This week's topic is still more focused on the filters. The one we used, as in Gaussian Filter, is not appropriate to use in our image prosscessing. As the instructor reconmmened us to take a look at the Matlab's built in filters, we try to use the edge detection to apply the filter after with a conference with TA.
Do list :
1) Do more research about the filters
2) Try every filter that make sense with its definition
3) Apply the most correct filter and do more correction
In researching the filters:
1) sample filtering http://lodev.org/cgtutor/filtering.html
2) useful link for k-means courses.washington.edu/matlab2/Lesson_15.html
Notes for K-means :
Basically, K-means does classify the center of different things.
In a 2-dimensional space( 2D coordinates). As each point point has a specific value, what K-means do is to pick up the value and put them into a cluster. For instance, K has a value of 3, then the function will pick up things and separate them into 3 clusters.
For the process of K- means:
1) pick up the center
2) decide each pixel goes to
3) repeat the process until things stop changing
In order to separate things appropriately, a conclusion came to the surface that we need to figure out an appropriate value for the K.
*There is a built-in k-means command in the Matlab.
Here is the list of problems we faced during the lab:
1) we can not determine which filter to use
2) cannot figure out the value for the filters
Here is the list of good things we had:
1) the "sobel" filter gives us some confidence of figuring out the filter
No comments:
Post a Comment