Halcon knowledge: regional topics [07]

One 、 summary

This article and halcon Area of ： Multiple areas （Region） features （6）​​​​​​​ Belong to the same blog , In this article, we will continue with the operators of matrices , Including moment features . Hamming distance, etc .

Two 、 Domain operator

2.1 regiongrowing Is an algorithm for merging regions

regiongrowing： Divide the image into regions of the same intensity --- The rasterization size is Row * Column The rectangular . To determine whether two adjacent rectangles belong to the same region , Only use the gray value of their center point . If the gray value difference is less than or equal to Tolerance, Then the rectangle is merged into a region .

If g_{1} and g_{2} Are the two gray values to be checked , Then merge them into the same region in the following cases ：

1)    | g_{1} - g_{2} | < torlerance

2)    | g_{1} - g_{2} | < torlerance &  | g_{1} - g_{2} | < 127

3)    255 - | g_{1} - g_{2} | < torlerance  & 127 < | g_{1} - g_{2} | 

2.2 expand_region Fill the area gap

expand_region Fill the gap between the input fields , This is due to the suppression of small regions in the segmentation operator , for example ,（ Pattern “ Images ”） Or separate overlapping areas （ Pattern “ Area ”）. Both of these uses stem from the expansion of the region . The operator adds or removes a pixel wide... To an area “ strip ” Come to work .

Extension occurs only when specified as non “ prohibit ” Region （ Parameters ForbiddenArea）. The number of iterations is determined by the parameter Iterations determine . By passing “ Maximum ”,expand_region Iterate until it converges , namely , Until it doesn't change . By passing 0, All non overlapping areas will be returned . Two modes of operation （“ Images ” and “ Area ”） There are differences in the following aspects ：

'Image'
         The input area is expanded iteratively , Until they touch another area or image boundary . under these circumstances , The image boundary is defined from (0,0) To (row_max,col_max) The rectangular . here ,(row_max,col_max) Corresponds to all input fields （ That is to say Regions and ForbiddenArea All areas passed in ） The bottom right corner of the smallest surrounding rectangle . because expand_region Process all areas at the same time , So the gap between regions is evenly distributed to all regions . The overlapping region is divided by evenly distributing the overlapping region to two regions .

'region'
         Do not extend the input area . contrary , Only the overlapping region is divided by evenly distributing the overlapping region to each region . Because the intersection with the original area is calculated after the shrink operation gap of the output area , That is, the segmentation is incomplete . This can be done by a second call to expand_region To prevent , Take the complement of the original region as “ Forbidden area ”.

2.3  get_region_index

   function ： Index of all areas including a given pixel . Is to give a coordinate , Which area does this coordinate fall into .

• get_region_index(Regions : : Row, Column : Index)

2.4 get  region_thickness

   function ： Check the width of the area near the spindle （ The thickness of the ）.

• get_region_thickness(Region : : : Thickness, Histogramm)

         Operator get_region_thickness Calculate each pixel in the section along the main axis （ See elliptic_axis） Area thickness of . The thickness at a certain point on the spindle is defined as the distance between the contour and the intersection of the vertical on the spindle at the farthest point on the corresponding point . Besides , Operator get_region_thickness Returns a histogram of area thickness . The length of the histogram corresponds to the maximum thickness that appears in the viewing area .

read_image(Image, 'fabrik')
get_image_size(Image, Width, Height)

auto_threshold(Image, Regions, 3 )

dev_open_window(0, 0, Width, Height, 'black', WindowHandle)

select_obj(Regions, RegSelect, 1)
get_region_thickness(RegSelect,Thickness, Histogramm)

         Only one area can be passed . If there are multiple connection components in this area , Only the first . All other components are ignored .

Such as in the figure above RegSelect It's like this . And obtained Thickness Just the first one alone region Enumeration of scan line widths of blocks . 

