In this lesson we will explore a non standard technique for image segmentation. It is based on the toggle concept. The toggle is defined as:
For this particular experiment we use a slight modification of this toggle, where the resultant image is binary:
The idea of this segmentation process is to estimate the foreground and the background of the image. The segmented image, the pixels will have value 1 if they are nearer to the correspondent foreground estimative and 0 if their are neared to the background estimative.
To estimate the foreground and background, we use closing and opening respectively, with a proper size and shape for structure element.
Below, we show the original grey-level image.
|
|---|
The foreground and background estimative are shown below. It was used a disk of diameter 12 for the foreground (closing) and diameter 35 for the background (opening).
|
|---|
The original toggle is shown at left and the toggle with binary output is show right which is the result of the segmentation.
|
|
|---|---|
| a) | b) |