الفهرس | Only 14 pages are availabe for public view |
Abstract The first part of the research proposes a new easy to use and light in weight system is used to measure the signal eye from the outer canthus, also an easy and efficient algorithm is proposed to distinguish between right and left eye movements. The proposed system is then used in multiple applications to help the disabled to depend on himself and use the computer to read books and browse. The second part of the thesis is proposing an algorithm is used to align the fundus images with different modalities based on retinal invariant features extraction, where for each retina there are footprint and do not change. These features are the angles between the branching of capillaries as well as the three ramifications connected to each three-point fork, then extract all properties that bound together like the distances between the original and the three branching related as well as the angles between each branching point. Every pixel in each fundus image is scanned for these features in order to extract all of the required matching points in one matrix. Features of the images are then compared in two stages to choose the best symmetrical points between the two images, which are sufficient for spacial alignment between the two images. All of these steps are should be done after the color images are converted to black and white image with the extraction of retinal blood vessels as segmentation step. Segmentation is also done based on proposed easy to build and efficient based on matched filters. Finally, the results presented for the proposed solutions and compared with other previous work. |