Research And Application Of Image Enhancement Method Based On Visual Attention Mechanism

As important information source of modern medical diagnosis,medical images make clinical diagnosis and treatment more efficient and accurate.Medical images play an irreplaceable role in the field of medicine.However,in the process of medical image formation,due to the influence of various adverse factors,medical images will appear unclear,low contrast and other issues.Low-quality medical images will interfere with the doctor's judgment of the condition,so the research of medical image enhancement methods becomes very important.Based on the principle of human visual attention mechanism,the region of interest of an image can be quickly found.This paper focuses on the research of medical image enhancement methods,and proposes an image enhancement method based on visual attention mechanism.This method is mainly divided into three parts: image salient region extraction,image enhancement and image quality evaluation.This paper mainly completes the following work:Firstly,the visual attention mechanism model based on feature integration and global contrast are studied and analyzed respectively.The advantages and disadvantages of the two models are analyzed according to the salient region extraction results of medical images.Secondly,because of the low contrast of medical images,a visual attention mechanism model based on the fusion of contrast features is proposed,this model introduces the image contrast features into the visual attention mechanism model based on feature integration.When calculating contrast saliency map,SLIC super-pixel segmentation algorithm is used to segment the original medical image,compared with other segmentation methods,this algorithm can effectively protect the edge texture information of the image.Then the contrast saliency value of each image area is calculated separately.Finally,the contrast saliency map of the medical image is obtained.The final saliency map of the image is obtained by fusing the contrast saliency map,brightness saliency map and directional saliency map of the medical image.Compared with the saliency map obtained by the existing visual attention mechanism model,the validity of the proposed visual attention mechanism model is verified.Thirdly,in light of the characteristics of medical images and the shortcomings oftraditional image enhancement methods,an image enhancement method based on the combination of visual attention mechanism and human visual perception characteristics was proposed.In the acquisition of the saliency map of the original medical image,an information histogram is constructed,different from the traditional histogram,the histogram can reflect the information of the medical image comprehensively.Because of the large amplitude of the information histogram,the dynamic range of the information histogram is optimized and adjusted according to the brightness threshold curve of the human eye,so as to enhance the medical image.The proposed image enhancement method and the existing image enhancement method are used to enhance the medical image,and a medical image quality evaluation algorithm considering both the characteristics of medical image and human eye perception is adopted to analyze the results of medical image enhancement.The results show that the image enhancement algorithm proposed in this paper is the best for medical image enhancement.Fourthly,The proposed image enhancement algorithm is programmed and realized by using MATLAB language.The medical image enhancement system is designed,the functions of user login,image salient area recognition and medical image enhancement are realized.