How can i calculate the texture of the soap only on the picture below

I want to calculate the texture properties of The soap only, but when I use the code below it Calculates the texture of the whole image. I have also tried to use regionprops but regionprops doesn’t have the properties for texture Please Help Me

                                                  soap105.jpg

RGB=imread('soap105.jpg'); RGB=imresize(RGB,[160 190]); I=rgb2gray(RGB); . After some gray level enhancement . I2=graycomatrix(I) I3=graycoprops (I2, all )

Answer 1

[Image Analyst]

It looks like graycomatrix takes a rectangular matrix so you'll just have to find the smooth areas and zero them out. Then calculate graycomatrix and do your own stats to ignore the zero parts. See attached code that finds the smooth areas.

Of course your best course of action is to just make sure there is no soap dust on the sample in the first place. It's always easier to make sure you start with a good image before image capture than to fix it post-capture in software.

[solomon adenew]

really thanks for your help but I want to calculate texture ppts of soap only , I may have a case where big filths may be found around the soap, which have the same color as the soap.Anyhow why you use the green component only? can I calculate graycomatrix for the smoothed image and then graycoprops to calculate texture properties

[Image Analyst]

You have essentially a monochrome image. The green channel has the lowest noise so I used that. I don't see any reason to use color processing for this image - do you?

Like I said, do your texture stuff on the masked image but when you look at the matrix you'll have to ignore any data that you got when the pixel was next to a zero pixel - in essence the first row and column of the GLCM.

[solomon adenew]

 Great thanks:- to make it clear I am doing a project that detect abnormal soaps from a soap industry(abnormal soaps in texture, color, size )from the conveior belt by putting a camera above the conveyior belt. what is  difficult for me is measuring the texture of soap only not including background.
  I have got a function code from the internet that calculate the texture of every objects in grayscale image by masking its binary image. but it has been difficult for me is how to pass the image data to this function code please help me. how can I pass the previous soap image, soap105.jpg to the code. here is the code

    function handles = MeasureTexture(handles,varargin)

% Help for the Measure Texture module: % Category: Measurement % % SHORT DESCRIPTION: % Measures several texture features for identified objects or for entire % images. % *********************************************************************** % % Given an image with objects identified (e.g. nuclei or cells), this % module extracts texture features for each object based on a corresponding % grayscale image. Measurements are recorded for each object. If "Image" is % chosen, the texture of the image overall is measured. % % How it works: % Retrieves objects in label matrix format and a corresponding original % grayscale image and makes measurements of the objects. The label matrix % image should be "compacted": that is, each number should correspond to an % object, with no numbers skipped. So, if some objects were discarded from % the label matrix image, the image should be converted to binary and % re-made into a label matrix image before feeding into this module. % % The scale of texture measured is chosen by the user, in pixel units. A % higher number for the scale of texture measures larger patterns of % texture whereas smaller numbers measure more localized patterns of % texture. It is best to measure texture on a scale smaller than your % objects' sizes, so be sure that the value entered for scale of texture is % smaller than most of your objects. For very small objects (smaller than % the scale of texture you are measuring), the texture cannot be measured % and will result in a value of NaN (Not a Number) in the output file. % % A range of texture scales may be specified, as a comma-separated list. % Measurements will be generated for all scales specified. % % Note that texture measurements are affected by the overall intensity of % the object (or image). For example, if Image1 = Image2 + 0.2, then the % texture measurements should be the same for Image1 and Image2. However, % if the images are scaled differently, for example Image1 = 0.9*Image2, % then this will be reflected in the texture measurements, and they will be % different. For example, in the extreme case of Image1 = 0*Image2 it is % obvious that the texture measurements must be different. To make the % measurements useful (both intensity, texture, etc.), it must be ensured % that the images are scaled similarly. In other words, if differences in % intensity are seen between two images or objects, the differences in % texture cannot be trusted as being completely independent of the % intensity difference. % % Features measured: Feature Number: % AngularSecondMoment | 1 % Contrast | 2 % Correlation | 3 % Variance | 4 % InverseDifferenceMoment | 5 % SumAverage | 6 % SumVariance | 7 % SumEntropy | 8 % Entropy | 9 % DifferenceVariance | 10 % DifferenceEntropy | 11 % InfoMeas | 12 % InfoMeas2 | 13 % GaborX | 14 % GaborY | 15 % % Texture Measurement descriptions: % % Haralick Features: % Haralick texture features are derived from the co-occurrence matrix, % which contains information about how image intensities in pixels with a % certain position in relation to each other occur together. For example, % how often does a pixel with intensity 0.12 have a neighbor 2 pixels to % the right with intensity 0.15? The current implementation in CellProfiler % uses a shift of 1 pixel to the right for calculating the co-occurence % matrix. A different set of measurements is obtained for larger shifts, % measuring texture on a larger scale. The original reference for the % Haralick features is Haralick et al. (1973) Textural Features for Image % Classification. IEEE Transaction on Systems Man, Cybernetics, % SMC-3(6):610-621, where 14 features are described: % H1. Angular Second Moment % H2. Contrast % H3. Correlation % H4. Sum of Squares: Variation % H5. Inverse Difference Moment % H6. Sum Average % H7. Sum Variance % H8. Sum Entropy % H9. Entropy % H10. Difference Variance % H11. Difference Entropy % H12. Information Measure of Correlation 1 % H13. Information Measure of Correlation 2 % H14. Max correlation coefficient % % *H14 is disabled because it is computationally demanding. % % Gabor "wavelet" features: % These features are similar to wavelet features, and they are obtained by % applying so-called Gabor filters to the image. The Gabor filters measure % the frequency content in different orientations. They are very similar to % wavelets, and in the current context they work exactly as wavelets, but % they are not wavelets by a strict mathematical definition. As currently % implemented, the frequency content of the object is measured along the x- % and y-axis (i.e. in two different orientations). The original reference % is Gabor, D. (1946). "Theory of communication" Journal of the Institute % of Electrical Engineers, 93:429-441.

