 |
| Home |
| Tutorials |
| Contact info |
| Tasks |
Recognition of Changes in a
Building
The objective of this section is to show you how you can use the Eigenface method to detect changes on a fixed structure. Before continuing, we should note that the images that were used for were modified digitally and not physically. The next step on our research will be to see how sensitive our algorithm is to physical changes.
First we start by using an image from our training set. We are going to observe how the maximum distances increases as imperfections are added.
Maximum Value = 42542 Minimum Value = 41150
In the next four images we observe at images outside our training set and how the error increases as imperfections are added.
Maximum Value = 43020 Minimum Value = 41976
Maximum Value = 42937 Minimum Value = 41903
The next two images show how small imperfection, on the right hand side of the building, changes our distances.
One imperfection was added
Maximum Value =43032 Minimum Value =42039
Two imperfections were added
Maximum Value = 43082 Minimum Value = 42098Even though the difference is small, the changes can still be noticed. Next we will show how adding imperfections to a known image can be detected. The original image is not shown but you can see where the changes have been made (one of the windows was modified).
Maximum Value = 43172 Minimum Value = 42239
Two windows were modified
Maximum Value = 43175 Minimum Value = 42258A change on top of the previous two changes was made
Maximum Value = 43239 Minimum Value = 42322
Finally the image was considerably altered. Look how the Euclidean distance increases dramatically.
Maximum Value = 44227 Minimum Value = 43049
This concludes our results with digitally modified pictures. The next step is to create an scale model of and observe how well physical changes can be detected.
