Hilditch refinement (implementation I)

For the current pixel , Scan its 8 Neighborhood , If the pixel value of the neighborhood is 255, be b[i]=1(i=0…8), The pixel value is 128( Indicates that the pixel is marked for deletion in the previous cycle ),b[i]=-1, If the pixel value is 0, be b[i]=0.

The following will be based on b[i] The value of the to 6 There are two conditions , If the conditions are met , The pixel value will be marked as GRAY(128).

1. b[0]=1, That is, the current pixel must be the front scenic spot .

2. 1-abs(b1) + 1 – abs(b3) + 1 –abs(b5) + 1 – abs(b7) >= 1, This condition indicates that the current pixel is a boundary point , That is, at least one of the four points in the East, West, north, South b[i]=0.

3. abs(b1)+…+abs(b8)>=2, This condition indicates that the endpoint cannot be deleted , namely p0 Only one point around the point is 1 or -1 The situation of .

4.  Statistics b1 To b8 be equal to 1 The number of , The quantity value must be greater than 1, This condition indicates that the endpoint cannot be deleted .、

5.  Connectivity detection , Use the following formula ： First, according to the current pixel around 3*3 Domain value , Record d[9] Array , If b[i] be equal to 0, be d[i]=0, otherwise d[i]=1, The final calculation d1-d1*d2*d3+d3-d3*d4*d5+d5-d5*d6*d7+d7-d7*d8*d1 Is it 1, by 1 Then the connectivity is satisfied , You can delete .

6. The last condition guarantees that when the contour is 2 Pixels wide , Delete only one side . Statistics sum Value , The duty of 8 When , You can delete .

It may not be easy to understand , Here are a few gif Show the refinement process , Red indicates the points deleted in this round .

Green in the figure below indicates ： This round meets the first condition , But points that do not meet the second condition .

Green in the figure below indicates ： Points that meet the first condition but do not meet the third condition .

Green in the figure below indicates ： Points that meet the first condition but do not meet the fourth condition .

The figure below is green ： Points that meet the first condition but do not meet the fifth condition .

The figure below is green ： Points that meet the first condition but do not meet the sixth condition ：

The sixth condition may not be easy to understand , Width （ Height ） Only 2 Pixel image , If there is no condition six , It's gone .

The first round of the first five conditions , No green .

Sixth , There's a line （ Column ） Rate .