% CellProfiler is distributed under the GNU General Public License. % See the accompanying file LICENSE for details. % % Developed by the Whitehead Institute for Biomedical Research. % Copyright 2003,2004,2005. % % Please see the AUTHORS file for credits. % % Website: http://www.cellprofiler.org % % $Revision$

% MBray 2009_03_20: Comments on variables for pyCP upgrade % % Recommended variable order (setting, followed by current variable in MATLAB CP) % (1) Input grayscale image (ImageName) % We should reword this to be "What did you call the greyscale images whose texture you % want to measure? % % (2) Input objects (ObjectNameList) % We should reword this to be "What did you call the objects within which you want % to measure texture?" % % (3) Feature scale (ScaleOfTexture) % We should reword this to be "What scale of texture do you want to measure?" % % (i) A button should be added that lets the user add/subtract images for (1) and objects % for (2) % (ii) The feature scale should let the user specify a range of texture % scales, so a module doesn't have to be added for each one. (not sure % whether we want a "range of textures" (not sure how that would be % entered) or instead add/subtract buttons to type in individual scales of texture.)

%%%%%%%%%%%%%%%%% %%% VARIABLES %%% %%%%%%%%%%%%%%%%%

[CurrentModule, CurrentModuleNum, ModuleName] = CPwhichmodule(handles);

%textVAR01 = What did you call the greyscale images you want to measure? %infotypeVAR01 = imagegroup ImageName = char(handles.Settings.VariableValues{CurrentModuleNum,1}); %inputtypeVAR01 = popupmenu

%textVAR02 = What did you call the objects that you want to measure? %choiceVAR02 = Image %choiceVAR02 = Do not use %infotypeVAR02 = objectgroup ObjectNameList{1} = char(handles.Settings.VariableValues{CurrentModuleNum,2}); %inputtypeVAR02 = popupmenu

%textVAR03 = %choiceVAR03 = Do not use %infotypeVAR03 = objectgroup ObjectNameList{2} = char(handles.Settings.VariableValues{CurrentModuleNum,3}); %inputtypeVAR03 = popupmenu

%textVAR04 = %choiceVAR04 = Do not use %infotypeVAR04 = objectgroup ObjectNameList{3} = char(handles.Settings.VariableValues{CurrentModuleNum,4}); %inputtypeVAR04 = popupmenu

%textVAR05 = %choiceVAR05 = Do not use %infotypeVAR05 = objectgroup ObjectNameList{4} = char(handles.Settings.VariableValues{CurrentModuleNum,5}); %inputtypeVAR05 = popupmenu

