pic2points.m

% -------------------------------------------------------------------------
% | Pic/Plot images to Coordinates Points                                 |
% | Amir Hossein Ashtari                                                  |
% | National University of Malaysia, Center for Artificial Intelligence   |
% | Technology (CAIT), Faculty Of Information Science and Technology      |
% | (c) 2015                                                              |
% -------------------------------------------------------------------------
function [MResult] = pic2points(Im,TshV,ImPlot,maxNum)
% -------------------------------------------------------------------------
% |                        How to use the code                            |
% -------------------------------------------------------------------------
% The provided source code gets an input image (black object pixels on 
% white background, e.g. an scanned plot with black ink on a white paper)
% The output is coordinates of all points/pixles in the image, therefore 
% the image is converted to coordinates of points and it is possible for 
% re-plotting the scanned image. In other words, the provided function is  
% an OCR for scanned plot to extract the image points coordinates. 
%
% ----------------------------------Inputs---------------------------------
%
%  (1)Im        - Input image, it can be gray, color or binary 
%                 image. The image can be Double = [0,1] or 
%                 Unsign integer = [0,255]. The image foreground must
%			      be black on white background.
%                    
%  (2)TshV      - Denotes the threshold value for image binarization.
%				  It is between 0~1 and by default is calculated by
%				  graythresh(Im).
%
%  (3)ImPlot    - If the image is a plot, ImPlot must be �1� or �true�. 
%                 The image is thinned to 1 pixel width line when ImPlot is 
% 	              true therefore finding the coordinate of drawn line in 
%	              plot would be possible.
%
%  (4)maxNum    - Denotes maximum number of points that should be	
%				  extracted from the image. If all points in the image
%				  are requested, leave the maxNum empty; otherwise enter 
%				  the arbitrary number to select randomly from all points.
%				  Actually it is a random resampling from all point for 
%                 output.
%    
% ----------------------------------Output---------------------------------
%
%  (1) MResult  - Matrix of pixels/points coordinates in the image. MResult
%				  has two columns, first column denotes x-coordinates
%				  and second column is for y-coordinates. 
%				  Number of rows equals to number of points in the image.
%   
% --------------------------------Examples---------------------------------
%
% example 1:
%
% Im = imread('./Images/Plot.png');
% CoordinateMatrix = pic2points(Im);
% scatter(CoordinateMatrix (:,1), CoordinateMatrix (:,2),'.'); 
%                                                          %Plot the points
%
% ----------------------------------
% example 2:
%
% Im = imread('./Images/Plot.png');
% CoordinateMatrix = pic2points(Im,0.547); % With defined threshold for 
%							               % binarization.
% scatter(CoordinateMatrix (:,1), CoordinateMatrix (:,2),'.'); 
%                                                          %Plot the points
% ----------------------------------
% example 3:
%
% Im = imread('./Images/Plot.png');
% CoordinateMatrix = pic2points(Im,0.547,1); % With defined threshold for 
%							                 % binarization and it is plot
%							                 % so line must be thin to 1 
%							                 % pixel.
% scatter(CoordinateMatrix (:,1), CoordinateMatrix (:,2),'.'); 
%                                                          %Plot the points
%
% ----------------------------------
% example 4:
%
% Im = imread('./Images/Plot2.png');
% CoordinateMatrix = pic2points(Im,0.547,1,1000); % With defined threshold 
%							               % binarization and it is plot 
%							               % so line must be thin to 1
%							               % pixel. 1000 is the number of
%							               % points that return by function
% scatter(CoordinateMatrix (:,1), CoordinateMatrix (:,2),'.'); 
%                                                          %Plot the points
%
% -------------------------------------------------------------------------
switch nargin
    case 1
        Pmode = false; % Not a plot
        randomPick = false;
        TshV = graythresh(Im);
    case 2
        Pmode = false; % Not a plot
        randomPick = false;
    case 3
        if strcmpi(ImPlot,'plot')
            Pmode = true; % It is a plot
        else
            Pmode = false; % It is an image
        end
        randomPick = false;
    case 4
        if strcmpi(ImPlot,'plot')
            Pmode = true; % It is a plot
        else
            Pmode = false; % It is an image
        end
        randomPick = true;
end
ImBW=im2bw(Im,TshV);
ImBW=1-ImBW; % When the plot drew by black ink on the white background
if Pmode
    ImBW = bwmorph(ImBW,'thin',Inf);
end
figure(); imshow(ImBW);
[r,c,v] = find(ImBW==1);
r = size(ImBW,1) - r; % Correct the coordinates from the image
TMresult = [c r];
if randomPick
    Sran = randperm(sum(v),maxNum);
    MResult = TMresult(Sran',:);
else
    MResult = TMresult;
end