config_demo.m

%Global configuration script 
%
% Holds all settings used in all parts of the code, enabling the exact
% reproduction of the experiment at some future date.
%
% Single most important setting - the overall experiment type
% used by esGenerateScanpath.m
%
% Authors
%   Giuseppe Boccignone <Giuseppe.Boccignone(at)unimi.it>
%
% License
%   The program is free for non-commercial academic use. Please 
%   contact the authors if you are interested in using the software
%   for commercial purposes. The software must not modified or
%   re-distributed without prior permission of the authors.
%
% Changes
%   20/01/2012  First Edition
%


%-------------------------------------------------------------------------
%The EXPERIMENT_TYPE identifies Feature Extraction and Salience Map methods
%-------------------------------------------------------------------------
% Features: 3DLARK
% Salience: SELFRESEMBLANCE
EXPERIMENT_TYPE = '3DLARK_SELFRESEMBLANCE'; 

%-------------------------------------------------------------------------
% DIRECTORIES - please change to your needs
%-------------------------------------------------------------------------
% Directory holding the code 
runpath = pwd;

RUN_DIR = [ runpath '/' ]

% Directory holding all the source frame
FRAME_DIR = [ RUN_DIR 'Datasets/demo/'] 

% Video name: 
VIDEO_NAME   = 'beverly01' %from CNRS dataset: clip beverly01
dir_offset   = 310; %first numbered frame of the directory
nNImageStart = 312-dir_offset; nNImageEnd=400-dir_offset;%select frames

% Directory holding all the experiment results
RESULT_DIR = [RUN_DIR 'results/'];

%-------------------------------------------------------------------------
% GENERAL PARAMETERS
%-------------------------------------------------------------------------

offset=1;
frameStep=2; %setting the frames to skip

%-------------------------------------------------------------------------
% SELF RESEMBLANCE SPATIO-TEMPORAL FEATURE & SALIENCY MAP PARAMETERS 
%-------------------------------------------------------------------------
%USING THE LARK PARAMETERES
WSIZE = 3;   % LARK spatial window size
WSIZE_T = 3; % LARK temporal window size
LARK_ALPHA = 0.42; % LARK sensitivity parameter
LARK_H = 1;  % smoothing parameter for LARK
LARK_SIGMA = 0.7; % fall-off parameter for self-resemblamnce

%if we perform a pyramid decomposition for efficiency
sLevels=4; %levels of piramid decomp

%-------------------------------------------------------------------------
% PROTO-OBJECTS PARAMETERS 
%-------------------------------------------------------------------------
PROTO      = 1; %using a proto-object representation
nBestProto = 15; %num max of retrieved proto-object


%-------------------------------------------------------------------------
% INTEREST POINT SAMPLER 
%-------------------------------------------------------------------------
% Type of interest operator to use
Interest_Point.Type                 = 'SelfResemblance';

% Scales at which features are extracted (radius of region in pixels).
Interest_Point.Scale                = [10:30];

% Maximum number of interest points allowed per image
Interest_Point.Max_Points           = 80; %30 original

% Parameters for particular type of detector
Interest_Point.Weighted_Sampling    = 1; %1= saliency weighted density, 0 = random sampling
Interest_Point.Weighted_Scale = 0;

windowSize=7; %spatial resolution of IP: this should be set as a function of the scale at which IP has been detected

WITH_PERTURBATION = true; % Adds some other IPs by sampling directly from the salience
                          % landscape    
                              
%-------------------------------------------------------------------------
% HISTOGRAM OPERATOR FOR IPs EMPIRICAL DISTRIBUTION
%-------------------------------------------------------------------------
xbinsize=20 ; ybinsize=20;

%-------------------------------------------------------------------------
% COMPLEXITY 
%-------------------------------------------------------------------------
complexity_type  ='SDL';   %Shiner-Davison-Landsberg (SDL) complexity
% complexity_type  ='LMC'; %Lòpez-Ruiz, Mancini, and Calbet complexity
% complexity_type  ='FC';  %Feldman and Crutchfield?s amendment replaces D with the Kullback-Leibler divergence


COMPL_EPS = 0.004;% 0.002
 

%-------------------------------------------------------------------------
% GAZE SAMPLING SETTINGS
%-------------------------------------------------------------------------

firstFOAonCENTER = 1; % if == 1 sets the first Foa on frame center

SIMPLE_ATTRACTOR = 0; % if == 1 using one point attractor in the potential
                      % otherwise using multipoints

%NMAX number of potential FOAS to determine the total attractor potential in
%LANGEVIN: 
%   HI...HNMAX. H=(X-X0)^2 , dH/dX=2(X-X0)
% the data
NMAX=10; %This is used if SIMPLE_ATTRACTOR=0; 

% Internal simulation: somehow related to visibility: the more the points
% that can be sampled the higher the visibility of the field

NUM_INTERNALSIM = 100; %maximum allowed number of candidate new  gaze position r_new
% If anything goes wrong retry:
MAX_NUMATTEMPTS = 5; %maximum allowed tries for sampling e new valid gaze position 


%Setting the parameters of the alpha-stable components
% -alpha is the exponent (alpha=2 for gaussian, alpha=1 for cauchian)
% -sigma is the standard deviation
% -beta  is symmetry parameter
% -delta  is location parameter (for no drift, set to 0)

%1. 'Normal gaze'
alpha_stable(1) = 2.0; gamma_stable(1) = 3.78; beta_stable(1) = 1;   
delta_stable(1) = 9; 

%2. 'Levy flight 1'      
alpha_stable(2) = 1.6; gamma_stable(2)= 22; beta_stable(2) = 1; 
delta_stable(2) = 60;   

%3. 'Levy flight 2' 
alpha_stable(3) = 1.4; gamma_stable(3) = 60; beta_stable(3) = 1;       
delta_stable(3) = 250;
 

%-------------------------------------------------------------------------
% FOR VISUALIZATION AND FILE SAVING
%-------------------------------------------------------------------------
VERBOSE = false; %comment visualization

VISUALIZE_RESULTS = 1;

%for saving need to set VISUALIZE_RESULTS=1
SAVE_FOV_IMG   = 0;  %if = 1, saving the foveated image
SAVE_SAL_IMG   = 0;  %if = 1, saving the salience map
SAVE_PROTO_IMG = 0;  %if = 1, saving the proto-object map
SAVE_IP_IMG    = 0;  %if = 1, saving the interest point map
SAVE_HISTO_IMG = 0;
SAVE_FOA_IMG   = 0;  %if = 1, saving the FOA image

%saving data
SAVE_FOA_ONFILE        = 0;
SAVE_COMPLEXITY_ONFILE = 0;