%textVAR06 = %choiceVAR06 = Do not use %infotypeVAR06 = objectgroup ObjectNameList{5} = char(handles.Settings.VariableValues{CurrentModuleNum,6}); %inputtypeVAR06 = popupmenu

%textVAR07 = %choiceVAR07 = Do not use %infotypeVAR07 = objectgroup ObjectNameList{6} = char(handles.Settings.VariableValues{CurrentModuleNum,7}); %inputtypeVAR07 = popupmenu

%textVAR08 = What is the scale of texture? A list of texture scales can be specified, separated by commas. %defaultVAR08 = 3 ScaleOfTexture = char(handles.Settings.VariableValues{CurrentModuleNum,8});

%%%%%%%%%%%%%%%% %%% FEATURES %%% %%%%%%%%%%%%%%%%

if nargin > 1 switch varargin{1} %feature:categories case 'categories' if nargin == 1 ismember(varargin{2},ObjectNameList) result = { 'Texture' }; else result = {}; end %feature:measurements case 'measurements' result = {}; if nargin >= 3 &&... strcmp(varargin{3},'Texture') &&... ismember(varargin{2},ObjectNameList) result = { ... 'AngularSecondMoment','Contrast','Correlation',... 'Variance','InverseDifferenceMoment',... 'SumAverage','SumVariance','SumEntropy',... 'Entropy','DifferenceVariance','DifferenceEntropy',... 'InfoMeas','InfoMeas2','GaborX','GaborY' }; end otherwise error(['Unhandled category: ',varargin{1}]); end handles=result; return; end

%%%VariableRevisionNumber = 2

%%% Set up the window for displaying the results ThisModuleFigureNumber = handles.Current.(['FigureNumberForModule',CurrentModule]); if any(findobj == ThisModuleFigureNumber) CPfigure(handles,'Text',ThisModuleFigureNumber); columns = 1; end

ScaleOfTexture = cellfun(@str2double,strread(ScaleOfTexture,'%s','delimiter',','));

% Loop through all the texture scales specified for TextureScaleNum = ScaleOfTexture(:)', %%% START LOOP THROUGH ALL THE OBJECTS for ObjectNameListNum = 1:6 ObjectName = ObjectNameList{ObjectNameListNum}; if strcmp(ObjectName,'Do not use') continue end

        %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
        %%% PRELIMINARY CALCULATIONS & FILE HANDLING %%%
        %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

        %%% Reads (opens) the image you want to analyze and assigns it to a variable,
        %%% "OrigImage".
        OrigImage = CPretrieveimage(handles,ImageName,ModuleName,'MustBeGray','CheckScale');

        if ~strcmp(ObjectName,'Image')
            %%% Retrieves the label matrix image that contains the segmented objects which
            %%% will be measured with this module.
            LabelMatrixImage = CPretrieveimage(handles,['Segmented', ObjectName],ModuleName,'MustBeGray','DontCheckScale');
            %%% For the cases where the label matrix was produced from a cropped
            %%% image, the sizes of the images will not be equal. So, we crop the
            %%% LabelMatrix and try again to see if the matrices are then the
            %%% proper size. Removes Rows and Columns that are completely blank.
            if any(size(OrigImage) < size(LabelMatrixImage))
                ColumnTotals = sum(LabelMatrixImage,1);
                RowTotals = sum(LabelMatrixImage,2)';
                warning off all
                ColumnsToDelete = ~logical(ColumnTotals);
                RowsToDelete = ~logical(RowTotals);
                warning on all
                drawnow
                CroppedLabelMatrix = LabelMatrixImage;
                CroppedLabelMatrix(:,ColumnsToDelete,:) = [];
                CroppedLabelMatrix(RowsToDelete,:,:) = [];
                clear LabelMatrixImage
                LabelMatrixImage = CroppedLabelMatrix;
                %%% In case the entire image has been cropped away, we store a single
                %%% zero pixel for the variable.
                if isempty(LabelMatrixImage)
                    LabelMatrixImage = 0;
                end
            end
            if any(size(OrigImage) ~= size(LabelMatrixImage))
                error(['Image processing was canceled in the ', ModuleName, ' module. The size of the image you want to measure is not the same as the size of the image from which the ',ObjectName,' objects were identified.'])
            end
        end

        %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
        %%% MAKE MEASUREMENTS & SAVE TO HANDLES STRUCTURE %%%
        %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

