the sensor system's ability to acquire data, the amount of data to be processed is also increasing, which puts forward higher requirements for signal processing capabilities and brings corresponding hardware equipment. Great challenge. In p...

AI Technical Summary

Problems solved by technology

[0005] In the CS algorithm of the prior art, the decomposition level has a great influence on the reconstruction result. If the decomposition level is too small, the reconstruction effect will be poor. With the increase of the decomposition level, the reconstruction effect will be enhanced. Finally, the original image is divided into high-frequency sub-bands and low-frequency sub-bands. The high-frequency sub-bands can be considered sparse, but the low-frequency sub-bands are the approximation signals of the original image at different scales, and cannot be considered sparse. Multiplying the measurement matrix Φ together with the high-frequency coefficients destroys the correlation between the coefficients of the low-frequency approximation components, resulting in poor reconstruction

Therefore, the transformation level of the layered DCT should be as large as possible, generally more than 3 layers, but usually the quality of the restored image is relatively poor

Images