aux_error_check.m

Standard model object recognition matlab code

function stat = aux_error_check (level)

% FUNCTION stat = aux_error_check (level)
%
% Do some rudimentary error checking before initializing the
% filters or performing the main computations.

global learn_file;
global all_oper;

% The output is not used for now.
stat = 0;

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Check #1:
% Check if level is defined correctly: see init_filt_oper.m
allowed_lvl = {[1], [1 1], [1 1 1], [1 1 1 1], ...
      [1 1 0 0 1], [1 1 0 0 1 1], [1 1 0 0 1 1 1], ...
      [1 1 0 0 1 1 1 1], [1 1 1 1 1 1], [1 1 1 1 1 1 1], ...
      [1 1 1 1 1 1 1 1], ...
      [0 0 1], [0 0 0 0 1], ...
      [0 0 1 1], [0 0 0 0 1 1], [0 0 0 0 1 1 1 1], ...
      [0 0 0 0 0 0 1 1], ...
      [0 0 1 1 1 1], [0 0 1 1 1 1 1 1] ...
      };

is_lvl = 0;
for i = 1:length(allowed_lvl)
  is_lvl = is_lvl+ strcmp(num2str(level), num2str(allowed_lvl{i}));
end
if is_lvl==0
  error('Level is not well defined.');
end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Check #2:
% Check if the learn_file is defined.
for i = 1:length(level)
  if level(i)>0
    if sum([3 5 7]==i) > 0 % if S2b, S2, or S3.
      if length(learn_file)==0 | length(learn_file{i})==0
	err_str = 'learn_file is not defined.';
	error(err_str);
      end
    end      
  end
end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Check #3:
% Check if init_filter_def.m is reasonable.

% (1) number of features in C1.
[a,b,c,d] = init_filter_def(2);
is_cond = sum(b(4,:)==4) > 0;
if ~is_cond
  error('Number of orientations in C1 error.');
end

% (2) number of features in S2, S3, S2b
for i = 1:length(level)
  if level(i) > 0 & sum([3 5 7]==i)>0
    [a,b,c,d] = init_filter_def(i);

    % Load the learn_file and check the # features.
    load(learn_file{i});
    scales = aux_get_filt_info(fieldnames(f));
    for j = scales
      siz_info = getfield(f,['size_s' num2str(j)]);
      feat_in_learn_file(scales) = siz_info(4);
    end

    if max(b(4,:)) > max(feat_in_learn_file)
      disp('There are not enough features in learn_file.');
      disp([learn_file{i}]);
      disp('Some ways of correcting the error: ');
      disp('   (1) Learn more features.');
      disp('   (2) Adjust init_filter_def.m (default_num)');
      disp('   (3) Augment learn_file with dummy features.');
      error(['Not enough features in ' learn_file{i}]);
    end
  end
end

% (3) number of features within layers
for i = 1:length(level)
  if level(i) > 0 & sum([4 6 8]==i)>0
    [a,b,c,d] = init_filter_def(i);
    if sum(b(3,:)~=b(4,:))>0 % b(3,:) should equal b(4,:).
      error(['Number of features not match within Layer ' num2str(i)]);
    end
  end
end

% (4) number of featuers between layers.
for i = 2:length(level)
  if level(i) > 0
    clear b_*;
    if i == 5 % S2
      [a,b_pre,c,d] = init_filter_def(2);
    else
      [a,b_pre,c,d] = init_filter_def(i-1);
    end
    
    if length(intersect([3 5 7],i)) > 0
      % scale mixing, according to C unit.
      [a,b,c,d] = init_filter_def(i-1);
      for m_idx = 1:max(d)
	which_scale = find(d==m_idx);
	b_pre_new(4,m_idx) = max(b_pre(4,which_scale));
      end
    else
      b_pre_new = b_pre;
    end
    
    [a,b_nex,c,d] = init_filter_def(i);      
    is_cond = (b_pre_new(4,:)~=b_nex(3,:));
    if sum(is_cond)>0
      error(['Number of features not match: Layer ' num2str(i)]);
    end
  end
end