        %%% Initialize measurement structure
        Haralick = [];
        HaralickFeatures = {'AngularSecondMoment',...
            'Contrast',...
            'Correlation',...
            'Variance',...
            'InverseDifferenceMoment',...
            'SumAverage',...
            'SumVariance',...
            'SumEntropy',...
            'Entropy',...
            'DifferenceVariance',...
            'DifferenceEntropy',...
            'InfoMeas1',...
            'InfoMeas2'};

        Gabor = [];
        GaborFeatures    = {'GaborX',...
            'GaborY'};

        if strcmp(ObjectName,'Image')
            ObjectCount = 1;
        else
            %%% Count objects
            ObjectCount = max(LabelMatrixImage(:));
        end

        if ObjectCount > 0 || strcmp(ObjectName,'Image')

            %%% Get Gabor features.
            %%% The Gabor features are calculated by convolving the entire
            %%% image with Gabor filters and then extracting the filter output
            %%% value in the centroids of the objects in LabelMatrixImage

            if ~strcmp(ObjectName,'Image')
                % Adjust size of filter to size of objects in the image
                % The centroids indicate where we should measure the Gabor
                % filter output
                tmp = regionprops(LabelMatrixImage,'Area','Centroid');
                Areas = cat(1,tmp.Area);
                MedianArea = median(Areas);

                % Round centroids and find linear index for them.
                % The centroids are stored in [column,row] order.
                Centroids = round(cat(1,tmp.Centroid));
            else
                MedianArea = size(OrigImage,1)*size(OrigImage,2);
            end

            sigma = sqrt(MedianArea/pi)/3;  % Set width of filter to a third of the median radius

            % Use Gabor filters with three different frequencies
            f = 1/(2*TextureScaleNum);

            % Angle direction, filter along the x-axis and y-axis
            theta = [0 pi/2];

            if ~strcmp(ObjectName,'Image')
                % Create kernel coordinates
                KernelSize = round(2.5*sigma); % The filter size is set somewhat arbitrary
            else
                KernelSize = max(size(OrigImage,1)/2,size(OrigImage,2)/2);
            end
            [x,y] = meshgrid(-KernelSize:KernelSize,-KernelSize:KernelSize);

            % Apply Gabor filters and store filter outputs in the Centroid pixels
            GaborFeatureNo = 1;
            Gabor = zeros(ObjectCount,length(f)*length(theta));                              % Initialize measurement matrix
            for m = 1:length(f)
                for n = 1:length(theta)

                    % Calculate Gabor filter kernel
                    % Scale by 1000 to get measurements in a convenient range
                    g = 1000*1/(2*pi*sigma^2)*exp(-(x.^2 + y.^2)/(2*sigma^2)).*exp(2*pi*sqrt(-1)*f(m)*(x*cos(theta(n))+y*sin(theta(n))));
                    g = g - mean(g(:));           % Important that the filters has DC zero, otherwise they will be sensitive to the intensity of the image

                    % Center the Gabor kernel over the centroid and calculate the filter response.
                    if strcmp(ObjectName,'Image')
                        % Cut patch
                        p = OrigImage;

                        if size(OrigImage,1) ~= size(g,1)
                            p = [p;zeros(size(g,1)-size(p,1),size(p,2))];
                        end

                        if size(OrigImage,2) ~= size(g,2)
                            p = [p zeros(size(p,1),size(g,2)-size(p,2))];
                        end
                        % Calculate the filter output
                        Gabor(1,GaborFeatureNo) = abs(sum(sum(g.*p)));
                    else
                        for k = 1:ObjectCount
                            %%% It's possible for objects not to have any pixels,
                            %%% particularly tertiary objects (such as cytoplasm from
                            %%% cells the exact same size as their nucleus).
                            if Areas(k) == 0,
                                Gabor(k, GaborFeatureNo) = 0;
                                continue;
                            end

                            xmin1 = Centroids(k,1)-KernelSize;
                            xmax1 = Centroids(k,1)+KernelSize;
                            ymin1 = Centroids(k,2)-KernelSize;
                            ymax1 = Centroids(k,2)+KernelSize;
                            xmin2 = max(1,xmin1);
                            xmax2 = min(size(OrigImage,2),xmax1);
                            ymin2 = max(1,ymin1);
                            ymax2 = min(size(OrigImage,1),ymax1);

