Segmentation

cv

1 Two Segmentation Methods

• Pixel based methods
• Grouping/clustering based methods

2 Pixel based methods

2.1 point detection

2.2 line detection

2.3 Hough transformation

Elegant method for direct object recognition

• Edges need not be connected
• Complete object need not be visible
• Key Idea: Edges VOTE for the possible model

2.4 Histogram

Four basic image types: dark, light, low contrast, high contrast.
Types of processing:

• (1)Hist. equalization;
  
• (2)Hist. matching (specification);
  
• (3)Local enhancement
• (4)Histogram based segmentation
  
  

3 Grouping/clustering based methods

3.1 K-means

Algorithm step:

Good and Bad：

3.2 Mean shift

Algorithm step:

Good and Bad：

3.3 Graph-based Segmentation

• Node for every pixel
• Edge between every pair of pixels
• Each edge is weighted by the affinity or similarity of the two nodes

3.4 Background Subtraction