|
Size: 2065
Comment:
|
Size: 2076
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 5: | Line 5: |
| Using images to search similar images in a huge database was a challenge. The result was promising and it had been used by the dotcors happily. The Master's thesis won the top prize from the university. Although the prize itself was just RMB 100, more or less 10 euro :), hey only few got it. | Using images to search similar images in a huge database was a challenge. The result was promising and it had been used by the dotcors happily ever after. The Master's thesis won the top prize from the university. Although the prize itself was just RMB 100, more or less 10 euro :), hey only few got it. |
This work, Content-based Retrieval on a Medical Image Database, was done for my Master's degree. During some other projects in coorperation with the doctors from the 4th Military Medical University, I found that doctors would always try to find similar medical images to support their diagnosis. However, the medial image database was huge, and tranditional way of retrieving a image was searching with the data from patients and their symptoms as keywords. However, these data are often either unrelated, or not a starting point but the end result of diagnosis. What they wanted were simply the images that looked similar to the images they had in hand.
Using images to search similar images in a huge database was a challenge. The result was promising and it had been used by the dotcors happily ever after. The Master's thesis won the top prize from the university. Although the prize itself was just RMB 100, more or less 10 euro :), hey only few got it.
* [blackandwhite.png Retrieval results using the top-left image.] * [color.png Retrieval results using the top-left image.] * [histogram.png The trick was to index the image features in the database. One of the features was the color histogram.]