ok. well it definately seems that filtering is a must before you actually do edge or probably any other type of detection. Indeed, notice how if a noisy image - say, especially one with just gaussian noise (which I added to image 5.6a) - must be filtered in order for edge detection to work. figure 5-7 illustrate what NOT doing filter will cause.
And yes this time with my improved matlab skills I used the embedded matlab commands to come up with the images below. Though understanding when to use uint8 and double and when to scale images is still confusing.
(Naq vg'f yngr naq V'z tehzcl naq zl tvey sevraq gubhtug V jnf vtabevat ure gbqnl!)
Here is my diary of what i did: "diary"
 5.4a |
5.4b |
 5.4c |
 |
5.4d |
 |
 5.4e |
 |
5.4f |
 |
5.4g |
 |
| Figure 5.4 ends
|
|
| 5.5a |
|
5.5b |
 |
5.5c |
 |
5.5d |
 |
5.5e |
 |
5.5f |
 |
| Figure 5.5 ends
|
|
5.6a |
5.6b |
5.6c |
 |
5.6d |
 |
5.6e |
 |
5.6f |
 |
5.6g |
 |
| Figure 5.6 ends
|
|
| 5.7a |
|
5.7b |
 |
5.7c |
 |
5.7d |
 |
5.7e |
 |
5.7f |
 |
| Figure 5.7 ends
|