                            % Cut patch
                            p = OrigImage(ymin2:ymax2,xmin2:xmax2);

                            % Pad with zeros if necessary to match the filter kernel size
                            if xmin1 < xmin2
                                p = [zeros(size(p,1),xmin2 - xmin1) p];
                            end
                            if xmax1 > xmax2
                                p = [p zeros(size(p,1),xmax1 - xmax2)];
                            end

                            if ymin1 < ymin2
                                p = [zeros(ymin2 - ymin1,size(p,2));p];
                            end
                            if ymax1 > ymax2
                                p = [p;zeros(ymax1 - ymax2,size(p,2))];
                            end

                            % Calculate the filter output
                            Gabor(k,GaborFeatureNo) = abs(sum(sum(g.*p)));
                        end
                    end
                    GaborFeatureNo = GaborFeatureNo + 1;
                end
            end

            if strcmp(ObjectName,'Image')
                [m,n] = size(OrigImage);
                BWim = ones(m,n);
                %%% Get Haralick features
                Haralick(1,:) = CalculateHaralick(OrigImage,BWim,TextureScaleNum);
            else
                %%% Get Haralick features.
                %%% Have to loop over the objects
                Haralick = zeros(ObjectCount,13);
                [sr sc] = size(LabelMatrixImage);
                props = regionprops(LabelMatrixImage,'PixelIdxList');   % Get pixel indexes in a fast way
                for Object = 1:ObjectCount
                    %%% Cut patch so that we don't have to deal with entire image
                    [r,c] = ind2sub([sr sc],props(Object).PixelIdxList);
                    rmax = min(sr,max(r));
                    rmin = max(1,min(r));
                    cmax = min(sc,max(c));
                    cmin = max(1,min(c));
                    BWim   = LabelMatrixImage(rmin:rmax,cmin:cmax) == Object;
                    Greyim = OrigImage(rmin:rmax,cmin:cmax);
                    %%% Get Haralick features
                    Haralick(Object,:) = CalculateHaralick(Greyim,BWim,TextureScaleNum);
                end
            end
        else
            Haralick = zeros(0,13);
            Gabor = zeros(0,2);
        end
        %%% Save measurements
        AllFeatures = cat(2,HaralickFeatures,GaborFeatures);
        Data = [Haralick Gabor];
        for FeatureNum = 1:length(AllFeatures)
            feature_name = CPjoinstrings('Texture',char(AllFeatures{FeatureNum}),ImageName,num2str(TextureScaleNum));
            handles = CPaddmeasurements(handles, ObjectName, feature_name, Data(:,FeatureNum));
        end

        %%% Report measurements
        FontSize = handles.Preferences.FontSize;

        if any(findobj == ThisModuleFigureNumber);
            % Remove uicontrols from last cycle
            delete(findobj(ThisModuleFigureNumber,'tag','TextUIControl'));

            % This first block writes the same text several times
            % Header

            if handles.Current.SetBeingAnalyzed == handles.Current.StartingImageSet
                delete(findobj('parent',ThisModuleFigureNumber,'string','R'));
                delete(findobj('parent',ThisModuleFigureNumber,'string','G'));
                delete(findobj('parent',ThisModuleFigureNumber,'string','B'));
            end

            uicontrol(ThisModuleFigureNumber,'style','text','units','normalized', 'position', [0 0.95 1 0.04],...
                'HorizontalAlignment','center','BackgroundColor',[.7 .7 .9],'fontname','Helvetica','tag','TextUIControl',...
                'fontsize',FontSize,'fontweight','bold','string',sprintf(['Average texture features for ',ImageName,', cycle #%d'],handles.Current.SetBeingAnalyzed));

            % Number of objects
            uicontrol(ThisModuleFigureNumber,'style','text','units','normalized', 'position', [0.05 0.85 0.3 0.03],...
                'HorizontalAlignment','left','BackgroundColor',[.7 .7 .9],'fontname','Helvetica','tag','TextUIControl',...
                'fontsize',FontSize,'fontweight','bold','string','Number of objects:');

