5207

4877

كارشناسي ارشد

معماري كامپيوتر

اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر

1388

نه،104ص.:مصور،جدول،نمودار

ص.ع.به فارسي و انگليسي

شادرخ سماوي،محمد داورپناه جزي

3/3/89

1396/09/29

كتابنامه

مهندسي برق و كامپيوتر

ID4877

به فارسي و انگليسي: قابل رويت در نسخه ديجيتالي

Compression of mammographic image using prediction methods Elham Mahmoudzadeh e mahmoudzadeh@ec iut ac ir Jan 18 2010 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 IranDegree M Sc Language FarsiS Samavi Prof Supervisor samavi96@cc iut ac irM Davarpanah Jazi Assist Prof Supervisor mdjazi@cc iut ac irAbstractBreast Cancer is one of the leading causes of death for women An early diagnosis is a requirement forreduction of the mortally rate from this disease Modern imaging technology has created the capability ofearly detection of this One of the most effective methods of early diagnosis is mammography Amammography is a low dose X ray imaging of the breast Every year millions of mammograms aregenerated world wide Due to large size and high quality of the mammogram images their transmissionover computer networks can be difficult and image compression is necessary On the other hand theseimages are usually stored for a long period of time on the patient s digital file to keep track of the treatment Archiving and retaining large amounts of data is expensive and requires compression methods Imagecompression methods are classified into lossy and lossless algorithms Lossless compression allows exactrecovery of the original image and is certainly the obvious choice for the mammograms where no alterationto the original image is tolerable Lossy compression methods do not allow exact recovery of the originaldata after compression although it allows much higher compression ratios Generally higher compressionratios are achieved at the expense of some levels of degradation of the original image Image degradationdue to lossy compression may be undetectable by a human observer In case of mammograms losslesscompression methods are used due to the sensitivity of fine details of these images which are diagnosticallyof great importance Often lossless image compression methods explore the correlation betweenneighboring pixels In the majority of the methods pixels values are first decorrelated and then data isencoded using a variable length approach It is known that prediction based coders are very competitive inlossless and high rate coding We propose a method based on adaptive prediction and context modeling forlossless compression of digital mammographic images First we modified the ALCM coefficients Thecoefficients are modified based on the region of the image that this predictor is to be applied Regiondetection is done by mask values of the five previous neighbors Two other predictors of MED and GAP are

شادرخ سماوي،محمد داورپناه جزي