2.5  hamming_distance

• hamming_distance(Regions1, Regions2 : : : Distance,

   function ： The Han Ming distance between the two regions .

operator hamming_distance Returns the Hamming distance between two regions , That is, the number of pixels in different regions （ distance ）, That is, the number of pixels that one region contains but the other region does not ：

 Distance： distance

 Similarity: Similarity coefficient

)

  2.6 hamming_distance_norm

• hamming_distance_norm(Regions1, Regions2 : : Norm : Distance, Similarity)

function ： Normalized Hamming distance between two regions .

         Operator hamming_distance_norm Returns the Hamming distance between two regions , That is, the number of pixels in different regions （ distance ）. Before calculating the difference ,Regions1 The region in is normalized to Regions2 Area in . The result is the number of pixels contained in one region but not in another ：

2.7 hamming_change_region 

hamming_change_region (Holes, OutputRegion, Width, Height, Error)

hamming_change_region Changed by Width and Height The region of the upper left part of a given image , The Hamming distance between the generated region and the input region is Distance. This is done by adding or removing distance points from the input area .

* Remove noise from a region
* 
read_image (Caltab, 'caltab')
get_image_size (Caltab, Width, Height)
threshold (Caltab, Region, 0, 120)
connection (Region, ConnectedRegions)
select_shape (ConnectedRegions, SelectedRegions1, 'convexity', 'and', 0.9, 1)
select_shape (SelectedRegions1, SelectedRegions2, 'area', 'and', 200, 1000)
union1 (SelectedRegions2, Holes)
for Error := 1 to 10000 by 1000
    * Change the region with the 'Error' hamming distance
    hamming_change_region (Holes, OutputRegion, Width, Height, Error)
    dev_clear_window ()
    dev_set_color ('red')
    dev_display (OutputRegion)
    hamming_distance (Holes, OutputRegion, Distance1, Similarity1)
    * Remove noise
    remove_noise_region (OutputRegion, NoiseRed, 'n_4')
    hamming_distance (Holes, NoiseRed, Distance2, Similarity2)
    dev_set_color ('green')
    dev_display (NoiseRed)
    stop ()
endfor

2.8 inner_circle

   function ： The largest circle inside an area .

• inner_circle(Regions : : : Row, Column, Radius)

Get the inner maximum inscribed circle of the area .

2.9 inner_rectangle1

   function ： The largest rectangle inside an area .

• inner_rectangle1(Regions : : : Row1, Column1, Row2, Column2)

The largest rectangle inside the area , The rectangular edge is parallel to the window . 

2.10 moments_region_2nd

   function ： Moment properties of the region .

• moments_region_2nd(Regions : : : M11, M20, M02, Ia, Ib)

        moment_region_2nd Calculation Regions Enter the geometric moment of the region in M11、M20 and M02. Besides , The major axis and minor axis of the input field are in Ia and Ib Calculates and returns . The covariance matrix is given by ：

         2 The order row correlation moment is at M20 Back in ,2 The order correlation moment is M02 Back in . Moment M11 Represents the covariance between the row and column coordinates of the region point .

r0 and c0 It's a region R The center of gravity coordinates of . Then the moment is defined as ：

2.11 moments_region_2nd_invar

   function ： A moment characteristic of a region .

• moments_region_2nd_invar(Regions : : : M11, M20, M02)

Operator moment_region_2nd_invar Calculate the scaling moment of the axis (M20, M02) And passing parallel to the coordinate axis (M11) The inertia coefficient of the axis at the center of .

Z0 and S0 It's a region R The center and area of F Coordinates of .Mij Then the moment is defined as ：

  among Z and S Is the coordinates of all points in the region .

2.12 moments_region_2nd_rel_invar

   function ： Calculate the parameters of the correlation moment .

• moments_region_2nd_rel_invar(Regions : : : PHI1, PHI2)

Operator moment_region_2nd_rel_invar Calculate the relative moment of scaling (PHI1, PHI2).

Moment PHI1 and PHI2 Defined as ：

  If more than one region is passed , The results are stored in tuples , The index of the value in the tuple corresponds to the index of the region in the input .

In the case of blank areas , If no other behavior is set , Then the values of all parameters are 0.0（ See set_system）.