            % Text for Gabor features
            uicontrol(ThisModuleFigureNumber,'style','text','units','normalized', 'position', [0.05 0.8 0.3 0.03],...
                'HorizontalAlignment','left','BackgroundColor',[.7 .7 .9],'fontname','Helvetica','tag','TextUIControl',...
                'fontsize',FontSize,'fontweight','bold','string','Gabor features:');
            for k = 1:2
                uicontrol(ThisModuleFigureNumber,'style','text','units','normalized', 'position', [0.05 0.8-0.04*k 0.3 0.03],...
                    'HorizontalAlignment','left','BackgroundColor',[.7 .7 .9],'fontname','Helvetica','tag','TextUIControl',...
                    'fontsize',FontSize,'string',GaborFeatures{k});
            end

            % Text for Haralick features
            uicontrol(ThisModuleFigureNumber,'style','text','units','normalized', 'position', [0.05 0.65 0.3 0.03],...
                'HorizontalAlignment','left','BackgroundColor',[.7 .7 .9],'fontname','Helvetica','tag','TextUIControl',...
                'fontsize',FontSize,'fontweight','bold','string','Haralick features:');
            for k = 1:10
                uicontrol(ThisModuleFigureNumber,'style','text','units','normalized', 'position', [0.05 0.65-0.04*k 0.3 0.03],...
                    'HorizontalAlignment','left','BackgroundColor',[.7 .7 .9],'fontname','Helvetica','tag','TextUIControl',...
                    'fontsize',FontSize,'string',HaralickFeatures{k});
            end

            % The name of the object image
            uicontrol(ThisModuleFigureNumber,'style','text','units','normalized', 'position', [0.35+0.2*(columns-1) 0.9 0.2 0.03],...
                'HorizontalAlignment','center','BackgroundColor',[.7 .7 .9],'fontname','Helvetica','tag','TextEachObjUIControl',...
                'fontsize',FontSize,'fontweight','bold','string',ObjectName);

            % Number of objects
            uicontrol(ThisModuleFigureNumber,'style','text','units','normalized', 'position', [0.35+0.2*(columns-1) 0.85 0.2 0.03],...
                'HorizontalAlignment','center','BackgroundColor',[.7 .7 .9],'fontname','Helvetica','tag','TextEachObjUIControl',...
                'fontsize',FontSize,'string',num2str(ObjectCount));

            if ObjectCount > 0
                % Gabor features
                for k = 1:2
                    uicontrol(ThisModuleFigureNumber,'style','text','units','normalized', 'position', [0.35+0.2*(columns-1) 0.8-0.04*k 0.2 0.03],...
                        'HorizontalAlignment','center','BackgroundColor',[.7 .7 .9],'fontname','Helvetica','tag','TextEachObjUIControl',...
                        'fontsize',FontSize,'string',sprintf('%0.2f',mean(Gabor(:,k))));
                end

                % Haralick features
                for k = 1:10
                    uicontrol(ThisModuleFigureNumber,'style','text','units','normalized', 'position', [0.35+0.2*(columns-1) 0.65-0.04*k 0.2 0.03],...
                        'HorizontalAlignment','center','BackgroundColor',[.7 .7 .9],'fontname','Helvetica','tag','TextEachObjUIControl',...
                        'fontsize',FontSize,'string',sprintf('%0.2f',mean(Haralick(:,k))));
                end
            end
            % This variable is used to write results in the correct column
            % and to determine the correct window size
            columns = columns + 1;
        end
    end
end
drawnow

function H = CalculateHaralick(im,mask,ScaleOfTexture)

warning off MATLAB:DivideByZero % This function calculates so-called Haralick features, which are % based on the co-occurence matrix. The function takes two inputs: % % im - A grey level image % mask - A binary mask % % Currently, the implementation uses 8 different grey levels % and calculates the co-occurence matrix for a horizontal shift % of 1 pixel. % % The original reference is: % Haralick et al. (1973) % Textural Features for Image Classification. % IEEE Transaction on Systems % Man, Cybernetics, SMC-3(6):610-621. % % BEWARE: There are lots of erroneous formulas for the Haralick features in % the literature. There is also an error in the original paper. %

% Number of greylevels to use Levels = 8;

% Quantize the image into a lower number of grey levels (specified by Levels) BinEdges = linspace(0,1,Levels+1);

% Find the max and min values within the mask and normalize so that the % intenisties within the mask are between 0 and 1. intensities = im(mask); Imax = max(intensities(:)); Imin = min(intensities(:)); if Imax ~= Imin % Avoid divide by zero im = (im - Imin)/(Imax-Imin); end

