function stats = grayrlprops(varargin)
%GRAYCOPROPS Properties of gray-level run-length matrix.
% -------------------------------------------
% STATS = GRAYCOPROPS(GLRLM,PROPERTIES) Each element in GLRLM, (r,c),
% is the probability occurrence of pixel having gray level values r, run-length c in the image.
% GRAYCOPROPS is to calculate PROPERTIES.
% -------------------------------------------
% Requirements:
% -------------------------------------------
% GLRLM mustbe an cell array of valid gray-level run-length
% matrices.Recall that a valid glrlm must be logical or numerical.
% -------------------------------------------
% Current supported statistics include:
% -------------------------------------------
% Short Run Emphasis (SRE)
% Long Run Emphasis (LRE)
% Gray-Level Nonuniformity (GLN)
% Run Length Nonuniformity (RLN)
% Run Percentage (RP)
% Low Gray-Level Run Emphasis (LGRE)
% High Gray-Level Run Emphasis (HGRE)
% Short Run Low Gray-Level Emphasis (SRLGE)
% Short Run High Gray-Level Emphasis (SRHGE)
% Long Run Low Gray-Level Emphasis (LRLGE)
% Long Run High Gray-Level Emphasis (LRHGE)
% --------------------------------------------
% Reference:
% --------------------------------------------
% Xiaoou Tang,Texture Information in Run-Length Matrices
% IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL.7, NO.11,NOVEMBER 1998
% ---------------------------------------------
% See also GRAYRLMATRIX.
% ---------------------------------------------
% Author:
% ---------------------------------------------
% (C)Xunkai Wei <xunkai.wei@gmail.com>
% Beijing Aeronautical Technology Research Center
% Beijing %9203-12,10076
% ---------------------------------------------
% History:
% ---------------------------------------------
% Creation: beta Date: 01/10/2007
% Revision: 1.0 Date: 12/11/2007
% 1.Accept cell input now
% 2.Using MATLAB file style
% 3.Fully vectorized programming
% 4.Fully support the IEEE reference
% 5. ...
% Check GLRLM
[GLRLM numGLRLM] = ParseInputs(varargin{:});
% Initialize output stats structure.
% 11 statistics for each GLRLM
numStats = 11;
% % count number of GLRLM
% numGLRLM = length(GLRLM);
% Initialization default 4*11 matrix
stats = zeros(numGLRLM,numStats);
for p = 1 : numGLRLM
%N-D indexing not allowed for sparse.
if numGLRLM ~= 1
% transfer to double matrix
tGLRLM = GLRLM{p};
else
tGLRLM = GLRLM;
end
% if numGLRLM ~= 1
% % transfer to double matrix
% tGLRLM = normalizeGLRL(GLRLM{p});
% else
% tGLRLM = normalizeGLRL(GLRLM);
% end
% Get row and column subscripts of GLRLM. These subscripts correspond to the
% pixel values in the GLRLM.
s = size(tGLRLM);
% colum indicator
c_vector =1:s(1);
% row indicator
r_vector =1:s(2);
% matrix element indicator
% Matrix form col and row: using meshgrid, you should transpose before using
% i.e. if tGLRLM is m*n, then this function return c_matrix n*m,
% r_matrix n*m.
[c_matrix,r_matrix] = meshgrid(c_vector,r_vector);
% Total number of runs
N_runs = sum(sum(tGLRLM));
% total number of elements
N_tGLRLM = s(1)*s(2);
%--------------------Prepare four matrix for speedup--------------
% 1.Gray Level Run-Length Pixel Number Matrix
% p_p = calculate_p_p(tGLRLM,c_matrix');
% 2.Gray-Level Run-Number Vector
% This vector represents the sum distribution of the number of runs
% with gray level i.
p_g = sum(tGLRLM);
% 3.Run-Length Run-Number Vector
% This vector represents the sum distribution of the number of runs
% with run length j.
