Use this URL to cite or link to this record in EThOS:
Title: Novel computational technique for determining depth using the Bees Algorithm and blind image deconvolution
Author: Yuce, Baris
ISNI:       0000 0004 2735 0852
Awarding Body: Cardiff University
Current Institution: Cardiff University
Date of Award: 2012
Availability of Full Text:
Access from EThOS:
Access from Institution:
In the past decade the Scanning Electron Microscope (SEM) has taken on a significant role in the micro-nano imaging field. A number of researchers have been developing computational techniques for determining depth from SEM images. Depth from Automatic Focusing (DFAF) is one of the most popular depth computation techniques used for SEM. However, images captured with SEM may be distorted and suffer from problems of misalignment due to internal and external factors such as interaction between electron beam and surface of sample, lens aberrations, environmental noise and artefacts on the sample. Distortion and misalignment cause computational errors in the depth determination process. Image correction is required to reduce those errors. In this study the proposed image correction procedure is based on Phase Correlation and Log-Polar Transformation (PCLPT), which has been extensively used as a preprocessing stage for many image processing operations. The computation process of PCLPT covers the pixel level interpolation process but it cannot deal with sub-pixel level interpolation errors. Hence, an image filtering stage is necessary to reduce the error. This enhanced PCLPT was also utilised as a pre-processing step for DFAF which is the first contribution of this research. Although DFAF is a simple technique, it was found that the computation involved becomes more complex with image correction. Thus, the priority to develop a less complicated and more robust depth computation technique for SEM is needed. This study proposes an optimised Blind Image Deconvolution BID) technique using the Bees Algorithm for determining depth. The Bees Algorithm (BA) is a swarm-based optimisation technique which mimics the foraging behaviour of honey bees. The algorithm combines exploitative neighbourhood search with explorative global search to enable effective location of the globally optimal solution to a problem. The BA has been applied to several optimisation problems including mechanical design, job shop scheduling and robot path planning. Due to its promise as an effective global optimisation tool,the BA has been chosen for this work. The second contribution of the research consists of two improvements which have been implemented to enhance the BA. The first improvement focuses on an adaptive approach to neighbourhood size changes. The second consists of two main steps. The first step is to define a measurement technique to determine the direction along which promising solutions can be found. This is based on the steepness angle mimicking the direction along which a scout bee performs its figure-of-eight waggle dance during the recruitment of forager bees. The second step is to develop a hybrid algorithm combining BA and a Hill Climbing Algorithm (HCA) based on the threshold value of the steepness angle. The final contribution of this study is to develop a novel technique based on the BA for optimising the blurriness parameter with BID for determining depth. The techniques proposed in this study have enabled depth information in SEM images to be determined with 68.23 % average accuracy.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: TJ Mechanical engineering and machinery