Matlab superpixels function (2D super pixel over segmentation of image)

grammar

[L,NumLabels] = superpixels(A,N)
[L,NumLabels] = superpixels(A,N,Name,Value)

explain

[L,NumLabels] = superpixels(A,N) 
% Calculate two-dimensional grayscale or  RGB  Images  A  Super pixels （ Expected number of super pixels ）.
%N  Specify the number of super pixels to create .
% This function returns  L(double  Label matrix of type ) and  NumLabels( The actual number of hyperpixels calculated ).

[L,NumLabels] = superpixels(A,N,Name,Value) 
% Use the name used to control various aspects of segmentation - Value pair group parameter calculation image  A  Super pixels .

superpixels The function uses simple linear iterative clustering (SLIC) Algorithm . This algorithm divides pixels into regions with similar values . In image processing operations （ Such as segmentation ） Using these regions in can reduce the complexity of these operations .

Example

Calculate input RGB Super pixel of image

% Read the image into the workspace 
A = imread('kobi.png');

% Calculate the super pixels of the image 
[L,N] = superpixels(A,500);

% Superimpose and display the boundary of super pixels on the original image 
figure(1)
BW = boundarymask(L);
imshow(imoverlay(A,BW,'cyan'),'InitialMagnification',67)

% Set the color of each pixel in the output image to  RGB  Mean color 
outputImage = zeros(size(A),'like',A);
idx = label2idx(L);
numRows = size(A,1);
numCols = size(A,2);
for labelVal = 1:N
    redIdx = idx{labelVal};
    greenIdx = idx{labelVal}+numRows*numCols;
    blueIdx = idx{labelVal}+2*numRows*numCols;
    outputImage(redIdx) = mean(A(redIdx));
    outputImage(greenIdx) = mean(A(greenIdx));
    outputImage(blueIdx) = mean(A(blueIdx));
end    
figure(2)
imshow(outputImage,'InitialMagnification',67)

Input parameters

A ： The image to be segmented
The image to be segmented , Specify as a two-dimensional grayscale image or a two-dimensional true color image . about int16 data ,A Must be a grayscale image . When parameters isInputLab yes true when , The input image must be of data type single or double.
A Data type of ：single | double | int16 | uint8 | uint16

N： Expected number of super pixels
The number of super pixels required , Specify as a positive integer .
N Data type of ： single | double | int8 | int16 | int32 | int64 | uint8 | uint16 | uint32 | uint64

name - Value pair group parameter
Specify optional 、 Separated by commas Name,Value For group parameters .Name For parameter name ,Value Is the corresponding value .Name Must be in quotation marks . You can specify multiple names in any order - Value pair group parameter , Such as Name1,Value1,…,NameN,ValueN.
such as ：

B = superpixels(A,100,'NumIterations', 20); 
% The image A Divided into 100 Two super pixels , The number of iterations is 20 Time 

common name Parameters ：
（1）‘Compactness’
Super pixel shape , Specify as numeric scalar .SLIC The compactness parameter of the algorithm controls the shape of super pixels . The higher the value , The more regular the shape of super pixels , That is, the closer to the square . The lower the value , Super pixels will better fit the boundary , Make its shape irregular . The allowable range is (0 Inf). The value of compactness is usually [1,20] Within the scope of . The default value is 10.
Compactness Corresponding value (value) Data type of ： single | double | int8 | int16 | int32 | int64 | uint8 | uint16 | uint32 | uint64

（2）‘IsInputLab’
Input image data in Lab* Color space , Designated as true or false. The default is false.
'IsInputLab’ Corresponding value (value) Data type of ： single | double | int8 | int16 | int32 | int64 | uint8 | uint16 | uint32 | uint64 | logical

（3）‘Method’
Algorithm for calculating super pixels , Specify one of the following values .superpixels The function uses simple linear iterative clustering (SLIC) Two variants of the algorithm .

value	significance
‘slic0’	superpixels Use after the first iteration SLIC0 The algorithm is optimized in an adaptive way ‘Compactness’. This is the default setting .
‘slic’	‘Compactness’ Remain unchanged in the clustering process .

'Method’ Corresponding value (value) Data type of ： char | string

（4）‘NumIterations’
The algorithm is used in the clustering stage The number of iterations , Specify as a positive integer （ The default is 10）. For most problems , There is no need to adjust this parameter .
'NumIterations’ Corresponding value (value) Data type of ：single | double | int8 | int16 | int32 | int64 | uint8 | uint16 | uint32 | uint64

Output parameters

L： Label matrix , Returns as an array of positive integers . value 1 Represents the first super-pixel region in the image ,2 Indicates the second area , And so on .
data type ： double

NumLabels： Calculated The number of superpixels , Returns as a positive integer .
data type ： double