Research On Compressive Sensing Coding Technology Of COVID-19 Images

In recent years,with the outbreak of the COVID-19,a medical image of chest film,as an auxiliary diagnostic basis for COVID-19,has become increasingly valuable.It can help doctors judge the severity of the disease and determine appropriate treatment plans.The acquisition speed and recovery quality of image data are particularly important in emergency medical operations in various regions.When obtaining medical images and transmitting them to doctors,image processing technology plays an important role in this process.The storage & transmission of images is one of the key links in medical image processing.Due to high infectivity of COVID-19 virus,medical institutions need to process a large number of patient image data in a short time,so the traditional storage & transmission methods may not meet the needs of real-time processing.At this time,compressive sensing has become an important solution to deal with this problem.However,there is no compressive sensing framework optimized for COVID-19 images nowadays,which will lead to additional data transmission pressure and storage cost.For this reason,this thesis starts from the construction of compressive sensing communication framework,and then studies the feature adaptive predictive coding of quantized block compressive sensing,and finally integrates them into three stages of a telemedicine image processing system to realize the research on compressive sensing coding technology for COVID-19 images.(1)During the construction of compressive sensing communication framework,the proposed framework integrates some functional modules such as image compressive sampling,quantization,entropy coding,decoding and reconstruction.This thesis investigates the main parameters affecting the system performance,which include the selection of observation matrix,the trade-off between quantization step and compression ratio,and propose an effective method to select a parameter combination that produces near-optimal quality at any given bit rate among all possible combinations of quantization step and compression ratio.By introducing different algorithms and comparing them with traditional frameworks,the proposed framework achieves higher structural similarity,efficient image transmission and processing,which proves the superiority and practicality of the proposed framework.(2)During the predictive coding research,in order to improve the encoding performance of quantized block compressive sensing,a feature adaptive predictive coding(FAPC)method is proposed for remote transmission of COVID-19 X-ray images.The FAPC method can adaptively calculate block-level prediction coefficients based on the main features of Covid-19 X-ray images,thereby provide the optimal prediction candidate set from feature guided candidate sets.The FAPC method can achieve efficient encoding of X-ray images and quickly transmit chest images for remote healthcare.The experimental results show that compared to existing predictive coding methods,the proposed FAPC method has sufficient competitive advantages in terms of rate distortion and complexity performance.(3)Based on the communication framework and predictive coding technology mentioned above,this thesis designs and implements a telemedicine image processing system by using AppDesigner in MATLAB.According to the needs of remote film reading,some functions have been implemented such as preprocessing of pneumonia images,comparison of compression and reconstruction,and soft-tissue segmentation,which provide doctors with fast and high-quality image data.