Vessel Segmentation Based On Direction Field And Morphological Transform

The application of image processing technology in medicine has been a century since 1895,the German scholar William Conrad Rontgen found the first X ray images.Image processing technology is playing a unique role in every aspects of medical.With the rapid development of modern computer technology and digital processing technology,more advanced technology is used in medical image processing,the speed and quality of medical image processing have been greatly improved,and some more methods in other fields also get into medical image processing field because of the convenience by computing and communication.That greatly accelerated the results of inquiry and progress in this traditional area.One of the traditional problems in this field is the segmentation of retinal blood vessels.The retinal image is image that directly taken and deep in microvascular.The image e is usually poor with noise,interference,and has fuzzy pixel boundary.Those factors sometimes lead to a hard auto segmentation of vascular retinal image and it is difficult to achieve deep precision.Therefore,it is of great significance to analyze and quantify retinal fundus images based on objective conditions by using modern image processing theory and computer digital technology.In order to further explore new ideas and methods of vascular segmentation,the orientation field and morphological transformation and the two methods of comprehensive application in segmentation of retinal image vessel was studied,the main research work includes:the image strengthening method of adaptive based on image histogram equalization;the image processing method based on morphological transform and the image reconstruction method based on gray gradient direction field detection,and the vascular segmentation implementation process of comprehensive orientation field and morphological transformation based on these studies.The result of these study is verified in Matlab and experimental programming object based on the DRIVE retinal database,firstly through image analysis and histogram equalization based on the enhanced image of small retinal image,and then through the special with the direction of the structure element to achieve the extraction of vascular skeleton,finally through the growth method to obtain the segmentation result.The experimental results show that the integrated use of direction field theory and morphological transformations in the retinal image segmentation in this field is a new idea,its theoretical basis and reliable,achieve good segmentation results via the convenient.