In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written

more »

... permission. Abstract Image enlargement is a vital process for printing images on large format printers. Image expansion introduces many new pixels to the original image. Proper values for these new pixels should be found otherwise distortion and significant degradation in the quality of the image will result. Several methods have been employed to minimize such distortion. The most popular of these methods are the pixel replication and the linear interpolation methods. This is because these methods are not computationally demanding. In this work, we develop a new image expansion method that preserves the quality of the edges in the expanded image, where other methods fail. First, the edges in the original image are extracted. For binary images we use Laplacian operator and for gray-level images we use Canny edge detector. We propose a new algorithm, which replicates the edge shape so that the edges in the expanded image do not appear zigzagged. Our algorithm expands the edges while it preserves the edge information such as the shape and the continuity of the edge. The edge-shape-replication algorithm simply, classifies the edges into two classes: short post-CCC and long post-CCC edges, each is expanded differently. Most of the original pixels keep their values after the expansion. However, the remaining pixels' values are calculated using one of three proposed techniques based on the location of the pixel whose value is to be calculated. The three locations of these pixels are pixels that lie on an edge, pixels that are adjacent to edges, and the remaining pixels areas. To compare performance of our proposed method with others, we reduced the test images to their quarter size then expanded them to their original sizes. The sum of absolute errors (SAD's) between the original image and the expanded images are calculated. Our proposed method is shown to outperform the replication, linear and cubic interpolation methods. -iii -