Content-based image retrieval of blood cells based on colour / Alagu Rathina Bairavi
The research in Content-based image retrieval is developing rapidly. It benefits many other fields, in particular the medical field as the need of having a better way of managing and retrieving digital images has increased. The aim of this thesis is to create an image retrieval system that will ass...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2009
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/5788/ http://studentsrepo.um.edu.my/5788/1/CBIR.pdf |
| _version_ | 1848772983436345344 |
|---|---|
| author | Bairavi, Alagu Rathina |
| author_facet | Bairavi, Alagu Rathina |
| author_sort | Bairavi, Alagu Rathina |
| building | UM Research Repository |
| collection | Online Access |
| description | The research in Content-based image retrieval is developing rapidly. It benefits many other fields, in particular the medical field as the need of having a better way of managing and retrieving digital images has increased.
The aim of this thesis is to create an image retrieval system that will assist hematologist to detect and classify unknown blood cell images speedily and accurately. The algorithm in this study is simple and fast. The problem that arises in this study was difficulty in differentiating one image from the other, as the level of homogeneity of the blood cells is quite high.
The method used in this study was segmentation of blood cell images that was performed based on threshold value automatically selected from a colour histogram. The segmented image is then subtracted with the blood cell images, which are already manually segmented and stored in the database. Images, that has less than 10% difference between automated threshold and manually selected threshold value are retrieved and displayed. Among the 100 images tested, only four images could not be segmented due to the appearance of the images. The number of images that has a difference of less than 10% between manually selected threshold value and automated threshold value are 91 images.
This study of image retrieval of blood cells based on colour generated good results for further prospective study. Additions of more features such as shape, texture, and spatial location will be required to further enhance accuracy of image retrieval of blood cells in the future. |
| first_indexed | 2025-11-14T13:35:11Z |
| format | Thesis |
| id | um-5788 |
| institution | University Malaya |
| institution_category | Local University |
| last_indexed | 2025-11-14T13:35:11Z |
| publishDate | 2009 |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | um-57882015-06-27T09:19:16Z Content-based image retrieval of blood cells based on colour / Alagu Rathina Bairavi Bairavi, Alagu Rathina QA76 Computer software The research in Content-based image retrieval is developing rapidly. It benefits many other fields, in particular the medical field as the need of having a better way of managing and retrieving digital images has increased. The aim of this thesis is to create an image retrieval system that will assist hematologist to detect and classify unknown blood cell images speedily and accurately. The algorithm in this study is simple and fast. The problem that arises in this study was difficulty in differentiating one image from the other, as the level of homogeneity of the blood cells is quite high. The method used in this study was segmentation of blood cell images that was performed based on threshold value automatically selected from a colour histogram. The segmented image is then subtracted with the blood cell images, which are already manually segmented and stored in the database. Images, that has less than 10% difference between automated threshold and manually selected threshold value are retrieved and displayed. Among the 100 images tested, only four images could not be segmented due to the appearance of the images. The number of images that has a difference of less than 10% between manually selected threshold value and automated threshold value are 91 images. This study of image retrieval of blood cells based on colour generated good results for further prospective study. Additions of more features such as shape, texture, and spatial location will be required to further enhance accuracy of image retrieval of blood cells in the future. 2009 Thesis NonPeerReviewed application/pdf http://studentsrepo.um.edu.my/5788/1/CBIR.pdf Bairavi, Alagu Rathina (2009) Content-based image retrieval of blood cells based on colour / Alagu Rathina Bairavi. Masters thesis, University of Malaya. http://studentsrepo.um.edu.my/5788/ |
| spellingShingle | QA76 Computer software Bairavi, Alagu Rathina Content-based image retrieval of blood cells based on colour / Alagu Rathina Bairavi |
| title | Content-based image retrieval of blood cells based on colour / Alagu Rathina Bairavi |
| title_full | Content-based image retrieval of blood cells based on colour / Alagu Rathina Bairavi |
| title_fullStr | Content-based image retrieval of blood cells based on colour / Alagu Rathina Bairavi |
| title_full_unstemmed | Content-based image retrieval of blood cells based on colour / Alagu Rathina Bairavi |
| title_short | Content-based image retrieval of blood cells based on colour / Alagu Rathina Bairavi |
| title_sort | content-based image retrieval of blood cells based on colour / alagu rathina bairavi |
| topic | QA76 Computer software |
| url | http://studentsrepo.um.edu.my/5788/ http://studentsrepo.um.edu.my/5788/1/CBIR.pdf |