Please use this identifier to cite or link to this item:
https://ptsldigital.ukm.my/jspui/handle/123456789/513306Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.advisor | Riza Sulaiman, Prof. Dr. Ir. | - |
| dc.contributor.author | Sophia Jamila Zahra (P62461) | - |
| dc.date.accessioned | 2023-10-16T04:35:23Z | - |
| dc.date.available | 2023-10-16T04:35:23Z | - |
| dc.date.issued | 2017-08-08 | - |
| dc.identifier.other | ukmvital:98995 | - |
| dc.identifier.uri | https://ptsldigital.ukm.my/jspui/handle/123456789/513306 | - |
| dc.description | Shape descriptor and leaf image retrieval have emerged as an important part of computer vision and image analysis applications. Numerous number of researchers have done algorithms to generate effective, efficient and steady methods in image processing particularly in shape representation, matching and image retrieval. Based on the extant literatures, existing leaf retrieval methods are insufficient to achieve adequate retrieval rate due to inherent difficulties of the available shape descriptors. In example, they used neighbors point to extract the features such as tangent angle, curvature and contour, while neighbor size has significant impact when noise is present on the image especially when the noise is widespread in the shape of image. Some of them also used multi-scale and scale space technique to reduce the sensitivity but computationally demanding in matching process. The other proposed such as centroid distance, local diameters, and complex coordinates sometimes failed to discriminate leaf shapes with large differences. This study aims to investigate the shape analysis for plant leaf retrieval studies, provide a suitable technique in feature extraction. Then a new similarity metrics is proposed to generate the final descriptor. Finally, a new framework of leaf retrieval is developed. A new image transform, known as harmonic mean transform (HMT) is proposed in this study, for use as a global descriptor method to extract leaf features. By using harmonic mean function, the signal carries the greater important information in signal acquisition. The selected image is extracted from the whole region where all the pixels are considered and accumulated in order to get a set of features. The sum of pixels in the shape are extracted vertically and horizontally then accumulated into sets of vector matrices, each of vector matrix contains set of sum in HMT which is rotated from zero to 180 degree. In addition, existing similarity measurement techniques such as mutual information (MI) is insufficient for achieving adequate results under different image conditions. Thus, a deformation invariant similarity metric known as distance based mutual information (DIMI) is proposed to produce the supportive features to improve the plant leaf retrieval rate. The proposed techniques are tested on standard dataset images sourced from “MPEG-7”,“UCS-SIPI”, “Swedish” and “ImageCLEF” datasets. The evaluation results demonstrate that the new methodology based on proposed techniques outperform the other state of the art techniques tested on standard image datasets using precision-recall, detection rate, and F-measure. Eventually, a new framework for plant leaf retrieval based on the aforementioned techniques is developed. The features are extracted based on comparison of similarity score for each image to a predefined class model. This leaf retrieval process, as a computer vision based technology is greatly accelerating the classification and retrieval rate compared to the existing methods. The accuracy for proposed method have achieved 96.25% where as the existing HSC is 87.31% for MPEG-7 part B dataset. Next, the MAP results on ImageClef dataset achieved 52.03% outperformed MARCH 46.25% correspondingly, and the accuracy proposed results on Swedish dataset performed slightly better 97.44% compared to MARCH”s result 97.33%. This technique can be used in a wide range of image processing applications. The comparison of the results has confirmed that performance of the proposed technique was superior from the other methods under different image conditions.,Certification of Master's/Doctoral Thesis" is not available | - |
| dc.language.iso | eng | - |
| dc.publisher | UKM, Bangi | - |
| dc.relation | Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat | - |
| dc.rights | UKM | - |
| dc.subject | Image analysis | - |
| dc.title | Shape analysis based on harmonic mean transform and similarity metric for plant leaf retrieval | - |
| dc.type | Theses | - |
| dc.format.pages | 168 | - |
| dc.identifier.callno | TA1637.Z334 2017 3 tesis | - |
| dc.identifier.barcode | 003235(2018) | - |
| Appears in Collections: | Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat | |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| ukmvital_98995+SOURCE1+SOURCE1.0.PDF Restricted Access | 551.34 kB | Adobe PDF |  View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.