p_r = sum(tGLRLM,2)';
% 4.Gray-Level Run-Length-One Vector
%
% p_o = tGLRLM(:,1); % Not used yet
% ----------------------End four matrix---------------------------
%
%------------------------Statistics-------------------------------
% 1. Short Run Emphasis (SRE)
SRE = sum(p_r./(c_vector.^2))/N_runs;
% 2. Long Run Emphasis (LRE)
LRE = sum(p_r.*(c_vector.^2))/N_runs;
% 3. Gray-Level Nonuniformity (GLN)
GLN = sum(p_g.^2)/N_runs;
% 4. Run Length Nonuniformity (RLN)
RLN = sum(p_r.^2)/N_runs;
% 5. Run Percentage (RP)
RP = N_runs/N_tGLRLM;
% 6. Low Gray-Level Run Emphasis (LGRE)
LGRE = sum(p_g./(r_vector.^2))/N_runs;
% 7. High Gray-Level Run Emphasis (HGRE)
HGRE = sum(p_g.*r_vector.^2)/N_runs;
% 8. Short Run Low Gray-Level Emphasis (SRLGE)
SGLGE =calculate_SGLGE(tGLRLM,r_matrix',c_matrix',N_runs);
% 9. Short Run High Gray-Level Emphasis (SRHGE)
SRHGE =calculate_SRHGE(tGLRLM,r_matrix',c_matrix',N_runs);
% 10. Long Run Low Gray-Level Emphasis (LRLGE)
LRLGE =calculate_LRLGE(tGLRLM,r_matrix',c_matrix',N_runs);
% 11.Long Run High Gray-Level Emphasis (LRHGE
LRHGE =calculate_LRHGE(tGLRLM,r_matrix',c_matrix',N_runs);
%----------------insert statistics----------------------------
stats(p,:)=[SRE LRE GLN RLN RP LGRE HGRE SGLGE SRHGE LRLGE LRHGE ];
end % end all run length matrixs
% ----------------------Utility functions--------------------
%-----------------------------------------------------------------------------
% function glrl = normalizeGLRL(glrl)
%
% % Normalize glcm so that sum(glcm(:)) is one.
% if any(glrl(:))
% glrl = glrl ./ sum(glrl(:));
% end
% function p_p = calculate_p_p(GLRLM,c) % Note: currently not used
%
% % p_p(i; j) = GLRLM(i,j)*j
% % Each element of the matrix represents the number of pixels of run length
% % j and gray-level i. Compared to the original matrix, the new matrix gives
% % equal emphasis to all length of runs in an image.
%
% term1 = c; % j index in matrix size
% term2 = GLRLM;
% p_p = term1 .* term2;
%---------------------------------
function SGLGE =calculate_SGLGE(tGLRLM,r_matrix,c_matrix,N_runs)
% Short Run Low Gray-Level Emphasis (SRLGE):
term = tGLRLM./((r_matrix.*c_matrix).^2);
SGLGE= sum(sum(term))./N_runs;
%------------------------------------
function SRHGE =calculate_SRHGE(tGLRLM,r_matrix,c_matrix,N_runs)
% Short Run High Gray-Level Emphasis (SRHGE):
%
term = tGLRLM.*(r_matrix.^2)./(c_matrix.^2);
SRHGE = sum(sum(term))/N_runs;
%------------------------------------
function LRLGE =calculate_LRLGE(tGLRLM,r_matrix,c_matrix,N_runs)
% Long Run Low Gray-Level Emphasis (LRLGE):
%
term = tGLRLM.*(c_matrix.^2)./(r_matrix.^2);
LRLGE = sum(sum(term))/N_runs;
%---------------------------------------
function LRHGE =calculate_LRHGE(tGLRLM,r_matrix,c_matrix,N_runs)
% Long Run High Gray-Level Emphasis (LRHGE):
%
term = tGLRLM.*(c_matrix.^2).*(r_matrix.^2);
LRHGE = sum(sum(term))/N_runs;
%----------------------------------------
%-----------------------------------------------------------------------------
function [glrlm num_glrlm] = ParseInputs(varargin)
% check stability of inputs
%
% first receive all inputs
glrlm = varargin{:};
% get numbers total
num_glrlm=length(glrlm);
% then for each element, check its stability
for i=1:num_glrlm
% The 'nonnan' and 'finite' attributes are not added to iptcheckinput because the
% 'integer' attribute takes care of these requirements.
% iptcheckinput(glrlm,{'cell'},{'real','nonnegative','integer'}, ...
% mfilename,'GLRLM',1);
iptcheckinput(glrlm{i},{'logical','numeric'},{'real','nonnegative','integer'},...
mfilename,'GLRLM',1);
% Cast GLRLM to double to avoid truncation by data type. Note that GLRLM is not an
% image.
if ~isa(glrlm,'double')
glrlm{i}= double(glrlm{i});
end
end
