| Medical ultrasound imaging technology is one of the most important detection methods in the field of medical image processing.It can display in detail the massive and complex information of body tissues.It has the advantages of high safety,non-invasiveness,convenient use and low price,and it is widely used in clinical monitoring of various diseases.The various stages of the process,especially the breast,abdominal organs and the development of pregnant women and fetuses play an irreplaceable role.However,during the acquisition and transmission of ultrasound images,due to the unevenness of the acoustic impedance of various tissues of the human body and the randomness of the spatial distribution,coupled with frequent switching of the operating frequency,speckle noise with different brightness and darkness is easily formed in the image It greatly reduces the quality of the image,blurs the edge details,and seriously affects the identification and positioning of the lesion area,making the inspection of the subtle lesions more complicated,and increasing the difficulty of medical diagnosis and treatment.Therefore,this paper proposes a denoising algorithm combined with transform domain frequency division and spatial domain denoising.The main research contents are as follows:First of all,with the development of ultrasound imaging technology,in order to improve the image quality of ultrasound imaging,dynamic frequency scanning technology is applied to ultrasound imaging equipment.It can automatically switch the working frequency according to the depth of the detection target,ensuring the resolution of ultrasound imaging.While increasing the penetration depth of the probe.Consider adopting a novel decomposition algorithm in recent years-Two-Dimensional Variational Mode Decom-position 2D-VMD,making full use of the advantages of 2D-VMD adaptive decomposition,which can be decomposed in the sub-image The feature information can be effectively extracted,and the noise can be separated to the greatest extent for processing,and the original image information can be retained.This paper chooses to divide the decomposed image into low-frequency and high-frequency images.The first sub-modal image containing a large amount of original image information is used as low-frequency image processing.Other sub-modalities contain texture information and noise,and they are all divided into high-frequency images.Since the divided image contains different image information,the low-frequency sub-modal image is subjected to anisotropic diffusion filtering for denoising.For the high-frequency sub-modal image,the BM3 D method is used to denoise,and the denoised image is reconstructed.Each sub-modal image gets the final image.In addition,in the decomposition process,most scholars did not give a detailed explanation on the selection of the parameter K,but only used empirical methods to determine or directly use the default value,so that the decomposition algorithm has problems of over-decomposition and underdecomposition.Therefore,this paper analyzes the information shared between sub-modal images,studies a method for adaptively determining the parameter K value based on VIF,and verifies the proposed method through experiments.Finally,a thorough experiment is carried out.A large number of experimental verifications are carried out on the composite graph,the simulated graph and the actual graph in turn,and the quality evaluation index is compared with other 10 different types of algorithms;in order to illustrate the efficiency of this algorithm,the average algorithm computing time is increased in the process of the comparison experiment.Comparison: At the same time,some detailed magnified view processing is done for the 4 types of actual ultrasound images,which further increases the intuitive denoising effect comparison.Research shows that this algorithm is better than other algorithms in denoising and maintaining structural details,but the timeliness of denoising is not good enough,and it needs to be improved in the future. |
PDF Full Text Request