Connected Components and Feature Detection

Here is the link to the initial notes which was basis on making this post and its executions:

https://drive.google.com/open?id=1txqXeQAygdaAXYIaxN0NMhxS930D5irdNl46Eazh1Ak

And here is the link for a discussion on a way of implementing connected components:

https://drive.google.com/open?id=1ruz_FDU1Aluk7weQX-ZmcgWZuPi3TggSvsLe51AP0-A

Connected Components and Feature Detection are two process needed in order to identify an object in a picture.

In this exercise, we are given an image with the goal of being able to match similar figures of the image to each other through its features.

Initially at the preprocessing stage I used another picture to sample out some techniques needed to be able to identify an object in an image.

These are the original images:

 testt_d_1

eight-thicknesseight-thickness_a


This is an image of a number 8 where the holes inside where filled and in order to get convex hull of the object. The convex hull is the shape most closely related to the object. It is used in feature detection.


 

In here the outline of the image is taken to get the perimeter of the object, supposedly only a letter was to be taken but due to the lack of implementation of Connected Components at the time the whole phrase was taken.

a_perimeter

 

 

 

In the next images techniques for getting a more whole binary image were applied. The techniques are opening, eroding, closing, dilation, and then the normal image, respectively. The normal image is not a technique but just the image after binarization.

a_oa_e
a_ca_da

And then to identify the object exactly, the boundary of the object is taken and drawn as a box to get the highest and lowest of x and y values the object has.

a_dummylabel

 

 

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s