5393

5053

كارشناسي ارشد

مخابرات

اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر

1389

[نه]،91ص.:مصور،جدول،نمودار

ص.ع.به فارسي و انگليسي

سعيد صدري

مهدي مهدوي

31/5/89

مهندسي برق و كامپيوتر

ID5053

به فارسي و انگليسي: قابل رويت در نسخه ديجيتالي

AbstractImage alignment and stitching are among the oldest and most important applications ofmachine vision Image alignment is used in every video camera with image stabilizationproperty Stitching algorithms which have made the creating of mosaic images are used inconstructing digital maps and stitching satellite pictures In constructing panoramas whichrecently has found important applications in virtual environment advertising and so on acombination of these algorithms is used Meanwhile each particular application is defined under special conditions while there areefforts to manage these applications automatically and without any preconditions In the earlyattempts for automatic image alignment and stitching these tasks have been based on thepixel intensity or color Hence every pixel of images is processed and then the transformationbetween images is derived Because of a huge number of variables pixels and blind searchon the transformation parameters these algorithms have high order of complexity whichleads to expensive memory and time cost In addition since these algorithms rely on the pixelintensity they are sensitive to illumination view rotation and zooming More recent approaches implement object recognition methods that eliminate the need forcomplete process of the entire images Image alignment is achieved by focusing on interestregions instead This leads to optimization of time and memory In addition theses methodsare insensitive to illumination change and any other transformation that may happen to theimages In this thesis we introduce image alignment and stitching and review some existingalgorithms We describe how object recognition approach can be applied in this procedure Furthermore we suggest two new algorithms for image alignment and stitching that improvesome stages of the algorithms and evaluate them

سعيد صدري

مهدي مهدوي