|Title:||Modified edge-directed interpolation|
|Subject:||Hong Kong Polytechnic University -- Dissertations|
Image processing -- Mathematical models
Image processing -- Digital techniques
|Department:||Department of Electronic and Information Engineering|
|Pages:||xiv, 100 leaves : ill. (some col.) ; 31 cm|
|Abstract:||Image interpolation is the most fundamental process of image processing, such as resolution enhancement and image restoration. It involves the estimation of unknown pixels from the neighboring known pixels. The interpolation depends on the prior knowledge of the original image and no new information is added by interpolation. Image interpolation can be categorized into non-adaptive interpolation and adaptive interpolation. The non-adaptive interpolation methods, such as nearest neighbor interpolation, bilinear interpolation, bicubic interpolation, e.t.c., estimate the unknown pixels by averaging the pixel values of the neighboring known pixels. The quality of the interpolated image depends on the interpolation methods used. The visual quality of the interpolated image increases with an increase in the number of known pixels being applied in the interpolation. However, those non-adaptive interpolation methods do not place any special considerations to the image edges, as a result, the averaging operation may smooth out the abrupt changes of the image edges. Hence, degradation of the visual quality of the interpolated images, especially in the area around the image edges and detailed features, are observed. Adaptive interpolation methods adapt the interpolation to the image features. Among a variety of adaptive interpolation methods, edge-directed approach is known to be more effective. Edge-directed approach interpolation, such as edge-directed interpolation proposed by , use edge map (formed by analyzing the pixel intensities) to guide the interpolation process. However, it is difficult to accurately locate image edges. As a result, the improvement in the visual quality of the interpolated images obtained by edge-directed interpolation are limited by the accuracy of the edge map. The edge detection accuracy problem in edge-directed interpolation method is remedied by the application of covariance-based edge-adaptive interpolation, such as that proposed in the new edge-directed interpolation (NEDI) method . The covariance structure reveals the edge information of the image. As a result, the NEDI adapts itself by studying the local covariance structure of an image. The unknown pixel is estimated by linear interpolation from the known pixels in the same local block with the locally stationary assumption. The weighting factors is optimized by linear prediction and solved by Weiner filtering. The interpolated image is observed to be able to preserve edge sharpness and continuity. However, the NEDI suffers from the prediction error and prediction error accumulation problems, which degrades the visual quality of the interpolated image. This dissertation depicts two novel interpolation methods, which effectively reduce the adverse effect of prediction error and prediction error accumulation problems. The two methods are entitled as Modified Edge-Directed Interpolation (MEDI) and Enhanced Edge-Directed Interpolation (EMEDI). The improvement of visual quality of the interpolated image obtained by the MEDI and the EMEDI is achieved by using merely the known pixels throughout the interpolation process. The EMEDI is an enhancement on the MEDI which extends the statistical study in all possible directions, thus enhances the visual quality of interpolated image. Both of the methods are proved to be effective through simulations, which showed that they are able to preserve edge and detailed features in the interpolated image. The new methods are further shown to be applicable in other industrial applications, such as resolution enhancement and demosaicing of color CCD samples, where simulation results are presented.|
|Rights:||All rights reserved|
Files in This Item:
|b20809049.pdf||For All Users (off-campus access for PolyU Staff & Students only)||24.84 MB||Adobe PDF||View/Open|
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item: