Edge Detection Of COVID-19 CT Image Based On Improved Canny And Morphological Algorithm

At the end of 2019,novel coronavirus pneumonia spread to all parts of the world at an extremely fast speed,causing great harm and threat to human health.As a commonly used imaging technique in clinical practice,CT can provide support for clinical diagnosis by reconstructing the image of the lesion site.Chest CT plain scan is the main screening and auxiliary diagnostic means of COVID-19 pneumonia,which plays an important role in virus screening.Through diagnosis,effective treatment for patients with early COVID-19 can prevent them from developing into sever e pneumonia.The edges of an image contain most of the information,and edge recognition,as a key step in the field of image processing,has a significant impact on feature description and matching of images.Extracting the weak edge of COVID-19 CT image to highlight the lung infection area is helpful for medical experts to analyze and judge COVID-19 virus.This paper mainly studies two edge detection algorithms,Canny and mathematical morphology.On the basis of the traditional algorithm,the two algorithms are respectively improved.The specific improvement work is as follows:(1)Improvements to traditional Canny edge detection algorithm:1Using a hybrid filter composed of adaptive median filtering and guided filtering for image denoising;2When calculat ing the image gradient,the second-order finite difference method is replaced by the 4-direction Sobel operator;3The adaptive linear interpolation method is used for non-maximum suppression.(2)Improvement of mathematical morphology edge detection algori thm:1A new edge detection operator is proposed based on basic morphological operations,and structural elements in the directions of 0°、45°、90°、135°and are selected to perform edge detection on the image.The detection results are fused using an even weight fusion method for edge fusion;2The structure elements with different sizes are selected for multi-scale edge detection of the image,and the detection results are fused by combining entity weighting and information entropy;3The image edges obtained in the first two steps are fused by combining entity weighting and information entropy to obtain the final morphological edge detection results.Finally,we apply the two improved image edge detection techniques to COVID-19 pneumonia CT images respectively,and compare them with other algorithms.The results show that the two improved algorithms can detect clearer image edges.