edge_sampling.m

一个学习自然场景类别的贝叶斯模型、基于“词袋”模型的目标分类。来源于Feifei Li的论文。是近年来的目标识别模型热点之一。

function Edge_Sampling(image_file_names,output_file_names,Interest_Point)

  %% Simple inerest operator that:
  %%    1. Runs Canny edge detector on image
  %%    2. Sample Interest_Point.Max_Points points from set of edgels, weighted according to their intensity
  %%    3. For each sample, set scale by drawing from uniform distribution ...
  %%        over Interest_Point.Scale
    
  %% Inputs: 
  %%      1. image_file_names - cell array of filenames of all images to be processed
  %%      2. output_file_names - cell array of output filenames
  %%      3. Interest_Point - structure holding all settings of the interest operator
  
  %% Outputs:
  %%      None - it saves the results for each image to the files
  %%      specified in output_file_names.
  %%      Each file holds 4 variables:
  %%          x - x coordinates of points (1 x Interest_Point.Max_Points)
  %%          y - y coordinates of points (1 x Interest_Point.Max_Points)
  %%          scale - characteristic scale of points (radius, in pixels)  (1 x Interest_Point.Max_Points)
  %%          score - importance measure of each point, determined by edge strength of pixels (1 x Interest_Point.Max_Points).

    
  %%% R.Fergus ([e-mail])  03/10/05.  
     
    
%%% DEBUG swtich. Normally set to 0 but if set to >=1 will start plotting
%%% out results so user can check operation of the fucntion    
DEBUG       = 0

%%% Default parameters section
if (nargin<3)
   fprintf('Edge_Sampling: No settings specified, so using defaults...\n');
   Interest_Points.Max_Points = 200;
   Interest_Points.Weighted_Sampling = 1;
   Interest_Points.Weighted_Scale = 1;
end

%%% Get total number of images
nImages = length(image_file_names);

%%% Loop over all images
for i = 1:nImages

  %%% Reset variables
  x = []; xx = [];
  y = []; yy = [];
  strength = [];
  scale = []; score = [];
  
  %%% read in image
  im=imread(image_file_names{i});
  
  %% Get size
  [imy,imx,imz]=size(im);
     
  %% Convert to grayscale if not already so...
  if (imz>1)
    im=rgb2gray(im);
  end
  
  %% Find canny edges using Oxford VGG code
  curves=vgg_xcv_segment(uint8(im),'canny_edges');   
 

  %%% Concatenate all edgel segments together into one big array
  for b=1:length(curves)
    xx = [ xx , curves{b}(1,:)]; %% x location
    yy = [ yy , curves{b}(2,:)]; %% y location
    strength = [ strength , curves{b}(3,:)]; %% edge strength
  end
  
  %% Total number of edge pixels exracted from image
  nEdgels = length(strength);
  
  
  if (nEdgels>0) %% check that some edgels were found in the image
    
    %% Obtain sampling density
    %% choose btw. uniform and weighted towards those edgels with a
    %% stronger edge strength 
    if Interest_Point.Weighted_Sampling
      sample_density = strength / sum(strength);
    else
      sample_density = ones(1,nPoints)/nPoints;
    end
    
    %% Choose how many points to sample
    nPoints_to_Sample = Interest_Point.Max_Points;
    
    %% draw samples from density
    samples = discrete_sampler(sample_density,nPoints_to_Sample,1);
    
    %% Lookup points corresponding to samples 
    x = xx(samples);
    y = yy(samples);
    
    %% now draw scales from uniform
    scale = rand(1,nPoints_to_Sample)*(max(Interest_Point.Scale)-min(Interest_Point.Scale))+min(Interest_Point.Scale);
    
    %% get scores for each points (its edge strength)
    score = strength(samples);
    
  else %%% No edgels found in image at allInterest_Point.Weighted_Sampling    = 1;


    %%% Set all output variables for the frame to be empty
    x = [];
    y = [];
    scale = [];
    score = [];
    
  end
  
    
  if DEBUG
    
    
    %%% Show image with edgels marked
    figure(1); clf;
    imagesc(im);
    colormap(gray);
    hold on;
    plot(xx,yy,'m.','MarkerSize',8)
    title('Raw edgels');
    
    %%% Show image with region marked
    figure(2); clf;
    imagesc(im);
    colormap(gray);
    hold on;
    
    for b=1:length(scale)
      plot(x(b),y(b),'r.');
      drawcircle(y(b),x(b),scale(b)*2+1,'g',1);
      hold on;
    end
    title(['Interest regions on image: ',num2str(i)]);
    
    pause
   
  end
    
  
  fprintf('Image: %d nPoints: %d\n',i,length(x));
  
  save(output_file_names{i},'x','y','scale','score');
  
  
end