% Do the quantization qim = zeros(size(im)); for k = 1:Levels qim(find(im >= BinEdges(k))) = k; %#ok Ignore MLint end

% Shift ScaleOfTexture step to the right im1 = qim(:,1:end-ScaleOfTexture); im1 = im1(:); im2 = qim(:,ScaleOfTexture+1:end); im2 = im2(:);

% Remove cases where at least one position is % outside the mask. m1 = mask(:,1:end-ScaleOfTexture); m1 = m1(:); m2 = mask(:,ScaleOfTexture+1:end); m2 = m2(:);

index = (sum([m1 m2],2) == 2); if isempty(index) H = [0 0 0 0 0 0 0 0 0 0 0 0 0]; return end im1 = im1(index); im2 = im2(index);

%%% Calculate co-occurence matrix % P = zeros(Levels); % for k = 1:Levels % index = find(im1==k); % if ~isempty(index) % P(k,:) = hist(im2(index),(1:Levels)); % else % P(k,:) = zeros(1,Levels); % end % end % % The line below is a fast 2D-histogram in matlab, and is equivalent to the % loop above. Ray & Kyungnam, 2007-07-18. P = full(sparse(im1,im2,1,Levels,Levels)); P = P/length(im1);

%%% Calculate features from the co-occurence matrix % First, pre-calculate a few quantities that are used in % several features. px = sum(P,2); py = sum(P,1); mux = sum((1:Levels)'.*px); muy = sum((1:Levels).*py); sigmax = sqrt(sum(((1:Levels)' - mux).^2.*px)); sigmay = sqrt(sum(((1:Levels) - muy).^2.*py)); HX = -sum(px.*log(px+eps)); HY = -sum(py.*log(py+eps)); HXY = -sum(P(:).*log(P(:)+eps)); HXY1 = -sum(sum(P.*log(px*py+eps))); HXY2 = -sum(sum(px*py .* log(px*py+eps)));

p_xplusy = zeros(2*Levels-1,1); % Range 2:2*Levels p_xminusy = zeros(Levels,1); % Range 0:Levels-1 for x=1:Levels for y = 1:Levels p_xplusy(x+y-1) = p_xplusy(x+y-1) + P(x,y); p_xminusy(abs(x-y)+1) = p_xminusy(abs(x-y)+1) + P(x,y); end end

% H1. Angular Second Moment H1 = sum(P(:).^2);

% H2. Contrast H2 = sum((0:Levels-1)'.^2.*p_xminusy);

% H3. Correlation H3 = (sum(sum((1:Levels)'*(1:Levels).*P)) - mux*muy)/(sigmax*sigmay); if isinf(H3), H3 = 0; end

% H4. Sum of Squares: Variation H4 = sigmax^2;

% H5. Inverse Difference Moment H5 = sum(sum(1./(1+toeplitz(0:Levels-1).^2).*P));

% H6. Sum Average H6 = sum((2:2*Levels)'.*p_xplusy);

% H7. Sum Variance (error in Haralick's original paper here) H7 = sum(((2:2*Levels)' - H6).^2 .* p_xplusy);

% H8. Sum Entropy H8 = -sum(p_xplusy .* log(p_xplusy+eps));

% H9. Entropy H9 = HXY;

% H10. Difference Variance H10 = sum(p_xminusy.*((0:Levels-1)' - sum((0:Levels-1)'.*p_xminusy)).^2);

% H11. Difference Entropy H11 = - sum(p_xminusy.*log(p_xminusy+eps));

% H12. Information Measure of Correlation 1 H12 = (HXY-HXY1)/max(HX,HY);

% H13. Information Measure of Correlation 2 H13 = real(sqrt(1-exp(-2*(HXY2-HXY)))); % An imaginary result has been encountered once, reason unclear

warning on MATLAB:DivideByZero

% H14. Max correlation coefficient (not currently used) % Q = zeros(Levels); % for i = 1:Levels % for j = 1:Levels % Q(i,j) = sum(P(i,:).*P(j,:)/(px(i)*py(j))); % end % end % [V,lambda] = eig(Q); % lambda = sort(diag(lambda)); % H14 = sqrt(max(0,lambda(end-1)));

H = [H1 H2 H3 H4 H5 H6 H7 H8 H9 H10 H11 H12 H13]; H(isnan(H))=0;

