The Research Of Image Enhancement Algorithm Based On Nonsubsampled Shearlet Transform

In recent years,with the rapid development of science and technology,image processing algorithms based on computer-aided systems are constantly updated,and their application fields are gradually increasing in people's lives,especially in medical images,remote sensing images and multi-focus images processing.The acquisition of medical images can provide useful information about the location of patient's disease,which enables doctors to diagnose and treat the patient's disease quickly and effectively,but due to the limitations of the image acquisition equipment and the interference of external factors(such as illumination intensity,temperature,etc.)in the process of shooting,the quality of the acquired image is usually low,which seriously affects the extraction of important information in the image and the subsequent processing of the image.As a special image recording geomorphological features,remote sensing images are widely used in the fields of military,national defense,people's livelihood,etc.,such as monitoring natural disasters,urban planning and construction,and dynamic monitoring of land use,etc.;However,due to the influence of atmospheric environment and sensor equipment,the acquired remote sensing images usually have the shortcomings of low definition and uneven brightness.Therefore,it is necessary to enhance these low-quality medical images and remote sensing images effectively,and the enhanced images can better reflect the real information.Due to the limited depth of field of the optical lens,it is difficult for people to obtain a clear panoramic image in photography,and because of the different focus,the multi-focus image contains different clear and blurred areas,which is not conducive to the extraction of information from the image;As a branch of image enhancement,image fusion technology can be applied to multi-focus image processing,which can effectively enhance the image.In view of the characteristics of medical images,remote sensing images and multi-focus images,it is of great significance to explore effective image enhancement algorithms in order to obtain images with good visual effects.In this paper,the problems of blurring and low contrast in medical image,remote sensing image and multi-focus image acquisition are studied in depth,the corresponding solutions are proposed,the experimental data and discussion analysisare given to verify the effectiveness and feasibility of the proposed methods.The main research contents and innovations of this paper are as follows:1.Medical image enhancement method based on nonsubsampled shearlet transform and guided filter.In order to solve the problem of low definition of the acquired medical image,a new medical image enhancement model is proposed.Firstly,the original image is decomposed by nonsubsampled shearlet transform,and a low-frequency sub-band and several high-frequency sub-bands are obtained,because the low-frequency component of the image contains a lot of background information of the image,these information will directly affect the contrast of the image.Guided filtering is a fast and effective contrast enhancement method,this method is used to process the low-frequency component in order to improve the overall contrast of the image;the high-frequency components of the image contain noise and detail information,an adaptive threshold method is used to process the high frequency part to reduce the noise interference,at the same time,the detail information of the image can be well preserved.Finally,all the effectively processed sub-bands are reconstructed by the inverse nonsubsampled shearlet transform,and the final enhanced image is obtained.The experimental results show that the proposed algorithm has obvious advantages in medical image enhancement and achieves good results in objective evaluation indexes.2.Medical image enhancement method based on gradient guided filtering and fuzzy contrast in nonsubsampled shearlet transform domain.As an important branch of medical image,brain image plays an important role in the analysis of human brain tissue,in order to improve the clarity and contrast of brain image and suppress noise interference,a medical image enhancement method based on nonsubsampled shearlet transform is proposed.Firstly,the input brain image is decomposed by nonsubsampled shearlet transform to obtain low-frequency sub-band and high-frequency sub-bands,gradient domain guided filtering is an effective image enhancement method with low computational complexity,it is used to process the low-frequency component of the image in order to improve the contrast;the improved fuzzy contrast method is used to process the high-frequency components of the image effectively to reduce the noise interference.Finally,the final enhanced image is reconstructed by the inverse nonsubsampled shearlet transform.The experimental results show that this algorithm has a good effect in detail preservation and contrast enhancement of brain image,and it has certain advantages in objective evaluation index data.3.Remote sensing image enhancement algorithm based on nonsubsampled shearlet transform and local laplacian filter.Due to the acquired remote sensing image has the disadvantages that the visual contrast and spatial resolution can not fully meet the application requirements,it is necessary to enhance remote sensing image effectively before analysis and interpretation,therefore,a new remote sensing image enhancement method is proposed.Firstly,the initial low-quality remote sensing image is decomposed by nonsubsampled shearlet transform,and the low-frequency component and the high-frequency components are obtained,respectively;then the low-frequency component of the initial image is processed by local laplacian filter algorithm in order to improve the contrast of the image and suppress a small amount of noise in the low-frequency component,the improved threshold algorithm is applied to the high-frequency components to eliminate noise interference;finally,all sub-bands are reconstructed by inverse nonsubsampled shearlet transform,and the enhanced remote sensing image is obtained.The experimental results show that this method has obvious advantages in subjective and objective evaluation of remote sensing image enhancement when compared with some newly proposed image enhancement algorithms.4.Multi-focus image fusion and enhancement algorithm based on nonsubsampled shearlet transform and spatial frequency-motivated parameter-adaptive pulse coupled neural network(SF-PAPCNN).Aiming at the problems of blurring and artifacts in image fusion methods,an image fusion model based on nonsubsampled shearlet transform is proposed.Firstly,the two multi-source images are decomposed by the nonsubsampled shearlet transform,and the corresponding low-frequency and high-frequency components are obtained,respectively;then the SF-PAPCNN model is used to fuse the low-frequency components,and the improved sum-modified-Laplacian(ISML)model is used to fuse the high-frequency components;finally,the low-frequency and high-frequency parts of the fusion are reconstructed by the inverse nonsubsampled shearlet transform,and the final fused image is obtained.The experimental results show that this algorithm can obtain clearer fusion image and more image details in multi-focus image fusion when compared with the classical and newly proposed fusion methods.