% % This function calculates Gabor features in a different way % % It may be better but it's also considerably slower. % % It's called by Gabor(Object,:) = CalculateGabor(Greyim,BWim,sigma); % function G = CalculateGabor(im,mask,sigma,flag) % % % % This function calculates Gabor features, which measure % % the energy in different frequency sub-bands. The Gabor % % transform is essentially equivalent to a wavelet transform. % % % % im - A grey level image % % mask - A binary mask % % sigma - Scale parameter for the Gaussian weight function % % % Use Gabor filters with three different frequencies % f = [0.06 0.12 0.24]; % % % Filter along the x-axis and y-axis % theta = [0 pi/2]; % % % Match the filter kernel size to the input patch size % [sr,sc] = size(mask); % if rem(sr,2) == 0,ty = [-sr/2:sr/2-1];else ty = [-(sr-1)/2:(sr-1)/2];end % if rem(sc,2) == 0,tx = [-sc/2:sc/2-1];else tx = [-(sc-1)/2:(sc-1)/2];end % [x,y]=meshgrid(tx,ty); % % % Calculate the Gabor features % G = zeros(length(theta),length(f)); % for m = 1:length(f) % for n = 1:length(theta) % % % Calculate Gabor filter kernel % g = 1/(2*pi*sigma^2)*exp(-(x.^2 + y.^2)/(2*sigma^2)).*exp(2*pi*sqrt(-1)*f(m)*(x*cos(theta(n))+y*sin(theta(n)))); % % % Use Normalized Convolution to calculate filter responses. This % % method only include object pixels for calculating the filter % % response and excludes surrounding background pixels. % % See Farneback, 2002. "Polynomial Expansion for Orientation and % % Motion Estimation". PhD Thesis % gr = real(g); % gi = imag(g); % B = [gr(:) gi(:)]; % Wc = diag(mask(:)); % r = inv(B'*Wc*B)*B'*Wc*im(:); % G(n,m) = sqrt(sum(r.^2)); % % % Direct way of calculating filter responses % %tmpr = sum(sum(real(g).*im)); % %tmpi = sum(sum(imag(g).*im)); % %G(n,m) = sqrt(tmpr.^2+tmpi.^2); % end % end % G = G(:)';

[Image Analyst]

Pass the filename or the array in via the input argument list. By the way, why are the soaps all sudsy on the conveyor belt? Is this some kind of QA/testing apparatus, or a manufacturing line (where they should not be sudsy)?

[solomon adenew]

what do u mean by sudsy? (English is my second language) would u give me an answer:- can I take the graycomatrix of a gray image which is masked by its binary,ROI. when we mask always regions outside of the ROI will be zero.if i take the graycomatrix, will the zeros has an effect, in your previous comment you said:if I have to take the graycomatrix I have to ignore the zeros. how can I ignor, even if, if I crop the image,the zeros will not be omitted totally. what I have to do? is there a solution? really thanks for your help

[Image Analyst]

Sudsy means covered with suds/lather/foam/bubbles.

You can mask the image and those pixels will be zero. That means in the GLCM there will be pixels represented that have a neighbor of value zero. I recommend you use the full GLCM of 256 gray levels, not the defaul one that chops up the range into something like 8 ranges of 0-31, 23-63, etc. Then you'll see that there are elements in the row and column representing zero. Not sure how you want to handle those or how they affect the properties in graycoprops() but you need to think about how to handle them.

[solomon adenew]

the lather on soap is by chance. I am doing a project on a soap factory, I will put a camera on the top of the conveyor belt, in which sops are flowing, then the camera continually will take snapshot of images to MATLAB, then the images will be analysed to check defects in texture, color, size and soap count and detection of errors in the uniformity of flow. I have got one suggestion from the internet that I can make the zeros outside ROI set to Nan. is it correct ??? is it possible?? is there any other effect?? thanks

[Image Analyst]

Try it and see. I'm not sure how graycomatrix will react to the presence of nans in the image. If you're lucky it will just ignore those locations. You can use the masks I made for you to set those pixels to nans if you want

smoothGray = greenChannel; % Initialize
smoothGray(~smoothAreas) = nan; % Set non-smooth areas (i.e. suds) to nan.

[solomon adenew]

s

[Image Analyst]

http://www.mathworks.com/matlabcentral/answers/13205-tutorial-how-to-format-your-question-with-markup

[solomon adenew]

thanks for your correction